Speculative Cross-Disease Connections

ME/CFS shares features with numerous other conditions. These overlaps may reflect shared mechanisms, common susceptibility factors, or convergent pathophysiology. This section explores speculative connections that might illuminate ME/CFS pathogenesis.

1 The Post-Infectious Syndrome Cluster

ME/CFS belongs to a family of post-infectious chronic conditions that may share core mechanisms:

Long COVID. The most obvious parallel:

Nearly identical symptom profile in many patients
Similar post-exertional malaise pattern
Common autonomic dysfunction
Suggests SARS-CoV-2 triggers the same “trap” as other pathogens
Epidemiological grounding: a large electronic health record cohort ($n = 147{,}377$) found elevated new-onset ME/CFS risk persisting up to four years after SARS-CoV-2 infection (adjusted hazard ratio 1.46 hospitalized, 1.56 non-hospitalized) (Hadidchi et al. 2025), and the RECOVER study found 51% of long COVID patients meet ME/CFS criteria (Jason et al. 2025) — together indicating that the SARS-CoV-2 → ME/CFS link is not a transient phenomenon but a sustained post-infectious outcome
Speculative link: Both may involve spike protein persistence or viral reservoir maintaining immune activation
Sleep physiology parallel: SleepFM (Thapa et al. 2026, Nature Medicine, n=65,000) demonstrates that cross-modal physiological decoupling during sleep predicts disease onset across 130+ conditions (Thapa et al. 2026); both ME/CFS and Long COVID show sleep abnormalities that could reflect this decoupling pattern (Long COVID: Chaganti 2025, Tang 2025; ME/CFS: alpha-delta sleep, LC-NE dysfunction), though no direct cross-disease decoupling comparison exists

Cognitive overlap: quantitative evidence. The cognitive profiles of ME/CFS and Long COVID show striking similarity. In a direct head-to-head comparison ($n = 42$ ME/CFS, $n = 73$ post-COVID), Azcue et al. found both conditions share a core pattern of impaired attention and slowed processing, but ME/CFS patients exhibited significantly worse sustained attention (83.3% vs. 56.2% impaired) and visuospatial ability. The authors concluded that the conditions share overlapping pathology with different precipitating triggers.

Large-scale post-COVID data provide indirect quantification relevant to ME/CFS. Hampshire et al. ($n = 112{,}964$) documented cognitive deficits equivalent to $~6$ IQ points ($-0.4$ SD) in patients with persistent symptoms ($\geq$ 12 weeks), with memory and reasoning most affected. Post-COVID patients also show reaction times $~3$ SD slower than controls, with 53.5% exceeding 2 SD below normal —a degree of slowing not explained by fatigue or depression alone. Structural neuroimaging reveals grey matter loss in orbitofrontal and parahippocampal regions even after mild COVID , potentially providing the anatomical substrate for the shared cognitive impairment.

These findings strengthen the case that Long COVID and ME/CFS involve convergent neurocognitive pathology rather than merely coincidental symptom overlap. The ME/CFS cognitive profile—established by meta-analysis (effect sizes $g = -0.55$ to $-0.82$ across domains ) and multi-site objective testing —is quantitatively comparable to or more severe than post-COVID cognitive impairment, consistent with ME/CFS representing a more established or severe form of the same underlying process (see Chapter Neurological and Neurocognitive Dysfunction, Section DecodeME Brain Tissue Enrichment for detailed analysis).

PEM and exercise response: a critical divergence.

Despite overlapping symptom profiles, Long COVID and ME/CFS may differ fundamentally in their response to physical exertion. A 2-day CPET study of 15 Long COVID patients (80% reporting PEM symptoms) found NO differences between Day 1 and Day 2 CPET performance . This null result stands in sharp contrast to the consistent CPET-2 deterioration observed across multiple ME/CFS cohorts Campen, Rowe, and Visser (2020; Lim et al. 2020).

Hypothesis: Distinct PEM Pathophysiology in Long COVID vs ME/CFS

The absence of 2-day CPET impairment in Long COVID patients reporting PEM symptoms suggests that PEM in Long COVID may have a different pathophysiological basis than PEM in ME/CFS. (Certainty: 0.50)

Evidence Base. Long COVID patients with self-reported PEM symptoms show NO Day 2 CPET deterioration , whereas ME/CFS patients consistently show significant CPET-2 declines across multiple studies Campen, Rowe, and Visser (2020; Lim et al. 2020). Both conditions report subjective PEM symptoms, but only ME/CFS shows the objective CPET-2 signature.

Alternative Interpretations. The Long COVID finding could reflect: (1) insufficient sample size (n=15), (2) different disease duration (Long COVID patients earlier in disease course), (3) selection bias (Long COVID patients able to complete CPET may represent a milder subset), or (4) genuine pathophysiological difference.

Clinical Implications. If PEM mechanisms differ between Long COVID and ME/CFS, treatment approaches may need to be condition-specific. Exercise-based rehabilitation that might be tolerable for some Long COVID patients could be harmful for ME/CFS patients if the underlying pathophysiology differs.

Research Implications. Direct head-to-head comparison studies using identical 2-day CPET protocols are needed. Longitudinal studies tracking Long COVID patients over time could determine whether CPET-2 impairment emerges later in the disease course, suggesting progression toward an ME/CFS-like state.

Falsifiable Predictions.

Long COVID patients meeting full ME/CFS criteria will show CPET-2 deterioration, while Long COVID patients not meeting ME/CFS criteria will not
Longitudinal follow-up of Long COVID patients will show that those who develop CPET-2 impairment are more likely to meet ME/CFS criteria over time
Molecular profiling of exercise responses in Long COVID vs ME/CFS will show distinct patterns (e.g., immune activation, metabolic signatures)

Treatment-response convergence: a strong shared-mechanism signal. Despite possible PEM-divergence, patient-reported treatment responses strongly converge across the two conditions. In the largest treatment-outcome survey to date (Eckey et al. 2025, $n = 3{,}925$ ME/CFS and long COVID patients, 150+ treatments), the Net Assessment Scores of treatments were highly correlated between ME/CFS and long COVID patients ($R^2 = 0.68$) (Eckey et al. 2025). The same four symptom/comorbidity clusters emerged largely independent of diagnosis label, and disease severity (capacity) — not the ME/CFS-versus-long-COVID label — was the strongest predictor of treatment response. Shared cluster membership predicted shared treatment benefit (e.g., a POTS-dominant patient benefited from autonomic modulators regardless of originating diagnosis). Only two treatments (midodrine; benfotiamine/thiamine) showed significantly different responses between the conditions (Eckey et al. 2025). This treatment-response convergence is functionally parallel to the symptom convergence documented above and argues that shared pathophysiology outweighs condition-specific differences for the majority of patients, even where a specific physiological signature (such as CPET-2) diverges.

Immune-profiling convergence. The convergent picture extends to immune readouts. The canonical long-COVID immune-profiling study by Klein et al. (2023, Nature) documented T-cell exhaustion, exhausted B cells, and exaggerated humoral responses to SARS-CoV-2 and to latent EBV/VZV in long COVID (Klein et al. 2023) — a cross-sectional long-COVID-only profile (no ME/CFS arm) that is consistent with the T-cell exhaustion and herpesvirus antibody dysregulation documented in ME/CFS, an inference drawn across separately-published studies rather than a head-to-head comparison (Chapter Immune System Dysfunction, Section t cells). A direct head-to-head comparison found no significant biomarker differences between ME/CFS and long COVID for lymphocytes, CD8⁺ T cells, NK cells, or IL-6/TNF/IL-4/IL-10 (Petrov et al. 2025). This immune convergence parallels the treatment-response convergence above. Two caveats temper the claim: the direct-comparison and divergence panels derive from the same Plovdiv research group with shared authorship (not independent replications; Section t cells, Cohort Overlap Constrains the Convergence Claim), and long-COVID immune changes appear partly reversible by 24 months in mild cohorts (Long COVID Immune Dysregulation as a Time-Limited Precursor State), in contrast to ME/CFS persistence.

Circulating cell-free mitochondrial DNA: a potential divergence point. Despite clinical overlap, Long COVID and ME/CFS may differ in circulating cell-free mitochondrial DNA (ccf-mtDNA) dynamics. In the EPILOC population-based cohort ($n = 228$), Matits et al. found reduced relative ccf-mtDNA in Long COVID patients ($n = 128$) compared to recovered controls ($n = 100$; partial $\eta^2 = 0.01$–$ 0.02$; $p = 0.089$ with full covariate adjustment, i.e. non-significant in the primary analysis; $p = 0.038$ only after excluding high-CRP outliers in a sensitivity analysis). Lower ccf-mtDNA correlated with worse general cognition, while CRP showed no independent association with cognitive function after controlling for ccf-mtDNA. Total cell-free DNA was not elevated, consistent with the reduction being mitochondria-specific rather than reflecting generalized cell death (though alternative explanations such as differential DNA stability or assay-specific effects cannot be excluded). The authors interpret this as impaired mitophagy: damaged mitochondria accumulate intracellularly instead of being released and cleared. A prior case series (Szögi et al. ; $n = 5$ Long COVID, $n = 5$ ciliary dyskinesia controls) also reported reduced ccf-mtDNA alongside mitochondrial ultrastructural abnormalities in Long COVID tissue biopsies; however, this was an extremely small sample with non-standard controls and should be considered preliminary rather than confirmatory.

By contrast, in ME/CFS, Tsilioni et al. found exosome-associated mtDNA elevated in serum after exercise challenge—not at rest (sample size, effect size, and p-values not reported in the abstract; the study has not been independently replicated). If confirmed by head-to-head studies using identical methodology, this divergence could point to fundamentally different mitochondrial dynamics: Long COVID involving impaired mitophagy (damaged mitochondria trapped intracellularly) versus ME/CFS involving exercise-triggered mitochondrial content release via exosomes (see Section Epigenetic Modifications in Chapter Genetic and Epigenetic Factors for the broader mitophagy context).

Limitation: ccf-mtDNA Comparison: Different Methods Preclude Direct Conclusions

The apparent Long COVID vs ME/CFS divergence in ccf-mtDNA must be interpreted cautiously: the studies use different compartments (free plasma vs. exosome-associated), different timing (resting vs. post-exercise), and different quantification methods (qPCR $\Delta$CT vs. exosome isolation). No study has measured free ccf-mtDNA at rest in a well-characterized ME/CFS cohort using the Matits et al. protocol. Additionally, the Matits effect size is very small (partial $\eta^2 \leq 0.02$) and loses statistical significance after full covariate adjustment. Physical activity itself increases ccf-mtDNA release; deconditioning in both patient populations may confound resting-state measurements independently of primary mitochondrial pathology.

Speculation: ccf-mtDNA as a Potential Long COVID vs ME/CFS Distinguishing Biomarker

Certainty: 0.25. If the pattern holds—low resting ccf-mtDNA in Long COVID versus elevated exosome-associated mtDNA post-exercise in ME/CFS —ccf-mtDNA dynamics could help distinguish the two conditions. This would imply different predominant mitochondrial failure modes: impaired clearance (Long COVID) versus excessive stress-triggered release (ME/CFS). However, no direct comparison using identical methodology exists, and the Long COVID finding itself is borderline (very small effect, lost significance with full covariate adjustment). The certainty is low because: (a) no ME/CFS resting ccf-mtDNA data exist; (b) the deconditioning confound applies to both conditions; (c) anxiety disorders show similarly low ccf-mtDNA, reducing specificity.

Testable prediction: A study measuring ccf-mtDNA by qPCR at rest and post-exercise in matched ME/CFS, Long COVID, and healthy control groups would show divergent trajectories: ME/CFS patients showing higher post-exercise ccf-mtDNA release than Long COVID patients, with both differing from controls. Falsified if resting ccf-mtDNA is equivalently low in both conditions, or if post-exercise trajectories are indistinguishable between the two conditions.

Clinical note: ccf-mtDNA measurement is a research-only assay. Until head-to-head studies exist, clinicians should distinguish Long COVID from ME/CFS using established clinical criteria (symptom duration, PEM pattern, exclusion criteria); ccf-mtDNA adds nothing to current diagnostic practice.

Speculation: Peripheral Blood Methylome as a Long COVID vs ME/CFS Distinguishing Signature

Certainty: 0.52. The peripheral blood mononuclear cell (PBMC) methylome separates ME/CFS, Long COVID (LC), and healthy controls into three distinct clusters by principal component analysis, while also showing a largely shared epigenetic makeup between the two conditions (Peppercorn et al. 2025). In a direct age/sex-matched RRBS comparison (n=5 each), ME/CFS had 214 differentially methylated fragments (DMFs) vs healthy controls and Long COVID had 429 (both a methylation difference of more than 10%, p < 0.05); 118 DMFs were shared, with a Pearson correlation of 0.88 between the two disease cohorts, but 26 of the shared fragments differed by more than 10% between the two conditions and 6 showed opposite-direction methylation (five hypermethylated in LC and hypomethylated in ME/CFS, one the reverse) (Peppercorn et al. 2025). This bidirectional-overlap-plus-specific-signature pattern is consistent with the paper’s per-locus vector model of directional methylation change (Per-Locus Dynamics: Vector Model for Bidirectional Methylation). The shared makeup favors treating ME/CFS and LC as mechanistically overlapping; the six opposite-direction loci (e.g., the FGD2 promoter hypermethylated in LC but hypomethylated in ME/CFS) are candidate condition-specific epigenetic signatures that could aid differential diagnosis and stratification, though unvalidated.

Alternative interpretations / competing mechanisms (Standing Epistemic #4). The apparent LC-vs-ME/CFS divergence may reflect disease stage rather than condition identity: the LC cohort was ~1 year from onset while the ME/CFS cohort averaged ~12 years, so LC’s larger DMF count could represent an earlier, more dynamic methylome that later settles into the more stable ME/CFS profile (or vice versa) (Peppercorn et al. 2025). SARS-CoV-2-specific effects (rather than any post-infectious mechanism) could also drive the LC-specific loci. The six opposite-direction DMFs rest on n=5 per cohort and are not validated for diagnosis, so they are hypothesis-generating, not diagnostic.

Falsifiable prediction. A stage-matched, larger (n≥50/group) RRBS or array study with LC and ME/CFS at comparable time-from-onset would find: (i) the Pearson R=0.88 shared-makeup relationship reproduces; (ii) the LC-vs-ME/CFS DMF amplitude difference shrinks or disappears when time-from-onset is matched; and (iii) the six opposite-direction DMFs either replicate as condition-specific or collapse with stage matching. Falsified if stage-matched cohorts show identical methylomes with no reproducible condition-specific DMFs.

Severity applicability: Unknown — the Peppercorn cohort was not stratified by severity.

Consequence: If the six opposite-direction methylation loci replicate, they could become a blood-based test helping clinicians separate Long COVID from ME/CFS by underlying biology rather than by symptom duration alone; even without that, the finding is a strong caution against pooling the two conditions in methylation research.

A third divergence point: brain MRS neurochemistry. Ultra-high-field 7 Tesla magnetic resonance spectroscopy (MRS, a non-invasive imaging method that measures tissue chemistry in a living brain region) provides an in vivo layer of evidence on the Long COVID / ME/CFS question (Godlewska et al. 2025). In a direct comparison (24 ME/CFS, 25 Long COVID, 24 healthy controls), ME/CFS patients showed elevated brain lactate in both the pregenual and dorsal anterior cingulate cortex (energy-metabolite consistent with energetic stress), whereas Long COVID patients showed reduced total choline (a membrane/lipid metabolite) specifically in the dorsal anterior cingulate (Godlewska et al. 2025). Resting calf-muscle metabolites did not differ between groups. The authors conclude that, despite similar clinical presentations, ME/CFS and Long COVID show distinct neurochemical signatures and should not be pooled as a single group until the mechanisms are understood (Godlewska et al. 2025). This mirrors the CPET-2 and ccf-mtDNA divergences above as a third, methodologically distinct line of evidence pointing to mechanism divergence.

Hypothesis: Distinct Brain Neurochemical Signatures Distinguish ME/CFS from Long COVID (7T MRS)

Ultra-high-field 7 Tesla MRS shows a separable brain chemistry profile for ME/CFS (elevated anterior-cingulate lactate) versus Long COVID (reduced dorsal-cingulate choline), supporting distinct rather than identical neurobiological mechanisms despite overlapping symptoms (Godlewska et al. 2025). (Certainty: 0.50.)

Evidence Base. In a single-protocol head-to-head study (24 ME/CFS, 25 Long COVID, 24 HC), ME/CFS brain lactate was elevated in both pgACC ($p = 0.004$) and dACC ($p = 0.006$), consistent with the independent-lab brain-lactate replication cluster (Mount Sinai: Murrough et al. 2010, Natelson et al. 2017; UAB: Mueller et al. 2020) (Godlewska et al. 2025) (Murrough et al. 2010) (Natelson et al. 2017) (Mueller et al. 2020). Long COVID showed reduced total choline in dACC ($p = 0.0002$), a finding without a stated ME/CFS counterpart in the same cohort. Resting calf-muscle metabolites did not differ between groups (Godlewska et al. 2025).

Alternative Interpretations / competing mechanisms (Standing Epistemic #4). The lactate elevation in ME/CFS does not uniquely identify a cause: it could reflect (1) mitochondrial dysfunction (impaired oxidation of lactate), (2) astrocyte-neuron lactate shuttle export failure, (3) hypoperfusion-driven anaerobic shift, or (4) inflammatory-cell glycolysis. The choline reduction in Long COVID is associative, not causal, and its direction is not uniform: Mueller et al. 2020 reported elevated choline in ME/CFS cingulate, and Pajuelo et al. 2024 found increased choline in the corpus callosum of older post-COVID patients (Mueller et al. 2020) (Pajuelo et al. 2024) — so choline changes are region-, age-, and condition-dependent, and the dACC reduction should not be generalized. A further confound is disease duration: long COVID patients are typically earlier in their disease course than established ME/CFS cohorts, so a temporal (rather than condition-specific) trajectory — later-progressing long COVID adopting an ME/CFS-style lactate-laden profile — could masquerade as a cross-condition divergence. This is why longitudinal follow-up (falsifiable prediction below) is the decisive test.

Independent-group neurochemical divergence at lower field. A separate Griffith University multimodal MRI study found posterior-cingulate neurochemical differences specifically between long COVID and COVID-recovered controls — higher $N$-acetylaspartate (NAA, a neuronal/energy marker) in long COVID and higher glutamine in recovered controls — while long COVID did not differ from never-infected controls on these metabolites (Thapaliya et al. 2025). This independently reinforces the general claim that long COVID and post-COVID-recovered groups carry distinguishable brain neurochemical profiles, though the specific metabolites (NAA/glutamine here vs lactate/choline in Godlewska) and brain regions (posterior cingulate here vs anterior cingulate) differ — so the two studies do not yet agree on which metabolites or regions carry the divergence, only that a divergence exists. The NAA elevation is itself of ambiguous direction (compensatory response to reduced metabolic efficiency versus osmotically driven change), so this should be read as corroborating the existence of a neurochemical divergence, not as identifying its specific content.

Clinical Implications. If the neurochemical divergence is real, the treat-as-one-population practice in some long COVID and ME/CFS trials may mask condition-specific responses. This is a research planning implication, not a change to current clinical care.

Research Implications. Cross-sectional MRS cannot establish causation. Direct head-to-head MRS with matched cohorts, plus perfusion measurement (to separate hypoperfusion from mitochondrial causes of lactate), and longitudinal follow-up (does Long COVID choline normalize or progress toward an ME/CFS lactate profile?) are needed.

Limitation (MRS interpretation caution). The divergence is a single-site 7T observation (24/25/24) not yet independently replicated at ultra-high field; 7T metabolite quantification is sensitive to voxel placement, T2/water-referencing, and patient arousal during scanning, and no severity/medication/diet stratification or perfusion co-measure was reported. The central evidence for ME/CFS (brain lactate elevation) is independently replicated at lower field strengths (Mount Sinai ventricular lactate; UAB regional LAC-MRS), but the Long COVID choline divergence rests on one cohort and the choline direction is region-, age-, and condition-dependent, so the divergence should be treated as a provisional hypothesis pending replication.

Falsifiable Predictions.

A head-to-head 7T MRS replication will find ME/CFS lactate elevated but Long COVID lactate not elevated in cingulate cortex (and vice versa for choline), i.e. the divergence is reproducible.
Long COVID patients who later meet full ME/CFS criteria will show a shift toward the ME/CFS lactate-elevation signature, whereas those who do not will retain the choline-reduction signature.
Concurrent arterial-spin-labeling perfusion will show whether the ME/CFS lactate elevation tracks cerebral hypoperfusion (implicating a vascular cause) or persists with normal perfusion (implicating a primary mitochondrial/metabolic cause).

Severity applicability: Unknown — the Godlewska cohort was not stratified by severity.

Consequence: If confirmed, this would change how researchers design trials — stopping the common practice of pooling ME/CFS and long COVID patients as a single group — and would give brain-imaging an early role in telling the two conditions apart by their underlying chemistry, though it is not yet ready to guide any clinical decision.

Speculation: Long COVID Dorsal-Cingulate Choline Reduction: Coagulation and “Brain Fog” Substrate

Certainty: 0.35. The reduced total choline (tCho) in the dorsal anterior cingulate of Long COVID patients may reflect altered membrane phospholipid metabolism (choline is a precursor for membrane phosphatidylcholine and a source of choline for the one-carbon/anticoagulation axis) (Godlewska et al. 2025). The authors link this to the association between blood clots and “brain fog,” and to earlier animal work showing that choline can prevent intravascular coagulation. The speculation is that a brain-region-specific choline deficit either (a) is a marker of endothelial/membrane stress in the setting of microclotting, or (b) contributes to a pro-coagulant local milieu relevant to cognitive symptoms. Evidence is associative and cross-sectional; there is no in-cohort coagulation assay and no demonstration that the dACC choline deficit causally drives symptoms.

Falsifiable prediction. (1) Long COVID patients with low dACC choline will show elevated markers of coagulation/fibrinolysis (D-dimer, fibrinogen, or microclot assay) relative to those with normal dACC choline. (2) If choline’s anticoagulant role is causal, dietary choline supplementation would raise dACC choline and correlatively lower coagulation markers — but subject to the caveat that choline status is driven by diet and the one-carbon cycle, not simply by tissue concentration, and no human trial supports this. (3) Falsified if dACC choline is unrelated to intravascular coagulation markers and to cognitive symptom severity.

Clinical note. This is a research hypothesis, not a clinical recommendation. There is no evidence that choline supplementation changes Long COVID brain chemistry or symptoms, and the choline-coagulation link is mechanistic speculation. No dietary or supplement guidance is warranted. Severity applicability: Unknown — cohort not severity-stratified.

Consequence: If replicated, this would point to a treatable axis (the one-carbon/choline/coagulation pathway) in at least a subset of Long COVID patients with brain fog, and would give MRS a candidate role as a stratification biomarker — but it is currently too early and too associative to guide any clinical decision.

Speculation: Shared Brainstem-Cerebellar-Limbic Vulnerability and the Bidirectional Microstructure Signature in Post-Infectious Fatigue

(Certainty: 0.28.) Two cross-disease observations from the long-COVID / ME/CFS neuroimaging literature may be read as one coherent picture rather than scattered inconsistencies (Origin: brainstorm). First, the regions that recur across independent studies — pons, midbrain, cerebellar tonsil, superior longitudinal fasciculus, and cingulum (Thapaliya et al. 2025) (Thapaliya et al. 2021) (K. Wu et al. 2026) — cohere into a single distributed circuit (the ascending brainstem-to-cerebellar-to-thalamic-to-cingulate limbic-cortical loop), suggesting a shared site of vulnerability in post-infectious fatigue conditions even though the specific metrics and directions differ. Second, the apparent contradiction in diffusion direction across studies (elevated FA in right SLF in long COVID (Thapaliya et al. 2025) and in the arcuate in ME/CFS (Zeineh et al. 2015), versus reduced FA in the cingulum (K. Wu et al. 2026) and reduced diffusivity in the recovered caudate (Thapaliya et al. 2025) (Lu et al. 2020), versus elevated-diffusivity edema on NII (Yu2026diffusion?)-neuroinflammation) may index a two-sided injury-repair balance rather than noise: restricted diffusion / elevated FA co-localising with elevated myelin-signal = remyelination and compensatory reorganisation on the recovery side, while elevated diffusivity / reduced FA = demyelination, edema, and injury. This reframes the long-standing “inconsistent findings” complaint (documented in the neuroimaging review (Maksoud et al. 2020)) as a predictable bidirectional signature that tracks where a patient sits on a damage-to-repair continuum.

Evidence base: All constituent findings are cited above and are cross-sectional; no study yet measures both the microstructural direction and a repair marker in the same recovering patient, so the synthesis is a hypothesis, not an established pattern. The circuit framing overlaps the existing speculation on cortico-cerebellar and brainstem glutamatergic vulnerability (Section Schizophrenia GWAS Glutamatergic Parallel — Same Pathway, Different Circuits); this speculation adds the microstructural-repair reading and the shared-region observation rather than re-stating the genetic argument.

Falsifiable prediction: Harmonised multi-site tractometry of the brainstem–cerebellar–thalamic–cingulate axis will show group-level alteration in ME/CFS and long COVID that is stronger in this circuit than in matched control tracts (e.g., visual or motor tracts), supporting circuit-specificity rather than diffuse change. Second, within-subject, tracts showing elevated FA should co-localise with elevated T1w/T2w (both read as remyelination) and with recovering regions, while tracts showing reduced FA should co-localise with edema signatures and with symptom severity. The synthesis is falsified if diffusion direction is random with respect to recovery state, or if the regional pattern is not circuit-specific.

Severity applicability: Unknown — the constituent cohorts were not severity-stratified, and severity-gated nulls (no overall FA difference in unselected cohorts (Arendt et al. 2026)) caution that the signature may be severity- or subset-specific.

Consequence: If the “confusing” scan results are actually a coherent picture of simultaneous damage and repair concentrated in one brain circuit, researchers can stop arguing about which direction is “right” and start measuring the damage-versus-repair balance and focusing imaging on the shared circuit — a research-direction change, not a clinical one.

2 Autoantibody Convergence: GPCR and Tissue-Specific Mechanisms

Hypothesis: Long COVID and ME/CFS Share Two Distinct Autoantibody Pathways

(Certainty: 0.50.) Long COVID and ME/CFS may share converging autoantibody mechanisms: GPCR autoantibodies (predominantly IgG, targeting functional receptors) and tissue-specific structural autoantibodies (predominantly IgM, targeting organ-restricted proteins). Evidence:

GPCR autoantibodies in both conditions: (Wallukat et al. 2021) detected functional autoantibodies against GPCRs in long COVID—a pattern similar to ME/CFS (Freitag et al. 2021). This convergence supports a shared autoimmune pathway.
Tissue-specific antibodies in long COVID: (Tatai et al. 2026) (n=114) identified cardiac (54%), vascular (34%), and pulmonary (34%) autoantibodies—predominantly IgM, undetected by routine ANA HEp-2. ME/CFS data on tissue-specific structural autoantibodies are absent.
Pathogenic complementarity: GPCR autoantibodies may cause autonomic dysfunction (vascular dysregulation, orthostatic intolerance); tissue-structural autoantibodies may cause organ-specific manifestations (cardiac dysfunction, vascular pathology, pulmonary involvement). Co-occurrence is predicted to produce more severe disease.

Certainty basis: GPCR overlap is supported by two independent studies across two conditions (Wallukat 2021 for long COVID; multiple Charité cohort studies for ME/CFS). Tissue-specific overlap is inferred from long COVID only—no ME/CFS tissue-specific autoantibody data exist. The dual-pathway framework is a mechanistic synthesis; certainty 0.50.

Testable prediction: Head-to-head comparison of ME/CFS and long COVID patients using both CellTrend GPCR ELISA and multi-tissue Western blot will show: (a) both populations have elevated GPCR autoantibodies compared to recovered and healthy controls; (b) both have elevated tissue-specific structural autoantibodies; (c) the autoantibody profile of post-COVID ME/CFS will be indistinguishable from infection-triggered non-COVID ME/CFS—supporting a shared post-infectious mechanism independent of trigger identity. Falsified if GPCR autoantibody profiles differ significantly between the two conditions.

Cross-reference: Section Increased Vulnerability to Iron-Dependent Cell Death for the detailed tissue-specific autoantibody evidence; Section GPCR Autoantibody-Driven Dysfunction for the GPCR autoantibody evidence.

Open Question: Is the IgM-dominant autoantibody profile in long COVID transient or stable?

(Tatai et al. 2026) found persistent IgM autoantibodies and de novo emergence at 141-day follow-up, suggesting ongoing immune dysregulation rather than resolution. Without longer follow-up (over 2 years), it is unknown whether IgM dominance represents: (a) a stable chronic state of failed class switching; (b) a prodromal phase that transitions to IgG dominance over time; or (c) transient extrafollicular activation that resolves within 1–2 years. Natural history data in ME/CFS would distinguish whether the IgM-to-IgG transition timeline differs between patients who recover and those who remain ill. No longitudinal ME/CFS data exist.

Direct measurement: Serum IgM and IgG autoantibody titers against tissue homogenates at diagnosis and at 6, 12, and 24 months in a prospective post-infectious cohort would establish trajectory. If IgG antibodies emerge over time as IgM wanes (class switching), this supports a maturing autoimmune response. If IgM persists without IgG emergence, this supports chronic extrafollicular activation with class-switch failure.

3 Schizophrenia Autoantibodies: A Historical Parallel for ME/CFS Subgroup Discovery

The arc of autoantibody discovery in schizophrenia — from early 20th century serological clues through modern unbiased screening — offers a historically instructive parallel for ME/CFS autoantibody research (Bartley and Ross 2020).

3.1 Historical Parallel: From Kraepelin to Phage Display

The schizophrenia autoantibody story spans nearly a century. In 1937, Lehmann-Facius found that 95% of schizophrenia patients harbored brain-reactive autoantibodies using a cerebrospinal fluid flocculation assay — a finding that could not be replicated due to assay limitations, a pattern that would recur repeatedly. For decades, the schizophrenia field pursued a unifying dopaminergic model, even as 60% of patients failed to respond to dopamine blockade, and the autoantibody hypothesis was periodically abandoned and resurrected as technology evolved (Bartley and Ross 2020).

The decisive breakthrough came in 2007 when Dalmau et al. identified 12 women with subacute psychosis caused by inhibitory NMDAR autoantibodies — a treatable autoimmune encephalitis masquerading as schizophrenia (Dalmau et al. 2008). Subsequent screening found peripheral anti-NMDAR autoantibodies in nearly 10% of chronic schizophrenia patients, yet these antibodies were virtually absent from CSF — a discrepancy that initially defied explanation (Ezeoke et al. 2013) (Pollak et al. 2014). Meta-analyses eventually established a pooled NMDAR seropositivity of 3.7–8.0% depending on IgG subclass stringency (Luykx et al. 2024) (Pearlman and Najjar 2014).

The technological inflection point came with phage display. Wilson and DeRisi used programmable phage display (REAP) to screen for autoantibodies at proteome scale, identifying novel autoantibodies that had eluded detection for decades (Pluvinage et al. 2024). In 2026, Nemani et al. (preprint) applied REAP screening (6,183 extracellular proteins) to 352 schizophrenia patients and 971 controls, finding that schizophrenia patients had nearly double the extracellular autoantibody burden of controls, targeting CNS antigens, neuroactive receptors, ion channels, synaptic proteins, and BBB antigens — with higher burden predicting worse antipsychotic response (Nemani et al. 2026).

3.2 Parallels to ME/CFS

Hypothesis: Schizophrenia Autoantibody Discovery Arc as a Model for ME/CFS

(Certainty: 0.45.) The schizophrenia autoantibody discovery trajectory — early positive findings not replicable due to assay limitations, decades of pursuit of a unifying non-autoimmune model, eventual recognition of autoantibody-positive subgroups through unbiased screening, and demonstration of immunotherapy-responsive cases — may be repeating in ME/CFS. The specific parallels:

Diagnostic heterogeneity masking subgroups: Just as Kraepelin’s “group of schizophrenias” concept was abandoned for a half-century in favor of a single dopaminergic model before being revived by autoantibody discoveries (Bartley and Ross 2020), ME/CFS may be prematurely consolidated under a single pathophysiological model despite evidence of autoantibody subgroups (Loebel et al. 2016) (Freitag et al. 2021).
Targeted testing blindness: In schizophrenia, screening for a handful of autoantibodies (NMDAR, VGKC, GAD65) found low prevalence (0.5–1.5% for NMDAR IgG) (Schou et al. 2016), leading many to conclude autoantibodies were irrelevant. The ME/CFS parallel: CellTrend ELISA screening for GPCR autoantibodies produces positive results in some cohorts (Freitag et al. 2020), but the most comprehensive screen to date (Germain 2025, 7,542 antibody-antigen interactions using REAP + Luminex) found no signal (Germain et al. 2025). This null may reflect the same limitation — targeted assays (GPCR ELISA) and even moderately broad screens (REAP with extracellular domain fragments) may miss conformational, post-translationally modified, or multi-subunit epitopes.
Unbiased screening as the solution: The schizophrenia field needed phage display (REAP) to reveal the true autoantibody burden (Nemani et al. 2026). ME/CFS has not undergone equivalent proteome-wide unbiased autoantibody screening. The Germain 2025 null, while the most comprehensive ME/CFS autoantibody screen to date, used REAP displaying individual extracellular domains — which may miss conformational epitopes from multi-loop GPCR structures or multi-subunit receptor complexes.
Immunotherapy-responsive hidden subgroup: In schizophrenia, immunosuppression (glucocorticoids, plasma exchange) benefits the autoantibody-positive subset (D. Endres et al. 2020) (Hansen et al. 2023) but failed in unselected trials (Cox et al. 2020). The ME/CFS parallel: immunoadsorption and daratumumab show benefit in open-label autoantibody-positive studies (Stein et al. 2025) (Fluge et al. 2025), but the sham-controlled IA-PACS-CFS trial — which enrolled patients with elevated GPCR autoantibodies — reported a null primary outcome (Rücker 2026), suggesting autoantibody selection alone may be insufficient or the CellTrend-based autoantibody criterion may not identify the pathogenic subgroup.
The “better prognosis” paradox: In schizophrenia, Luykx et al. (2024) found that NMDAR-seropositive patients had less severe negative symptoms and better psychosocial functioning (Luykx et al. 2024) — paradoxical if autoantibodies simply worsen disease. The ME/CFS analog: Azcue et al. (2026) found M2 and M4 muscarinic receptor autoantibodies were lower in ME/CFS vs controls, while β2-adrenergic autoantibodies correlated with specific symptom domains rather than global severity (N. Azcue et al. 2026). In both conditions, autoantibody profiles may define distinct clinical subgroups with different disease trajectories rather than uniformly more severe pathology.

Certainty basis: The historical parallel rests on structurally analogous research trajectories across two fields, not direct evidence. The Dalmau discovery (2007) and Nemani 2026 REAP schizophrenia screen provide the positive proof-of-concept. The Germain 2025 null in ME/CFS — the most comprehensive autoantibody screen to date — is a significant disanalogy: unlike pre-REAP schizophrenia (where targeted assays found positive signals), REAP itself found null in ME/CFS. Additionally, the IA-PACS-CFS sham-controlled null in autoantibody-enriched patients (Rücker 2026) challenges the core treatment prediction. The methodological analogy is suggestive but not probative. Certainty 0.45: biased downward from 0.55 to account for Germain 2025 REAP null and IA-PACS-CFS null, both of which materially weaken the direct parallel.

Evidence gaps:

Germain 2025 performed a REAP screen (7,542 interactions) in ME/CFS and found null (Germain et al. 2025) — whether this reflects biological absence or a domain-fragment limitation of the specific REAP library used is unresolved. A complementary REAP screen using full-length native protein libraries is needed to distinguish these possibilities
No CSF autoantibody profiling in ME/CFS using unbiased platforms (Bynke 2020 found no CSF GPCR autoantibodies by CellTrend ELISA, but unbiased screening may detect targets invisible to GPCR panels)
The schizophrenia-ME/CFS comparison is cross-disease analogy, not direct evidence of shared mechanism

Falsifiable prediction: REAP screening of ME/CFS plasma against a proteome-scale extracellular protein library (identical to the Nemani 2026 protocol) will identify a set of autoantibody targets — including at least some not currently measured by GPCR ELISA — that are significantly enriched in ME/CFS vs healthy controls (q < 0.05 after FDR correction), with the overall autoantibody burden correlating with immunoadsorption/daratumumab treatment response. Falsified if the total extracellular autoantibody burden is equivalent between ME/CFS and healthy controls (no q-value below 0.10 in an adequately powered cohort of n ≥ 100 per group), or if the identified targets do not differentiate treatment responders from non-responders.

3.3 Methodological Lessons for ME/CFS

Several specific methodological lessons from the schizophrenia autoantibody literature apply directly to ME/CFS research:

Plasmapheresis/dialysis trials in unselected schizophrenia were negative (Cox 2020 systematic review of 9 studies, n=105): 6/8 dialysis studies were null, 1 beneficial, 1 harmful; the single plasmapheresis trial was ineffective (Cox et al. 2020). The IA-PACS-CFS null in autoantibody-enriched ME/CFS (Rücker 2026) is a key disanalogy: unlike the schizophrenia apheresis trials (which did not select for autoantibody positivity), the ME/CFS trial enrolled autoantibody-positive patients and still produced a null result, challenging the simple “selection solves it” narrative.

The IgG subclass problem. In schizophrenia, NMDAR IgA/IgM antibodies are common (7–8% prevalence) but only IgG subclass is clearly pathogenic (0.5–1.5%) (Pollak et al. 2014) (Schou et al. 2016). In ME/CFS, the same issue arises: CellTrend ELISA detects total IgG against GPCRs, but functional bioassays (distinguishing activating vs blocking IgG) may be more specific (Loebel et al. 2016). Both fields need functional characterization of antibody pathogenicity, not just binding assays.

Rare-to-common paradigm. Rare autoimmune encephalitis (NMDAR, LGI1, CASPR2) in neurology clinics provided the proof-of-concept that autoantibodies cause psychiatric symptoms, which then motivated screening in idiopathic psychosis (Casanova and Abel 2022). Similarly, rare autoimmune conditions (myasthenia gravis, Lambert-Eaton) may provide the proof-of-concept for autoantibody-mediated fatigability that motivates research in idiopathic ME/CFS.

Clinical screening tools. The Neuropsychiatric Checklist for Autoimmune Psychosis (Tebartz van Elst 2025) systematically integrates red-flag symptoms, FDG-PET, CSF, EEG, and MRI findings for assessing autoimmune psychosis likelihood (Tebartz van Elst et al. 2025). No equivalent diagnostic algorithm exists for autoimmune ME/CFS. Developing such a tool — integrating autonomic testing, immunoadsorption response history, GPCR autoantibody profiles, unbiased autoantibody screening, and CSF markers — would enable systematic identification of the autoimmune ME/CFS subgroup.

3.4 Summary Chart: Schizophrenia Autoantibody Milestones and ME/CFS Parallels

Schizophrenia autoantibody milestones with corresponding ME/CFS status
[Schizophrenia Milestone, ME/CFS Status, Certainty]	[Lehmann-Facius 1937: 95% seropositive by flocculation (unreplicable),, Similar: early GPCR autoantibody reports (Loebel et al. 2016) with subsequent replication challenges; Germain 2025 comprehensive null (Germain et al. 2025), 0.35]	[Dalmau 2007: anti-NMDAR encephalitis defined a treatable autoimmune psychosis subgroup, Similar: GPCR autoantibody pathology suggested by immunoadsorption response (Stein et al. 2025); daratumumab response (Fluge et al. 2025), 0.60–0.65]
[Targeted screening found 3.7–8% NMDAR-Ab+ but only 0.5–1.5% IgG subclass (Pollak et al. 2014), CellTrend ELISA: 29–91% GPCR-Ab+ depending on cohort and threshold (Loebel et al. 2016) (Bynke et al. 2020); REAP/Luminex null (Germain et al. 2025), 0.30–0.50]	[Nemani 2026: REAP proteome-wide screening found 2× autoantibody burden in schizophrenia vs controls, Not performed in ME/CFS — equivalent unbiased screen is the single highest-yield experiment, 0.00 (not done)]	[Plasmapheresis in unselected schizophrenia: null (Cox et al. 2020), IA-PACS-CFS in autoantibody-enriched ME/CFS: null (Rücker 2026) (key disanalogy), 0.35 (parallel weakened by null in enriched patients)]
[Endres 2020: 94% immunotherapy response in autoantibody-confirmed autoimmune psychosis, Stein 2024: 70% IA response in antibody-positive post-COVID ME/CFS; Fluge 2025: 60% daratumumab response, 0.50]	[Tebartz van Elst 2025: clinical diagnostic algorithm for autoimmune psychosis, No equivalent diagnostic algorithm for autoimmune ME/CFS, 0.00 (not developed)]	[Kraepelin/Bleuler “group of schizophrenias” reconceptualization driven by autoantibody evidence, ME/CFS subgroup reconceptualization driven by autoantibody evidence — partially underway, 0.40]

Speculation: ME/CFS Autoantibody Subgroup Discovery Will Follow the Schizophrenia Trajectory

(Certainty: 0.45.) If the schizophrenia trajectory is a model for ME/CFS, then: (1) unbiased proteome-wide autoantibody screening (REAP) will identify a panel of novel autoantibody targets enriched in a subset of ME/CFS patients; (2) this autoantibody-positive subgroup will show preferential response to immunomodulatory therapy; (3) the initial discovery will come from a well-characterized cohort with clinical features suggestive of autoimmune involvement (infection-triggered onset, inflammatory markers, multi-system involvement). The Germain 2025 null may reflect either (a) genuine absence of autoantibodies in chronic, pre-COVID ME/CFS, or (b) technological limitations of REAP with individual extracellular domain fragments for detecting conformational epitopes. The schizophrenia experience — where decades of negative targeted screening preceded REAP-based discovery — suggests option (b) cannot be excluded without a dedicated proteome-wide screen using complementary platforms.

Falsifiable prediction: A head-to-head comparison of three autoantibody screening platforms (CellTrend GPCR ELISA, REAP with full-length native protein libraries, and multi-tissue Western blot) in the same ME/CFS cohort (n ≥ 100) will show a Jaccard similarity < 0.4 between GPCR ELISA and REAP target lists, and at least one platform will identify targets significantly enriched in ME/CFS vs controls (q < 0.05 after FDR correction). Falsified if all three platforms converge on equivalent null results (no q < 0.10 for any target on any platform).

Gap: Zero ME/CFS studies have used proteome-wide unbiased autoantibody screening. This is the single most important experiment to resolve the autoantibody controversy in ME/CFS.

3.5 References for Cross-Disease Autoantibody Comparison

For detailed evidence on schizophrenia autoantibody prevalence, unbiased screening methodology, and treatment response in autoantibody-confirmed subgroups, see Appendix H, Schizophrenia — Autoantibody Parallels to ME/CFS.

Speculation: The “Many Schizophrenias” Lesson for ME/CFS Subtyping

(Certainty: 0.40 — strong historical parallel, zero direct evidence in ME/CFS.)

Kraepelin and Bleuler both concluded from clinical observation that what they called “schizophrenia” was actually a group of diseases with different etiologies — “we should speak of schizophrenias in the plural” (Bartley and Ross 2020). This insight was marginalized for decades as the field pursued a single unifying model (dopamine, then NMDA). The cost was substantial: no mechanistically novel treatment for schizophrenia emerged between the 1970s and the 2020s.

The parallel for ME/CFS: if the field pursues a single unifying pathophysiological model (energy metabolism, neuroinflammation, autoimmunity, or any single mechanism) while ignoring subgroup heterogeneity, the same stagnation will occur. Autoantibody-based subgrouping — still controversial in ME/CFS — may ultimately prove as fruitful as it has in schizophrenia, but only if the field is willing to “speak of ME/CFSs in the plural.”

Falsifiable predictions: (1) The four autoantibody-defined subgroups will show non-equivalent responses to immunomodulatory therapy, tested by a significant interaction term (group × treatment, p < 0.05) in a biomarker-stratified RCT. (2) The double-negative subgroup will show no clinically meaningful response (effect size < 0.2 vs placebo), demonstrating that the apparent “treatment failure” of unselected trials (rituximab, BC007) reflects subgroup dilution rather than treatment inefficacy. Prediction (1) is falsified if the interaction term is non-significant (p ≥ 0.05). Prediction (2) is falsified if the double-negative subgroup shows a clinically meaningful response (effect size ≥ 0.3 vs placebo).

3.6 Extended Cross-Disease Autoantibody Hypotheses

Hypothesis: IgG Subclass Profiling Resolves the GPCR Autoantibody Contradiction in ME/CFS

(Certainty: 0.50.) The schizophrenia literature reveals a critical IgG subclass problem: among anti-NMDAR antibody-positive patients, the vast majority carry IgA/IgM (non-pathogenic), and only 0.5–1.5% carry IgG (the established pathogenic subclass) (Pollak et al. 2014) (Schou et al. 2016). Total autoantibody prevalence of 7.98% collapses to 1.46% IgG. ME/CFS GPCR autoantibody research uses total IgG ELISA (CellTrend), which detects all IgG subclasses combined. If ME/CFS autoantibodies follow the same pattern — mostly IgG4 (non-inflammatory, blocking) rather than IgG1 (inflammatory, complement-fixing) — then total IgG titers may misrepresent true pathogenic burden. IgG subclass-specific assays (IgG1, IgG2, IgG3, IgG4) would reveal which subclass drives ME/CFS pathology, enabling: (a) better patient selection for immunoadsorption, (b) correlation of subclass with clinical severity, (c) identification of a “pathogenic isotype signature.”

Evidence base. Pollak 2014 (meta-analysis: 7.98% positive any Ig class, only 1.46% IgG) (Pollak et al. 2014); Schou 2016 (n=925, 11.6% anti-neuronal Ab overall, only 0.5% NMDAR IgG) (Schou et al. 2016); Hartwig 2020 (ME/CFS IgG fails to activate β2-AdR — functional deficit) (Hartwig et al. 2020); existing paper ch28 (IgG4-IgG1 isotype switch speculation, certainty 0.42). No ME/CFS study has profiled GPCR autoantibodies by IgG subclass.

Falsifiable prediction. In n ≥ 100 ME/CFS patients with elevated GPCR autoantibodies (CellTrend-positive), IgG subclass profiling will show: (1) IgG1 and IgG3 (inflammatory, complement-fixing) subclasses correlate more strongly with disease severity (r ≥ 0.4) than total IgG (r ≤ 0.2); (2) IgG4 (non-inflammatory, blocking) autoantibodies are present in ≥ 30% of patients and negatively correlate with symptom severity; (3) the IgG1/IgG4 ratio discriminates immunoadsorption responders from non-responders with AUC ≥ 0.75. Falsified if IgG1/IgG3 correlate no more strongly with severity than total IgG, or if IgG4 shows no negative correlation with symptoms.

Speculation: ME/CFS Autoantibodies as Active BBB Disruptors Creating a Feed-Forward Loop

(Certainty: 0.40.) In schizophrenia, Nemani 2026 found autoantibodies against BBB antigens, suggesting that autoantibody-mediated BBB disruption is part of the disease mechanism — not just a consequence (Nemani et al. 2026). In ME/CFS, the existing paper models BBB permeability (P_BBB) as a function of inflammatory cytokines (TNF-α, IL-1β) with cooperative saturation kinetics, but does not include autoantibodies as direct BBB disruptors. If ME/CFS autoantibodies target BBB endothelial antigens (claudin-5, occludin, ZO-1), they would actively increase BBB permeability, creating a feed-forward loop: autoantibodies → BBB leak → CNS entry of peripheral autoantibodies + cytokines → neuroinflammation → more BBB damage → worse CNS symptom burden. This mechanism is distinct from the current cytokine-driven BBB leak model and would predict that BBB-stabilizing interventions (PEA, luteolin) must be combined with autoantibody reduction for efficacy, explaining the high interaction effect of P_BBB in global sensitivity analysis (ch32: S_T − S_1 ≈ 0.09).

Evidence base. Nemani 2026 (schizophrenia REAP: autoantibodies against BBB antigens) (Nemani et al. 2026); existing paper ch28 (BBB model P_BBB driven by TNF-α, IL-1β, no autoantibody term); ch32 (GSA: P_BBB interaction effect S_T − S_1 ≈ 0.09); ch14a (BBB vulnerability). No study has tested ME/CFS sera for BBB antigen autoantibodies.

Falsifiable prediction. ME/CFS sera (n ≥ 100) screened against BBB endothelial antigen panels (claudin-5, occludin, ZO-1, GLUT1, LRP1) by ELISA will show: (1) ≥ 15% of ME/CFS patients have autoantibodies against ≥ 1 BBB antigen vs ≤ 5% of controls; (2) BBB autoantibody-positive patients have elevated CSF/serum albumin ratio (Q_Alb > 7.0), indicating objective BBB disruption; (3) patient IgG increases endothelial monolayer permeability in vitro (Transwell assay, ≥ 20% increase vs control IgG); (4) BBB autoantibody status predicts preferential CNS symptom burden (cognitive ≥ 3 SD below norm) independent of peripheral GPCR autoantibody status. Falsified if no significant enrichment of BBB antigen autoantibodies in ME/CFS vs controls.

Speculation: Conformational Epitope Hypothesis Resolves the Germain 2025 REAP Null in ME/CFS

(Certainty: 0.40.) Germain 2025 screened 7,542 protein interactions using REAP + Luminex and found NO ME/CFS-specific autoantibody signal (Germain et al. 2025). This null is puzzling given replicated GPCR autoantibody findings by CellTrend ELISA (Freitag et al. 2020) (N. Azcue et al. 2026). The conformational epitope hypothesis resolves the discrepancy: REAP uses linear peptides expressed on phage (~50–150 amino acid fragments of extracellular domains); GPCRs have complex multi-pass transmembrane topology with highly conformational extracellular loops. CellTrend ELISA uses full-length GPCRs expressed on HEK cells (native conformation). If ME/CFS GPCR autoantibodies recognize conformational epitopes — dependent on proper 3D folding, glycosylation, and membrane embedding — they would be invisible to REAP/Luminex but detectable by cell-based ELISA. The schizophrenia parallel: NMDAR autoantibodies nearly always recognize conformational epitopes on native GluN1/GluN2 subunits — cell-based assays are the gold standard, not linear peptide ELISAs (Dalmau et al. 2008). This explains both why targeted (conformation-preserving) assays find GPCR autoantibodies in ME/CFS and why unbiased (linear-epitope) screening missed them.

Evidence base. Germain 2025 (REAP + Luminex, 7,542 interactions, no signal) (Germain et al. 2025); Freitag 2020 (Freitag et al. 2020), Azcue 2026 (N. Azcue et al. 2026) (CellTrend cell-based ELISA: GPCR autoantibodies found); Dalmau 2008 (NMDAR-Ab: conformational epitopes require cell-based assay) (Dalmau et al. 2008); Nemani 2026 (schizophrenia REAP: found non-GPCR targets) (Nemani et al. 2026).

Falsifiable prediction. Paired testing of ME/CFS sera (n ≥ 100) across three platforms — (1) CellTrend cell-based GPCR ELISA (conformation-preserving), (2) REAP/Luminex linear extracellular domain peptides, (3) in-house full-length GPCR expressed on HEK cells (independent validation) — will show: GPCR autoantibodies detected by platforms (1) and (3) but NOT by platform (2). Conversely, platform (2) will detect ≥ 3 novel linear-epitope targets (non-GPCR, likely ion channels or transporters) missed by both cell-based platforms. A fourth platform using membrane protein arrays (full-length GPCRs in nanodiscs) will partially rescue REAP detection for GPCR targets. Falsified if all platforms converge on equivalent null.

Speculation: Early Autoantibody Intervention Within a Reversible Therapeutic Window

(Certainty: 0.35.) The multi-attractor ODE model (ch33) defines attractor basins with autoantibody acquisition driving attractor migration. This raises a critical clinical question: is attractor migration reversible if autoantibodies are removed early, or does epigenetic consolidation create a one-way ratchet? The model predicts a therapeutic window of ~6 months after autoantibody acquisition during which early immunoadsorption could prevent irreversible attractor migration. Beyond this window, epigenetic consolidation deepens and migration becomes irreversible even if autoantibodies are cleared. Retrospective analysis of natural history data (Nacul 2020, Chu 2019) should show that patients who received early immunomodulatory treatment (within 2 years of autoantibody-positive test) have lower rates of progression to severe disease (OR ≤ 0.4) vs untreated patients. The schizophrenia parallel: Luykx 2024 found NMDAR-Ab+ patients had LESS severe symptoms — potentially representing a “not yet migrated” attractor state that remains treatment-responsive (Luykx et al. 2024).

Evidence base. Existing paper ch33 (attractor migration hypothesis, cert 0.40); ch28 (B cell model: autoantibody kinetics, plasma cell half-life); ch33 (timescale hierarchy: autoantibody τ ≈ months); Luykx 2024 (NMDAR-Ab+ schizophrenia: less negative symptoms, better functioning).

Falsifiable prediction. A threshold autoantibody burden θ_A (≥ 3 specificities with titer > 75th percentile) triggers transition from immune-dominant to neurovascular-dominant attractor. Attractor migration is reversible only if autoantibody burden is reduced below θ_A within τ_epi (epigenetic consolidation timescale, ≈ 6–12 months). After τ_epi, migration becomes irreversible. Retrospective analysis of natural history data will show OR ≤ 0.4 for early-treated vs untreated autoantibody-positive patients. Falsified if early autoantibody reduction does not reduce progression to severe disease.

Speculation: Autoimmune ME/CFS Diagnostic Checklist — Learning from Autoimmune Psychosis

(Certainty: 0.45.) The Neuropsychiatric Checklist for Autoimmune Psychosis (Tebartz van Elst 2025) integrates red-flag symptoms, FDG-PET, CSF, EEG, and MRI into a clinical scoring system. No equivalent exists for ME/CFS. A proposed Autoimmune ME/CFS Checklist would combine: clinical red-flags (infection trigger, sudden onset, family autoimmunity history, multi-system involvement), laboratory biomarkers (GPCR autoantibody elevation, CSF oligoclonal bands, BBB disruption markers, NK dysfunction), and imaging (FDG-PET hypometabolism, TSPO-PET neuroinflammation).

Evidence base. Tebartz van Elst 2025 (autoimmune psychosis checklist); Endres 2020 (145 AP cases, 94% immunotherapy response). No ME/CFS equivalent exists.

Falsifiable prediction. An Autoimmune ME/CFS Checklist (scored 0–20) will identify a subgroup (score ≥ 12) with: ≥ 2× higher GPCR autoantibody titers vs low-scorers; ≥ 40% IA response rate vs ≤ 10% in low-scorers; and a distinct FDG-PET frontotemporal hypometabolism pattern.

Speculation: Autoantibody-Driven Attractor Migration in the Multi-ODE Model

(Certainty: 0.35.) Extending the existing ODE model (ch33) with an autoantibody vector A(t) = (A_1(t), …, A_k(t)) — with subclass-resolved kinetics — would predict: (a) a critical burden threshold for attractor migration; (b) whether migration is reversible or a one-way epigenetic ratchet; (c) a timescale (months to years) determined by plasma cell half-life. The schizophrenia parallel: Luykx 2024 found NMDAR-Ab+ patients had less severe symptoms — a potential “pre-migration” state still treatment-responsive.

Falsifiable prediction. Patients below a critical autoantibody burden threshold recover to pre-migration state after immunoadsorption; patients above stabilize without full recovery. This tests whether early autoantibody removal prevents progression to the severe/locked attractor.

Speculation: REAP Proteome-Wide Autoantibody Screening — the Highest-Yield Single Experiment

(Certainty: 0.55.) Nemani et al. applied REAP (6,183 extracellular proteins) to 352 schizophrenia patients and 971 controls, finding ~2× autoantibody burden and novel targets (ion channels, synaptic proteins, BBB antigens) missed by targeted ELISAs for decades. All ME/CFS autoantibody research is target-driven (GPCR ELISA). The Germain 2025 REAP null used individual domain fragments, which may miss conformational epitopes from multi-loop GPCRs. A dedicated screen using full-length native protein libraries (n ≥ 200, infection-triggered cohort) is the single most important experiment to resolve the autoantibody controversy.

Falsifiable prediction. ME/CFS screened by REAP (>2,500 targets) will show higher autoantibody burden vs controls (d ≥ 0.3), ≥ 5 novel targets beyond the GPCR panel, including ≥ 2 ion channel or synaptic protein specificities. Replication cohort (n ≥ 100) confirms top 10 hits.

Speculation: Schizophrenia GWAS Glutamatergic Parallel — Same Pathway, Different Circuits

(Certainty: 0.45.) Schizophrenia GWAS also shows glutamatergic synapse enrichment — the same biological pathway identified by Maccallini 2026 at the gene-set level in ME/CFS (Maccallini 2026). The convergent signal suggests glutamatergic synaptic dysfunction is a shared genetic vulnerability, with circuit specificity determining clinical expression: prefrontal-temporal circuits in schizophrenia versus cortico-cerebellar and brainstem circuits in ME/CFS (Hirsch et al. 2025) (Wirth and Scheibenbogen 2026).

Pomaglumetad (mGluR2/3 agonist) failed in schizophrenia Phase III trials despite confirmed target engagement. This failure may reflect circuit mismatch: mGluR2/3 agonism modulates prefrontal glutamate release relevant in schizophrenia but not necessarily in ME/CFS, where cerebellar and brainstem circuits may be the primary vulnerability. Pomaglumetad’s failure in schizophrenia does not predict failure in ME/CFS if the target circuits differ. A glutamatergic modulator trial stratified by glutamatergic PRS in ME/CFS would test whether circuit-specific modulation benefits patients with high glutamatergic genetic loading (Maccallini 2026) (DecodeME Consortium, Ponting, et al. 2025).

Evidence base. Maccallini 2026 meta-GWAS (glutamatergic synapse enrichment); DecodeME single-gene hits in glutamatergic transmission (SHISA6, UNC13C); Hirsch 2025 comparative GWAS confirming glutamatergic enrichment across conditions (Hirsch et al. 2025).

Falsifiable prediction. ME/CFS patients stratified by high vs. low glutamatergic PRS will show differential response to glutamatergic modulators (memantine benefit in high-PRS subgroup), supporting circuit-specific glutamatergic pathology distinct from schizophrenia.

Speculation: Epilepsy: Glutamatergic Hyperexcitability Threshold

Epilepsy is the archetypal glutamatergic hyperexcitability disorder. Anti-epileptic drugs (lamotrigine, levetiracetam, zonisamide) are already discussed in Chapter Neurological and Neurocognitive Dysfunction for their microglial-suppressive and anti-kindling properties. Maccallini 2026 glutamatergic synapse enrichment provides a genetic rationale: ME/CFS may share a sub-convulsive hyperexcitability phenotype with epilepsy, manifesting as altered neural network excitability without overt seizure activity (Maccallini 2026).

(Certainty: 0.35)

Falsifiable prediction: Interictal-like discharges on sleep EEG (already documented as alpha-delta sleep in ME/CFS) will be more common in high Glu-PRS ME/CFS patients than low Glu-PRS patients, and sleep EEG abnormality burden will correlate with glutamatergic PRS.

Speculation: Autism: Cerebellar Development and Glutamatergic Circuits

Cerebellar Purkinje cell loss is among the most replicated neuropathological findings in autism. Both ME/CFS and autism show: (a) glutamatergic synapse genetic enrichment in GWAS, (b) Purkinje cell vulnerability to metabolic and inflammatory stress, (c) sensory hypersensitivity, and (d) autonomic dysfunction. Maccallini 2026 independent cerebellar neuronal signal provides a genetic bridge: if cerebellar glutamatergic dysfunction is a shared developmental vulnerability, it could explain the disproportionate ME/CFS prevalence in neurodivergent populations (Maccallini 2026).

(Certainty: 0.35)

Falsifiable prediction: A cerebellar cell-type PRS derived from Maccallini enrichment data will correlate with Sensory Profile questionnaire scores (especially auditory and tactile sensitivity) in both autism and ME/CFS cohorts, and will be elevated in ME/CFS patients with co-occurring autism traits.

Speculation: Angelman Syndrome (UBE3A) as a Cross-Disease Model for ME/CFS Mitochondrial and GABAergic Vulnerability

Angelman syndrome (AS) arises from loss of function of UBE3A (E6AP, an E3 ubiquitin ligase), producing a neurodevelopmental disorder dominated by intellectual disability, epilepsy, gait ataxia, and disrupted sleep (Buiting, Williams, and Horsthemke 2016). In an animal model, Ube3a loss impairs mitochondrial oxidative phosphorylation and lowers neuronal ATP (Su et al. 2011), and in silico work links UBE3A-dependent mitochondrial transcripts to energy metabolism (Panov et al. 2020). AS models also show elevated reactive-oxygen species (Simchi et al. 2023) and reduced tonic GABAergic (GABA-A) inhibition, an excitatory/inhibitory imbalance (Watanabe et al. 2026), alongside fragmented sleep architecture (Qu et al. 2024). Each of these domains — mitochondrial/energy deficit, oxidative stress, GABAergic E/I imbalance, and sleep fragmentation — has been independently proposed in ME/CFS (Le et al. 2025) (Shankar et al. 2025) (L. Davis et al. 2025) (Wirth and Scheibenbogen 2026) (Liu et al. 2025).

No direct clinical co-occurrence or mechanistic link between Angelman syndrome and ME/CFS has been reported (0 PubMed hits for AS–ME/CFS). The analogy is therefore hypothesis-generating, not an established shared pathophysiology. It is additionally limited by two structural disanalogies: (a) AS is a congenital, monogenic, neurodevelopmental disorder, whereas ME/CFS is adult-onset and polygenic, so the animal/in-silico evidence — which describes a developing nervous system — may not transfer to the adult-acquired setting; and (b) UBE3A dosage sensitivity — both loss (AS) and duplication (Dup15q) produce overlapping neurodevelopmental outcomes — means the proteostasis/dosage model is not specific to a single direction of UBE3A perturbation (Kurmashev 2026).

(Certainty: 0.20, cross-disease analogy, animal/in-silico/overview evidence discounted for non-ME/CFS population and for developmental stage.)

Falsifiable prediction. If the analogy holds, then ME/CFS should show impaired mitochondrial oxidative phosphorylation/ATP and reduced tonic GABA-A inhibition consistent with the AS-like phenotype, regardless of whether the deficit is acquired or genetic; UBE3A-pathway variant enrichment in glutamatergic/mitochondrial gene-set analyses is an optional corroborating (not falsifying) test, since an acquired ME/CFS deficit would show pathway pathology without genetic enrichment. Where such carriers are identified, iPSC-derived neurons should recapitulate the reduced tonic inhibition and lower ATP seen in AS models, though an iPSC null would be non-decisive given technical recapitulation limits.

Consequence: This speculation does not change current clinical care — it is a research hypothesis. If it holds, it would point researchers toward UBE3A/ubiquitin-pathway and GABAergic tonic-inhibition circuits as one possible shared vulnerability underlying energy and brain-fog symptoms, offering a concrete genetic and biochemical test. Severity applicability: unknown — no ME/CFS-population data; the mechanism, if real, would be expected to affect all severity levels without stratification.

Evidence base. Buiting 2016 (overview); Su 2011 (mitochondrial ATP, mouse); Panov 2020 (in silico mitochondrial transcriptome); Simchi 2023 (ROS); Watanabe 2026 (tonic GABA-A inhibition); Qu 2024 (sleep); Kurmashev 2026 (UBE3A dosage).

Limitation: HSAT2 Cluster: Single-Anchor Architecture Risk

All environments in the HSAT2 hypothesis cluster — across Chapters 14a, 14d, 16, 17, 18, and 20 — share a single foundational mechanistic anchor: Evdokimova et al. 2019 (bioRxiv, Ewing sarcoma, unpublished). If this anchor fails to replicate in non-cancer biology, the downstream speculations (exosomal propagation, MDSC expansion, NK suppression, epigenetic combination strategies) are individually unmotivated. The certainty values assigned to each environment are conditional on the anchor’s plausibility; readers should weight them accordingly.

Speculation: Shared Exosomal HSAT2/HERV-K Signature as a Convergent Mechanism in Long COVID and ME/CFS

(Certainty: 0.30 — both conditions share NK/T-cell exhaustion phenotypes; HERV transcriptional activation after SARS-CoV-2 is directly documented; checkpoint gene dysregulation overlap confirmed; the specific exosomal propagation loop remains a cross-disease inference. Certainty assigned: 0.30. The parent loop hypothesis (Exosomal HSAT2/HERV-K RNA as a Self-Perpetuating MDSC Expansion Mechanism) constrains the permissible range to 0.25–0.45; any revision to the parent’s certainty should be propagated here. All mechanistic anchors ultimately trace to a single 2019 bioRxiv preprint in Ewing sarcoma cancer biology.)

Long COVID and ME/CFS share post-infectious onset, persistent immune activation despite viral clearance, NK cell dysfunction, and T-cell exhaustion. Several lines of new evidence strengthen the convergence:

HERV reactivation confirmed in COVID-19: Grandi et al. found 282 HERV loci differentially expressed in COVID-19 PBMCs (n = 26), with convalescent patients showing a distinct persistent HERV transcriptional signature — confirming that SARS-CoV-2 infection triggers broad HERV de-silencing, consistent with the pericentromeric dismantling pathway
HERV-K antibody persistence in ME/CFS post-COVID: ME/CFS patients showed stronger and more persistent HERV-K IgG responses after mild/asymptomatic COVID-19 than healthy donors , consistent with ongoing epigenetic derepression in ME/CFS rather than recovered individuals
Shared PD-1/CTLA-4 checkpoint dysregulation: Eaton-Fitch et al. demonstrated that ME/CFS and Long COVID share 7 overlapping checkpoint dysregulation genes — ME/CFS more immunosuppressed, Long COVID more immune-activated — with the ME/CFS phenotype consistent with established MDSC-mediated suppression MDSC Expansion via Exosomal HSAT2 as the Upstream Cause of NK Cytotoxicity Loss

If the exosomal HSAT2/HERV-K propagation loop operates in both conditions:

Long COVID and post-infectious ME/CFS should share an elevated plasma exosomal HSAT2/HERV-K signature, distinguishing them from recovered post-COVID controls
The subgroup of Long COVID patients who progress to ME/CFS may be those whose exosomal loop failed to self-terminate
The ME/CFS vs. Long COVID difference in immune phenotype (suppressed vs. activated) could reflect MDSC dominance in ME/CFS versus insufficient MDSC formation in earlier-stage Long COVID
Treatments targeting the loop (NRTIs, methyl-donor support, EV depletion) could have cross-condition efficacy

This is the strongest cross-disease prediction of the HSAT2 hypothesis, because Long COVID provides a well-characterized post-infectious cohort where recovery vs. persistence can be studied prospectively.

Falsifiable prediction: Long COVID patients with persistent fatigue and PEM at ≥ 6 months post-infection will show higher exosomal HSAT2 than matched recovered post-COVID individuals. Exosomal HSAT2 at the 3-month timepoint will predict persistence vs. recovery at 12 months (AUROC > 0.70). If recovered and persistent Long COVID show equivalent HSAT2, the loop does not explain the bifurcation. The Eaton-Fitch 2024 ME/CFS-vs-Long COVID checkpoint divergence should track with exosomal HSAT2 levels if the MDSC-driven suppression model is correct.

Limitations: Grandi 2023 has no Long COVID or ME/CFS arm (n = 26, COVID-19 only). HERV-K antibody elevation (Apostolou 2022) is indirect and could reflect prior reactivation rather than ongoing expression. The ME/CFS-vs-Long COVID immune suppression/activation divergence (Eaton-Fitch 2024) has multiple plausible explanations beyond the HSAT2 loop. Cross-condition inference assumes equivalent viral de-silencing mechanisms for SARS-CoV-2 and ME/CFS-triggering viruses. Not replicated.

Speculation: Fibromyalgia: Partial HSAT2 Overlap via HSF1-Oxidative Axis Without the MDSC Arm

(Certainty: 0.20 — no HSAT2 or MDSC data in fibromyalgia; mechanistic distinction inferred from different immunological profiles.)

Fibromyalgia shares oxidative stress elevation and neuroinflammatory features with ME/CFS, but the NK cytotoxicity loss that defines the ME/CFS immune signature is not consistently reported in fibromyalgia. If the HSAT2 derepression pathway operates via two arms — HSF1-driven transcription (activated by oxidative/heat stress) and MDSC expansion (activated by EV-delivered HSAT2 in myeloid cells) — fibromyalgia may share the first arm without the second.

This predicts a partial HSAT2-overlapping profile: fibromyalgia would show elevated cell-autonomous HSAT2 transcription driven by oxidative stress, but with normal MDSC frequency and preserved NK cytotoxicity. ME/CFS would show both elevated cellular HSAT2 AND elevated MDSCs AND depressed NK cytotoxicity. This mechanistic distinction could explain why fibromyalgia does not show the characteristic NK depletion — and why treatments targeting the MDSC arm (arginase-1 inhibitors, PDE5 inhibitors) would not be predicted to benefit fibromyalgia.

Falsifiable prediction: Fibromyalgia plasma EV HSAT2 will be elevated versus healthy controls (AUC > 0.65) but MDSC frequency (CD33+HLA-DR−/lo) will not differ from controls. ME/CFS will show both elevated EV HSAT2 and elevated MDSCs. This head-to-head comparison would mechanistically distinguish the two conditions.

Limitations: Zero HSAT2 or MDSC data in fibromyalgia. NK cytotoxicity in fibromyalgia is not consistently measured. Mechanistic reasoning is cross-condition extrapolation. Not replicated.

Open Question: Does the CENPA/Senescence Stromal Arm of the HSAT2 Loop Partially Operate in hEDS and Dysautonomia?

Hypermobile Ehlers-Danlos syndrome is characterized by connective-tissue stromal involvement — fibroblasts and extracellular matrix remodeling are central to its phenotype. The proposed HSAT2→CENPA mislocalization→p53-dependent senescence chain in stromal fibroblasts [NO SPECULATION] predicts an elevated burden of senescent fibroblasts in any tissue where EV-delivered HSAT2 acts. If hEDS fibroblasts are constitutively abnormal (altered ECM gene expression, increased mechanical sensitivity), they may be more susceptible to EV-HSAT2-triggered CENPA induction and subsequent senescence.

This does not make hEDS an HSAT2-driven disease; the primary connective tissue pathology in hEDS is independent of the HSAT2 loop. The question is whether the downstream stromal senescence arm partially co-activates in hEDS patients with comorbid ME/CFS — contributing to the connective tissue features seen in that overlap group — rather than being driven by the HSAT2 mechanism de novo.

The distinction between a primary HSAT2-high ME/CFS patient and an hEDS patient who coincidentally experiences stromal senescence would be testable: ME/CFS patients should show elevated NK cytotoxicity deficits and MDSC expansion alongside the senescence score CD33+HLA-DR− MDSC Frequency as a Functional Readout of the HSAT2 Loop, while hEDS patients without ME/CFS would show normal NK cytotoxicity and normal MDSC frequency despite any elevated senescent fibroblast burden.

What would establish this: Skin biopsy senescence score (p16^INK4a+ fibroblast burden, SA-β-gal) in four groups: ME/CFS only, hEDS only, ME/CFS+hEDS, healthy controls — with concurrent NK cytotoxicity and MDSC quantification from blood. The ME/CFS-specific prediction is elevated senescence score AND elevated MDSCs AND depressed NK function; hEDS-only prediction is mildly elevated senescence score but normal NK and MDSC.

Limitations: Zero skin biopsy or NK/MDSC data in hEDS without comorbid ME/CFS. hEDS fibroblast gene expression is abnormal at baseline; CENPA/senescence assays may need fibroblast-passage-standardized protocols. hEDS diagnostic criteria remain controversial regarding specificity . Not explored in any hEDS or dysautonomia context. Certainty: 0.15.

Speculation: ME/CFS as a “Cancer Immune-Evasion Environment Without the Cancer”: MDSC-NK Suppression as a Shared Mechanism

(Certainty: 0.30 — MDSC-NK suppression mechanism cross-validated across cancer, chronic HCV, and post-COVID; ME/CFS MDSC data absent.)

The MDSC-mediated NK suppression machinery that enables cancer immune evasion is well-established . In solid tumors, MDSCs expand from myeloid precursors in response to tumor-derived signals, create an immunosuppressive microenvironment, suppress NK and CD8+ cytotoxicity, and facilitate escape from immune surveillance. The entire suppressive apparatus — arginase-1 depletion of L-arginine, iNOS-derived reactive nitrogen, TGF-β, and IL-10 — is the same pathway proposed to operate in ME/CFS via EV-HSAT2-driven MDSC expansion [NO SPECULATION].

The “inverse mirror” framing: in cancer, MDSCs protect a malignant population from immune clearance; in ME/CFS, MDSCs may be expanded by the same general myeloid-programming mechanism (EV-delivered danger signals) but in the absence of a tumor. The immunosuppressive milieu is reproduced by the HSAT2-EV cargo acting as a tumor-analogue signal rather than by the tumor itself. ME/CFS would represent a pathological state where the immune-evasion infrastructure is activated chronically without a survival benefit for any specific cell population — a self-sustaining immune suppression loop without an object.

This analogy has a testable molecular prediction: if ME/CFS MDSCs are transcriptomically equivalent to cancer-associated MDSCs (tumor microenvironment-type), tadalafil and other MDSC-differentiating agents from oncology would be predicted to work [NO SPECULATION]. If ME/CFS MDSCs cluster more closely with chronic-viral MDSCs (HCV, post-COVID type), the molecular targets may differ subtly. If they cluster with tumor-associated MDSCs despite post-viral context, EV-HSAT2 as a tumor-analogue signal is supported.

Falsifiable prediction: ME/CFS PBMC bulk RNA-seq or single-cell RNA-seq (MDSC gate) will show a transcriptomic profile clustering with chronic-viral MDSCs (HCV, post-COVID) rather than tumor-associated MDSCs in a multi-disease reference atlas. If ME/CFS MDSCs cluster with tumor-associated MDSCs, the “cancer-analogue signal” hypothesis is strengthened.

Limitations: No ME/CFS MDSC data exists; the cross-disease analogy is entirely proxy-based. MDSC transcriptomics vary substantially by tumor type and chronic virus species; the reference atlas itself may not contain an appropriate post-viral chronic disease comparator. Analogy is not equivalence — even if the transcriptomics cluster, downstream treatment implications require prospective testing. Not replicated in ME/CFS or any post-viral chronic fatigue condition.

Simpler competing explanation: chronic antigen stimulation (persistent viral epitopes, autoantigen presentation) drives MDSC expansion in ME/CFS without requiring EV-HSAT2 as a tumor-analogue signal. This simpler mechanism is well-supported in HCV, HIV, and post-COVID contexts and would produce the same MDSC-NK phenotype without invoking the cancer-mirror analogy. The EV-HSAT2 cargo signal is an additional layer, not a required one.

Speculation: Accelerated Biological Aging in ME/CFS as a Measurable Proxy for HSAT2 Inflammaging Burden

(Certainty: 0.35 — epigenetic aging acceleration in ME/CFS has preliminary support; HSAT2 as mechanistic driver of acceleration is indirect inference.)

De Cecco et al. established in healthy aging that LINE-1 derepression tracks with epigenetic clock acceleration (GrimAge); repeat de-silencing and biological aging advance together. The HSAT2 inflammaging model [NO SPECULATION] proposes that ME/CFS represents an acute post-viral short-circuit of the same process. This predicts measurable biological age acceleration in ME/CFS, with the acceleration correlating with HSAT2 expression levels.

Early epigenetic clock studies in ME/CFS are consistent with this picture (preliminary data suggest 3–8 year GrimAge acceleration), though sample sizes are small and methodology varies. The prediction is testable with existing methods: GrimAge or DunedinPACE measurements in a cohort where EV HSAT2 is simultaneously quantified would directly test the HSAT2-inflammaging chain.

Falsifiable prediction: ME/CFS patients (mean chronological age 35–45) will show GrimAge or DunedinPACE biological age 3–8 years above matched controls; this acceleration will correlate (Spearman ρ > 0.4) with plasma EV HSAT2 RNA load in the same cohort.

Limitations: Epigenetic clock estimates in ME/CFS are inconsistent across published studies; methodology (clock type, blood cell composition correction) varies. HSAT2 specifically is not the causal driver of epigenetic clock acceleration in De Cecco 2019 (LINE-1 is); HSAT2 is proposed as a parallel locus with analogous mechanism. Not replicated.

Cross-disease retrotransposon activation: Evidence summary table.

Table Accelerated Biological Aging in ME/CFS as a Measurable Proxy for HSAT2 Inflammaging Burden summarizes the current state of HSAT2 and retrotransposon research across ME/CFS and related conditions. This comparative framework is essential for diagnostic specificity assessment.

Cross-disease evidence summary for HSAT2 and retrotransposon activation
[Disease, Evidence for HSAT2/Retrotransposon Activation, Certainty, Mechanistic Link to ME/CFS]	[	[Cancer (pancreatic, colon), HSAT2 validated as biomarker; TRAP-ddPCR and hybridization capture methods established , 0.60–0.65, HSAT2 exosome transfer drives MDSC expansion and NK suppression in cancer ; same pathway proposed in ME/CFS [NO SPECULATION]]	[Long COVID, HERV activation documented (HERV-W ENV correlates with severity) ; HERV transcriptome tracks clinical stages, 0.45, Direct HSAT2 measurements absent; EV-mediated retrotransposon transfer proposed as shared mechanism [NO SPECULATION]]
[Fibrotic diseases (IPF, liver, cardiac), No HSAT2 or retrotransposon literature found, 0.00, None established; hypothetical link via HSF1-oxidative stress axis in fibromyalgia [NO SPECULATION]]	[Post-viral conditions (EBV, CMV, other herpesviruses), No HSAT2 literature found; HERV data suggests class-wide activation, 0.30, Proposed as trigger for sustained HSAT2 activation analogous to COVID-19 HERV findings]	[ME/CFS, Exosomal HSAT2/HERV-K hypothesis; MDSC expansion and NK suppression documented, 0.30–0.45, Proposed as “cancer immune-evasion environment without cancer” [NO SPECULATION]]	]

Hypothesis: HSAT2 as a Pan-Retrotransposon Stress Response vs Disease-Specific Signature

(Certainty: 0.45 — supported by HERV COVID-19 data and HSAT2 cancer progression data ; ME/CFS-specific pattern not yet measured)

Retrotransposon activation (HSAT2 in humans, HERV class-wide) may represent a conserved cellular stress response to viral infection or inflammatory insult, with disease-specific patterns emerging from tissue context and chronicity. In acute viral infection (COVID-19), HERV activation is transient and stage-specific . In cancer, HSAT2 is progressively elevated and correlates with tumor stage (early vs late-stage discrimination) . ME/CFS may represent an intermediate state: elevated but stable rather than progressive.

This pan-retrotransposon stress response framework predicts measurable differences in kinetics across diseases:

Acute viral infection: Retrotransposon activation transient, tracks clinical stage (HERV in COVID-19)
Cancer: Retrotransposon elevation progressive, correlates with disease stage (HSAT2 in pancreatic/colon cancer)
Post-viral syndromes (ME/CFS, Long COVID): Retrotransposon elevation at acute trigger, then plateau at persistent intermediate level

Falsifiable predictions:

Long COVID patients with persistent fatigue at ≥6 months post-infection will show elevated HSAT2 that remains stable over time (unlike cancer progression), distinguishing ME/CFS-like Long COVID from recovered Long COVID. The Eaton-Fitch 2024 immune checkpoint divergence between ME/CFS and Long COVID should track with exosomal HSAT2 levels if the MDSC-driven suppression model [NO SPECULATION] is correct,
ME/CFS patients will show elevated HSAT2 that correlates with symptom severity but not with disease duration. Longitudinal tracking over 12–24 months will show plateaued elevation (trend slope not different from zero, p > 0.05) despite symptom fluctuation, distinguishing ME/CFS from cancer.
Antiviral therapy (valganciclovir, ganciclovir) responders among EBV/CMV-onset ME/CFS patients will show HSAT2 reduction correlated with symptom improvement (Spearman ρ>0.4). Non-responders will maintain elevated HSAT2 despite viral load suppression, suggesting epigenetic locking.

Limitations: No direct EBV/CMV→HSAT2 data exists; inference is based on COVID-19 HERV activation pattern , which may not generalize across herpesviruses. Antiviral trials in ME/CFS have shown mixed results; confounding factors (viral suppression vs immune modulation) are unresolved. Not replicated.

Hypothesis: Post-Acute HSAT2 Persistence vs Cancer Progression: Kinetic Distinction

(Certainty: 0.40 — HSAT2 kinetics established in cancer ; inferred for post-viral syndromes)

HSAT2 kinetics may distinguish disease categories. In cancer, HSAT2 levels progressively rise with tumor stage . In post-viral syndromes (Long COVID, ME/CFS), HSAT2 may elevate acutely then plateau at a persistent intermediate level. This kinetic distinction could serve as a diagnostic rule-out: progressive HSAT2 elevation → suspect malignancy; plateaued elevation → suspect post-viral syndrome.

Falsifiable prediction: Longitudinal tracking of HSAT2 in two cohorts: (a) cancer patients (n=50, serial measurements over treatment) will show monotonic increase or decrease correlating with tumor burden; (b) ME/CFS/Long COVID patients (n=50, 12-month follow-up) will show stable or fluctuating levels without monotonic trend (trend slope not different from zero, p > 0.05). Kinetic modeling will achieve >80% classification accuracy.

Limitations: No longitudinal HSAT2 data in ME/CFS or Long COVID exists; this is a proposed diagnostic framework requiring validation. Cancer kinetics data come from cross-sectional staging rather than true longitudinal. Confounding factors (treatment effects on HSAT2) unknown in cancer. Not replicated.

Speculation: Fibrotic Disease HSAT2 Gap: HSF1-Mediated Derepression Hypothesis

(Certainty: 0.25 — HSF1-HSAT2 pathway established ; fibrotic HSAT2 data absent)

Fibrotic diseases (IPF, liver fibrosis, cardiac fibrosis) share HSF1 activation and oxidative stress pathways with ME/CFS . HSAT2 is HSF1-regulated; if HSF1 drives HSAT2 derepression in fibrosis as in ME/CFS, then (a) fibrotic disease patients should show elevated HSAT2, and (b) HSAT2 levels should correlate with fibrosis severity scores. The absence of HSAT2 literature in fibrosis may reflect measurement gaps rather than true absence.

Falsifiable prediction: IPF patients (n=30) will show elevated serum HSAT2 RNA compared to matched COPD controls (non-fibrotic lung disease). HSAT2 levels will correlate with GAP scores (ρ>0.35). If HSAT2 is absent in IPF despite HSF1 activation, disease-specific co-factors (tissue-specific epigenetic context) are dominant over HSF1-HSAT2 axis.

Limitations: Zero HSAT2 or retrotransposon literature in fibrotic diseases; hypothesis is entirely extrapolative. HSF1 activation in fibrosis is well-documented, but tissue-specific epigenetic context may prevent HSAT2 derepression. Not replicated.

Open Question: Exosomal HSAT2 Tissue Origin: Stromal vs Immune Cell Source

Evdokimova et al. shows exosomal HSAT2/HERV-K in cancer is packaged by tumor cells. In ME/CFS, the tissue origin is unknown: are EVs from immune cells, stromal fibroblasts, endothelial cells, or neuronal tissue? Tissue origin affects pathogenic potential (immune-modulating vs tissue-damaging) and informs treatment targeting.

Research approach: Flow cytometry with cell-type-specific surface markers (CD45 for immune, CD90 for fibroblast, CD31 for endothelial) on EV preparations from ME/CFS plasma, followed by single-cell RNA-seq on sorted EVs to identify HSAT2 expression concentration by cell type.

Potential impact: If EV-HSAT2 is primarily immune cell-derived, therapies targeting immune suppression (MDSC depletion, T-cell exhaustion reversal) are most rational. If EV-HSAT2 is stromal-derived, antifibrotic or connective-tissue-targeted approaches may be relevant. Not yet tested.

Speculation: HSAT2 Methylation as a Reversible ME/CFS Epigenetic Lock

(Certainty: 0.20 — HSAT2 methylation patterns established ; DNMT3A overexpression shown to silence repetitive elements)

If HSAT2 activation in ME/CFS is maintained by DNA hypomethylation, then targeted remethylation (DNMT3A activators, SAMe supplementation) could reverse HSAT2 expression and potentially improve symptoms. This is distinct from cancer where hypomethylation is more extensive and less reversible.

Falsifiable prediction: In vitro treatment of HSAT2-expressing cells (primary fibroblasts or immune cells from ME/CFS patients) with DNMT3A activator (RG-108) will increase HSAT2 promoter methylation (>2-fold) and reduce HSAT2 RNA (>50%) within 72 hours. In a small clinical trial (n=15), oral SAMe (1600 mg/day, 8 weeks) will reduce plasma HSAT2 and correlate with symptom improvement (effect size d≥0.8).

Limitations: DNMT3A activators are oncogenic risks; clinical use in ME/CFS would require careful monitoring. SAMe supplementation trials in ME/CFS have shown mixed results. HSAT2-specific methylation in patient cells not tested. Not replicated.

Recommendation: Cross-Disease HSAT2 Network Analysis: Identifying Shared vs Unique Pathways

Bioinformatic analysis of HSAT2-interacting proteins and regulatory networks across disease transcriptomes could reveal (a) shared retrotransposon response pathways (HSF1, CTCF, DNMTs), and (b) disease-specific co-factors (tissue-specific TFs, viral-specific integration sites). This network approach would guide targeted interventions.

Research design: Compare transcriptomes from ME/CFS, Long COVID, pancreatic cancer, and IPF. Identify (a) core HSAT2 co-expression module (10–15 genes) conserved across ≥3 diseases, and (b) disease-specific modules unique to each condition. Core module should be enriched for HSF1 targets (FDR < 0.01). Disease-specific modules would reveal tissue or virus-specific co-factors.

Potential outcomes: Shared pathway validation strengthens pan-retrotransposon stress response hypothesis. Disease-specific modules identify novel therapeutic targets (e.g., fibrotic-specific co-factors). Network-based drug repurposing candidates for core pathway modulation.

Speculation: Retrotransposon-Induced Autoimmunity as ME/CFS-Specific Mechanism

(Certainty: 0.15 — autoimmunity is established ME/CFS feature; retrotransposon immunogenicity plausible but untested)

If HSAT2/HERV-K expression in ME/CFS (but not cancer) triggers autoantibody production via molecular mimicry or nucleic acid immune complex formation, this could explain ME/CFS-specific features (autoimmunity, post-exertional malaise) absent from cancer. Retrotransposon-derived RNA/protein could act as neoantigens.

Falsifiable prediction: ME/CFS serum will show higher anti-HSAT2 protein or anti-HSAT2 RNA antibody titers compared to healthy controls and cancer patients (effect size d≥1.2). Antibody levels will correlate with symptom severity (ρ>0.3) but not with disease duration. Long COVID patients will show intermediate titers, suggesting partial overlap.

Limitations: Autoimmunity in ME/CFS has multiple plausible explanations beyond retrotransposon trigger. No direct HSAT2-autoimmunity link exists. Retrotransposon protein immunogenicity is theoretical. Not replicated.

4 Viral Clearance Dynamics: Long COVID vs ME/CFS

Building on the clinical similarities and differences outlined above, a detailed comparison of viral clearance mechanisms reveals important distinctions between Long COVID and ME/CFS that may explain divergent natural histories and treatment responses.

Viral Detection and Persistence Patterns.

Aspect	Long COVID	ME/CFS
Primary trigger	SARS-CoV-2 (single pathogen)	Multiple pathogens (EBV, HHV-6, enteroviruses, etc.)
Viral detection	Spike protein and RNA detectable in gut tissue up to 12 months	No consistent viral detection in blood/tissue
Antibody patterns	Variable: some show waning, others persistent	Elevated herpesvirus antibodies common
Time since onset	Typically months to 1-2 years	Often years to decades
Viral clearance evidence	Mixed: some show clearance, others persistence	No evidence of active replication

Immune Response Differences.

Long COVID Immune Profile:

More activated immune phenotype
Elevated but not exhausted T cells
Persistent inflammatory signatures
Evidence of ongoing immune response to residual antigen

ME/CFS Immune Profile:

More exhausted/suppressed phenotype
Marked T cell and NK cell exhaustion
Mixed inflammation (some markers elevated, others suppressed)
Evidence of failed immune surveillance

Potential Explanations for Divergence.

Time Course Hypothesis: Long COVID represents an earlier stage of post-viral pathology that may progress to ME/CFS-like exhaustion over time
Pathogen-Specific Effects: SARS-CoV-2 may trigger different immune responses than herpesviruses or other ME/CFS triggers
Host Susceptibility: Different genetic or epigenetic susceptibility factors may determine outcome
Treatment Effects: Early COVID treatments (antivirals, immunomodulators) may alter disease course
Heterogeneity: Both conditions likely encompass multiple subtypes with different mechanisms

Hypothesis: Long COVID to ME/CFS Progression Model

Certainty: 0.35. A subset of Long COVID patients will progress to a ME/CFS-like state characterized by immune exhaustion, persistent energy limitation, and established metabolic dysfunction. The rate and predictors of this progression can identify high-risk patients for early intervention.

Evidence Base. The Eaton-Fitch 2024 finding that ME/CFS shows more immunosuppressed and Long COVID more immune-activated phenotypes suggests a potential progression: activated → exhausted. Longitudinal data from other post-viral conditions show similar progression patterns over years.

Progression Markers. Potential predictors of progression from Long COVID to ME/CFS:

Immune exhaustion markers: Increasing PD-1, CTLA-4, Tim-3 expression over time
Metabolic deterioration: Worsening mitochondrial function, energy crisis
Autonomic dysfunction: Development or worsening of POTS, orthostatic intolerance
Neurological symptoms: Progression from mild cognitive symptoms to significant dysfunction
PEM development: Emergence of classic post-exertional malaise pattern

Clinical Implications. If progression occurs, early identification of high-risk Long COVID patients could:

Enable targeted interventions to prevent progression
Inform prognosis and treatment planning
Guide research on progression mechanisms
Support disability claims and workplace accommodations

Testable Predictions.

Long COVID patients with high exhaustion marker levels at 6 months will meet ME/CFS criteria at 24 months
Immune activation-to-exhaustion transition will correlate with functional decline
Early intervention (within 6 months) will prevent progression in high-risk patients
Biomarker panels will predict progression with >75% accuracy

Limitations. Not all Long COVID patients may progress; some may recover or remain stable. The progression timeline and predictors are unknown. ME/CFS and Long COVID may be distinct conditions rather than progression stages.

Treatment Response Implications.

The different immune phenotypes suggest different therapeutic approaches:

Long COVID (More Activated):

May benefit from anti-inflammatory agents
Immunomodulation to reduce overactive responses
Antiviral therapy if viral persistence present
Targeted symptom management

ME/CFS (More Exhausted):

May require immune enhancement rather than suppression
Metabolic support and energy optimization
Neuromodulation and CNS-targeted approaches
Pacing and energy management

Research Priorities.

Key questions for distinguishing these conditions:

What biomarkers reliably differentiate Long COVID from ME/CFS?
What factors determine which Long COVID patients recover vs progress?
Can early intervention prevent progression to ME/CFS?
Are there distinct subtypes within each condition?
How should treatment be tailored to immune phenotype?

Open Question: Continuum or Distinct Entities?

Are Long COVID and ME/CFS points on a continuum of post-viral pathology, or are they distinct entities with overlapping symptoms? If they represent a continuum, what factors determine progression from early Long COVID to established ME/CFS? If they are distinct, what are the distinguishing biological features? Answering this question requires longitudinal studies tracking Long COVID patients over years with comprehensive immune and metabolic profiling.

Post-Treatment Lyme Disease Syndrome. Chronic symptoms after Lyme treatment:

Fatigue, cognitive dysfunction, pain
Controversial whether active infection persists
Speculative link: Borrelia may trigger same epigenetic/autoimmune locks; the specific pathogen matters less than the host response pattern

Post-Dengue Fatigue Syndrome. Chronic fatigue following dengue infection:

Well-documented in endemic areas
Similar symptom profile to ME/CFS
Speculative link: Dengue’s immune evasion strategies may be particularly effective at triggering the “safe mode” lock

Gulf War Illness. Multi-symptom illness in Gulf War veterans:

Fatigue, cognitive problems, pain, GI symptoms
Multiple potential triggers (infections, chemical exposures, vaccines, stress)
Speculative link: Multiple simultaneous stressors may be more likely to establish multiple locks simultaneously

Post-Meningitis Fatigue Syndrome. Chronic fatigue following viral or bacterial meningitis:

31–34% of meningitis survivors report persistent fatigue at 1–2 year follow-up, independent of acute illness severity
53% report rapid exhaustion after cognitive effort—consistent with post-exertional cognitive malaise as described in ME/CFS, though ME/CFS diagnostic criteria were not applied
CFS prevalence of 12.6% after viral meningitis in a small controlled study ($n = 83$), though non-significant after adjustment (OR 1.4, CI 0.5–3.6)
Consistent with the Dubbo cohort finding that post-infectious CFS develops at $~11%$ regardless of pathogen identity (EBV, Coxiella, Ross River virus—meningitis pathogens were not studied but the rate is comparable)
Speculative link: Meningitis causes direct neuroinflammation, blood-brain barrier disruption, and glymphatic system dysfunction ; these overlap with documented features of ME/CFS (see Chapter [NO CHAPTER], Sections [NO SECTION] and [NO SECTION]). Unlike peripheral infections, meningitis delivers the inflammatory insult directly to the CNS, potentially explaining the high rates of persistent cognitive symptoms. The implications of microglial priming for meningitis in existing ME/CFS patients are explored in Speculation [NO SPECULATION]

Bacterial infections as ME/CFS triggers: the Q fever paradigm. The post-infectious trigger literature is dominated by viral pathogens, but the Q Fever Fatigue Syndrome (QFS) literature provides strong evidence that bacterial infections can trigger persistent post-infectious illness. In a 10-year follow-up of 368 QFS patients, 98.9% experienced PEM, with a median of 12 chronic complaints and severely compromised quality of life (EQ-5D-5L: 0.63) . QFS is caused by Coxiella burnetii, an intracellular bacterium—demonstrating that neither viral persistence nor viral immune evasion is necessary for chronic post-infectious sequelae. This broadens the post-infectious trigger framework beyond viruses and supports the pathogen-agnostic host-response model from the Dubbo cohort .

Open Question: Common Post-Infectious Pathway?

What if all these conditions—ME/CFS, Long COVID, post-Lyme, post-meningitis fatigue, Q fever fatigue syndrome, Gulf War Illness—represent the same underlying “locked sickness behavior” state triggered by different insults? The specific trigger might influence which symptoms predominate, but the core pathophysiology could be identical. This would explain why they’re so similar clinically yet have different apparent causes. The inclusion of both viral (EBV, SARS-CoV-2, enteroviruses) and bacterial (Coxiella, Neisseria, Borrelia) triggers, as well as non-infectious insults (chemical exposures, vaccination stress), strengthens the case for a pathogen-agnostic host-response mechanism.

Limitation: Cross-Disease Extrapolation: Shared Symptoms Do Not Establish Shared Mechanisms

This chapter maps symptom overlaps between ME/CFS and numerous other conditions (Long COVID, POTS, MCAS, EDS, fibromyalgia, autoimmune diseases, mitochondrial diseases, cancer cachexia). While clinically suggestive, shared symptoms across conditions do not establish shared etiology. Key epistemic boundaries:

Symptom overlap may reflect convergent endpoints from distinct mechanisms rather than a common pathway; fatigue, cognitive dysfunction, and autonomic symptoms are final common manifestations of many unrelated pathologies.
The “speculative links” offered for each condition are hypothesis-generating, not evidence-based—none has been tested by studies designed to distinguish shared mechanism from coincidental co-occurrence.
Prevalence estimates for comorbidity (e.g., MCAS in ME/CFS, SFN in EDS) vary widely across studies due to inconsistent diagnostic criteria, referral bias, and lack of population-based data.
The “multi-lock” framing applied to cross-disease connections has not been empirically validated as a model distinct from simpler explanations such as shared genetic susceptibility or diagnostic overlap.

5 Bimodal Onset: Cross-Disease Lessons from Vitiligo

ME/CFS shares the rare feature of bimodal age at onset with a small number of other conditions, most notably autoimmune vitiligo, Hodgkin lymphoma, and inflammatory bowel disease . The combination of adolescent and early-middle-age peaks appears distinctive for ME/CFS, though systematic comparison of peak-age combinations across all bimodal-onset diseases has not been performed .

The vitiligo precedent.

Autoimmune vitiligo provides the strongest precedent for what bimodal onset analysis can reveal about disease biology. Jin et al. demonstrated that vitiligo has two onset-age peaks (mean 10.3 and 34.0 years) and that early-onset vitiligo harbours a specific MHC class II enhancer variant haplotype (rs145954018del-rs9271597A) with an extraordinary odds ratio of 8.10 ($p = 2.4 \times 10^{-86}$). This haplotype upregulates HLA-DQB1 expression in monocytes and dendritic cells. The late-onset subgroup lacked this genetic signal entirely, establishing that bimodal onset can reveal fundamentally distinct genetic architectures within a single clinical diagnosis.

Implications for ME/CFS.

The vitiligo finding motivates a specific research program for ME/CFS:

Genetic stratification: GWAS of early-onset vs. late-onset ME/CFS could reveal subtype-specific genetic signals, as it did in vitiligo
HLA analysis: The DecodeME dataset ($n gt 17{,}000$) contains genetic data that could be stratified by onset age; HLA fine-mapping would be the logical first target given the vitiligo precedent
Mechanistic divergence: The clinical differences between early and late onset (severity, triggers, familial clustering) are consistent with distinct biological substrates, though this remains unproven

Speculation: Early-Onset ME/CFS Has a Distinct Genetic Architecture Revealed by Bimodal Stratification

The vitiligo precedent demonstrates that bimodal disease onset can unmask genetic subtypes with very large effect sizes (OR > 8) . The clinical differences between early and late onset in ME/CFS (severity OR 2.15, infectious trigger enrichment, familial clustering OR 1.43) are consistent with early-onset ME/CFS having stronger genetic loading, as the familial aggregation signal suggests. The infectious mononucleosis association (OR 2.32 for early onset) may reflect age-specific immune vulnerability (adolescent thymic output, EBV primary infection timing) interacting with genetic susceptibility at HLA or immune-regulatory loci. Certainty: 0.40 (strong cross-disease precedent from vitiligo; clinical correlates consistent with genetic substructure; no direct evidence yet from ME/CFS GWAS stratified by onset age). Testable prediction: GWAS of DecodeME participants stratified by onset age (early: $\leq$ 20 years vs. late: $\geq$ 30 years) will show at least one locus with significantly different allele frequency between subgroups. HLA region is the most likely candidate based on vitiligo precedent and existing ME/CFS HLA associations.

Limitation: Cross-Disease Analogy Limits: ME/CFS Is Not Vitiligo

The vitiligo analogy, while mechanistically suggestive, has important limitations. Vitiligo is a classical autoimmune disease with well-characterized HLA associations; ME/CFS has no such established autoimmune etiology. The bimodal peaks in ME/CFS (16 and 37) differ from vitiligo (10 and 34) and may reflect different biological processes entirely. The familial clustering signal in ME/CFS (OR 1.43) is far weaker than the genetic signal in early-onset vitiligo (OR > 8), suggesting that if a genetic subtype exists in early-onset ME/CFS, its effect sizes are likely modest. The bimodal pattern could also reflect age-specific exposure (EBV infection timing) rather than genetic susceptibility windows.

Why these specific ages? Two mechanistic hypotheses.

Hypothesis: Thymic–EBV Synchrony: The Early Peak as a Pubertal “Perfect Storm” Window

Age ~16 coincides with three simultaneous biological events: (1) accelerating thymic involution (thymus halves in size during puberty; naive T cell output drops sharply), (2) peak EBV primary infection incidence (ages 15–19 in Western populations), and (3) pubertal sex hormone surge. The early onset peak may represent EBV primary infection occurring during the narrow window when thymic output is collapsing. EBV-driven B cell transformation creates a large pool of novel EBV-specific memory B cells just as thymic output of naive regulatory T cells (nTregs) is declining. Without adequate nTreg replenishment, EBV-induced autoreactive clones that would normally be suppressed escape peripheral tolerance. This predicts: (a) early-onset patients will have lower recent thymic emigrants (CD31+ naive CD4) than age-matched controls; (b) the ratio of RTEs to EBV-specific memory B cells at seroconversion will predict ME/CFS development; (c) HLA variants affecting thymic selection will be enriched in early-onset ME/CFS. Certainty: 0.45 (mechanistically coherent; consistent with McGrath IM enrichment OR 2.32, Katz 2009 ~13% CFS after IM, and Pricoco 2024 differential adolescent recovery; no direct measurement of thymic output in ME/CFS by onset age).

Speculation: Hodgkin Lymphoma Bimodal Onset as Cross-Disease Model for EBV-Driven Early Peak

Hodgkin lymphoma (HL) has a well-characterized bimodal age distribution with peaks at ~20 and ~55 years. The early HL peak is EBV-associated (~70% of young-adult HL is EBV+ mixed cellularity subtype), while the late peak is EBV-negative (nodular sclerosis subtype). This partially parallels the ME/CFS pattern: early peak enriched for EBV/IM triggers, late peak with less infectious trigger predominance. However, the age peak mismatches are substantial (HL ~20, ~55 vs. ME/CFS ~16, ~37), and both diseases have different cellular mechanisms (malignant transformation vs. functional immune dysregulation). HL was selected as the strongest-fit bimodal-EBV comparator among known diseases with bimodal onset; many bimodal-onset conditions lack EBV associations, so this is a selected rather than sampled comparison. Both diseases may share impaired EBV immune control during young adulthood, but diverge in cell-type consequences. Family history of HL may be enriched in early-onset ME/CFS families (testable), and genetic variants predisposing to EBV+ HL (HLA-A01, HLA-B37) may overlap with early-onset ME/CFS susceptibility. HL’s established HLA associations contrast with ME/CFS’s absence of confirmed genetic signals, and similarity in age patterns alone is weak evidence for shared mechanism (Grotmol et al. 2011). Certainty: 0.20 (established HL bimodal epidemiology; age peaks differ substantially; no direct evidence of shared genetic susceptibility; HL was selected as best-fit comparator, not as a random sample of bimodal diseases).

Speculation: Type 1 Diabetes and ME/CFS: Shared Pubertal Thymic Vulnerability Window

Type 1 diabetes (T1D) incidence peaks at puberty (ages 10–14), coinciding with pubertal thymic involution and autoimmune susceptibility. T1D is a classical autoimmune disease where autoreactive T cells targeting pancreatic beta cells escape thymic deletion. The parallel with early-onset ME/CFS is mechanistic rather than epidemiological: the T1D peak (10–14) does not align with the ME/CFS early peak (16), and no study has found elevated T1D rates in ME/CFS families or vice versa. The absence of reported T1D-ME/CFS familial co-occurrence, despite both conditions being extensively studied, is a negative signal that weighs against shared genetic susceptibility. The comparison rests on temporal coincidence — puberty is a period of multiple physiological transitions, of which thymic involution is only one — and two diseases peaking at different ages may reflect different mechanisms within the pubertal window. This predicts that families with both T1D and ME/CFS, if they exist, will show enrichment for immune-regulatory gene variants affecting thymic selection (AIRE, FOXP3), and that early-onset ME/CFS patients may have higher rates of islet autoantibodies (GAD65, IA-2) than late-onset patients or controls. Certainty: 0.20 (T1D pubertal peak well-established; thymic involution-autoimmunity link mechanistically plausible; no data on T1D-ME/CFS familial co-occurrence or overlap of thymic-selection gene variants).

Speculation: XCI Escape as a Constant Female Susceptibility Mechanism Across Both Onset Peaks

The ~80% female ratio in ME/CFS is identical between the early and late onset peaks (S. J. McGrath et al. 2026), despite divergent hormonal trajectories (pubertal hormone surge in the early peak, perimenopausal transition in the late peak). If sex hormones were the primary driver of female susceptibility, the ratio should differ between peaks. This paradox points to a non-hormonal mechanism present from birth: X chromosome dosage effects on immune gene expression. The X chromosome carries the highest density of immune-related genes in the genome (TLR7, TLR8, CD40L, FOXP3, CXCR3). X chromosome inactivation (XCI) is incomplete for these immune genes—“escapee” genes produce higher expression in XX individuals regardless of hormonal state (XCI escape in immune genes is well-established; Carrel & Willard 2005). This creates a baseline immune activation difference that is constant across age and therefore constant across both peaks, answering the parity paradox. Importantly, the observed constant ratio does not exclude alternative or additive explanations—ascertainment bias, coincidental net-equivalence of different hormonal mechanisms at each peak, or sex-specific epigenetic programming independent of XCI escape could also produce (or contribute to) the parity. The XCI-escape mechanism is proposed as one coherent candidate, not the only one. This hypothesis predicts that: (a) XCI escape skewness will be higher in female ME/CFS patients than controls; (b) the magnitude will not differ between early- and late-onset; (c) males with Klinefelter syndrome (XXY) will have elevated ME/CFS risk approaching female rates; (d) females with Turner syndrome (XO) will have reduced risk. Certainty: 0.30 (XCI escape biology well-established; female predominance constant across peaks from survey data; no direct XCI-escapee measurement in ME/CFS; Klinefelter/ME/CFS comorbidity data do not exist).

6 Butyrate Across ME/CFS and Related Conditions

Butyrate deficiency has been observed across multiple chronic conditions with shared pathophysiological features. This cross-disease pattern suggests common mechanisms—possibly involving gut microbiome dysfunction, barrier breakdown, and immune dysregulation.

Speculation: X1: Butyrate-IBS-ME/CFS Continuum

Butyrate deficiency observed across IBS, ME/CFS, and fibromyalgia suggests a shared microbiome dysfunction spectrum. The severity of butyrate deficiency may help distinguish conditions and inform treatment targets.

Evidence:

IBS: Reduced butyrate producers (Faecalibacterium prausnitzii, Roseburia) in IBS-D and IBS-C, correlating with symptom severity ((C. Guo et al. 2023))
ME/CFS: Reduced butyrate producers in gut microbiome dysbiosis ((C. Guo et al. 2023))
Fibromyalgia: Preliminary evidence of reduced butyrate-producing bacteria, but less consistent than IBS/ME/CFS

Cross-condition analysis:

IBS and ME/CFS share GI symptoms, barrier dysfunction, and butyrate deficiency patterns
Fibromyalgia overlaps with ME/CFS in fatigue and central sensitization, but lacks consistent IBS-like GI dysfunction
Butyrate deficiency severity may distinguish IBS (moderate, GI-focused) from ME/CFS (severe, multi-system)

Mechanistic link: Butyrate deficiency reduces:

Tight junction maintenance (increased intestinal permeability)
Treg differentiation (immune dysregulation)
Vagal afferent activation (autonomic dysfunction)
HIF-1α stabilization (metabolic reprogramming)

Clinical implications:

ME/CFS patients with severe butyrate deficiency may benefit from targeted microbiome interventions (butyrate-producing probiotics, high-fiber diets)
Butyrate supplementation trials in ME/CFS may show differential response based on butyrate deficiency severity
Cross-disease butyrate deficiency panel could guide treatment selection across conditions

Cross-references: See Section [NO SECTION] for ME/CFS butyrate deficiency, Section [NO SECTION] for barrier mechanisms, and Section [NO SECTION] for vagal coupling.

Certainty: 0.60 (butyrate deficiency documented in IBS and ME/CFS; fibromyalgia evidence preliminary; mechanistic overlap across conditions is plausible but not proven)

Speculation: X2: Butyrate-Long COVID Fatigue Parallel

Long COVID post-viral fatigue shows butyrate deficiency patterns similar to ME/CFS, supporting cross-disease mechanistic hypotheses and potential for shared treatment protocols.

Evidence:

Long COVID: Gut microbiome dysbiosis with reduced butyrate producers (Kim et al. (Kim et al. 2026))
Long COVID fatigue: Similar to ME/CFS post-exertional fatigue (certainty: 0.50 from distinct PEM pathophysiology analysis Distinct PEM Pathophysiology in Long COVID vs ME/CFS)
Butyrate mechanisms: Similar in both conditions (HDAC inhibition, Treg upregulation, barrier maintenance)

Cross-condition comparison:

Fatigue patterns: ME/CFS PEM (2-day CPET deterioration) distinct from Long COVID (some patients report PEM but no CPET deterioration)
Butyrate deficiency: Documented in both, but severity may differ
Immune phenotype: ME/CFS more suppressed, Long COVID more activated (Eaton-Fitch 2024)
Viral persistence: SARS-CoV-2 RNA detectable in some Long COVID patients, not in ME/CFS

Mechanistic link: Butyrate deficiency in Long COVID may contribute to:

Persistent fatigue via vagal dysfunction (similar to ME/CFS)
Immune dysregulation through reduced Tregs and HDAC inhibition
Barrier dysfunction contributing to viral antigen persistence

Treatment implications:

Butyrate supplementation may benefit Long COVID patients with severe butyrate deficiency
Combined butyrate + antiviral protocols could address both microbiome dysfunction and viral persistence
Cross-condition butyrate-targeted trials could inform both ME/CFS and Long COVID treatment protocols

Cross-references: See Section [NO SECTION] for Long COVID vs ME/CFS comparison, Section [NO SECTION] for butyrate-vagal coupling, and Section [NO SECTION] for butyrate distribution.

Certainty: 0.55 (butyrate deficiency documented in Long COVID and ME/CFS; mechanistic parallels plausible; clinical trials needed to test shared treatment protocols)

Speculation: X3: Butyrate-IBD Barrier Comparison

Butyrate mechanisms (HDAC inhibition, tight junction maintenance) are central to IBD pathophysiology. Comparing barrier repair dynamics may reveal ME/CFS-specific dysfunction patterns.

IBD-butyrat mechanisms:

HDAC inhibition stabilizes tight junction proteins (claudin-1, occludin, ZO-1)
Butyrate reduces intestinal inflammation via Treg differentiation
Butyrate promotes colonocyte energy production and proliferation
Butyrate deficiency contributes to barrier breakdown in IBD

ME/CFS-butyrat mechanisms:

Butyrate deficiency documented in ME/CFS microbiome dysbiosis ((C. Guo et al. 2023))
Barrier dysfunction observed in ME/CFS (Section [NO SECTION])
Immune dysregulation via reduced Tregs and HDAC inhibition
Vagal coupling through serotonin-vagal pathway (Section [NO SECTION])

Cross-condition analysis:

IBD: Active inflammation with barrier breakdown, butyrate supplementation shows efficacy in clinical trials
ME/CFS: Chronic low-grade inflammation with barrier dysfunction, butyrate supplementation not yet tested in controlled trials
Severity: IBD butyrate deficiency correlates with active inflammation; ME/CFS butyrate deficiency may correlate with systemic symptoms

Distinguishing patterns:

IBD butyrate deficiency is primarily gut-localized with prominent GI symptoms
ME/CFS butyrate deficiency is systemic, with multi-organ dysfunction despite gut-localized deficiency
IBD responds well to butyrate supplementation; ME/CFS response not yet established

Mechanistic insight: Comparing barrier repair kinetics after butyrate challenge may reveal:

Whether ME/CFS barrier repair is fundamentally impaired (similar to IBD)
Whether ME/CFS requires higher butyrate doses due to impaired utilization
Whether ME/CFS-specific factors (immune dysregulation, autonomic dysfunction) limit butyrate efficacy

Cross-references: See Section [NO SECTION] for ME/CFS barrier dysfunction, IBD butyrate supplementation evidence, and Section [NO SECTION] for butyrate distribution.

Certainty: 0.50 (butyrate mechanisms well-established in IBD; ME/CFS butyrate deficiency documented; butyrate efficacy in ME/CFS not tested; barrier comparison is hypothesis-generating)

Speculation: X4: Butyrate-Autonomic Neuropathy Diabetic Parallels

Diabetic autonomic neuropathy shows vagal dysfunction; butyrate deficiency may contribute to similar dysautonomia in ME/CFS. Comparing mechanisms may reveal treatment targets.

Diabetic autonomic neuropathy (DAN):

Vagal dysfunction observed in 30–50% of diabetic patients
Reduced HRV, impaired baroreflex sensitivity
Mechanisms: hyperglycemia-induced oxidative stress, advanced glycation end-products (AGEs), inflammation
Evidence-based treatments: tight glycemic control, lifestyle modifications, possible butyrate supplementation

ME/CFS dysautonomia:

Vagal dysfunction observed in 30–40% of ME/CFS patients (POTS prevalence >30%)
Reduced HRV, impaired baroreflex sensitivity
Mechanisms: uncertain, but includes immune dysregulation, inflammation, neurotransmitter abnormalities (Wirth & Scheibenbogen 2025 (Wirth and Scheibenbogen 2025))
Current treatments: salt/fluids, compression, ivabradine, fludrocortisone

Cross-condition mechanisms:

Both conditions show reduced HRV and vagal efferent output
Both conditions involve low-grade inflammation contributing to vagal dysfunction
BUTYRATE DEFICIENCY MAY CONTRIBUTE TO DYSAUTONIA IN BOTH CONDITIONS via vagal afferent sensitization

Butyrate-DAN connection:

Butyrate enhances vagal afferent firing rate (ODE model in Section [NO SECTION])
Butyrate deficiency reduces vagal afferent input → impaired vagal efferent output
Butyrate improves mitochondrial function → supports neuronal energy metabolism

Butyrate-ME/CFS connection:

ME/CFS butyrate deficiency documented in gut microbiome dysbiosis ((C. Guo et al. 2023))
Butyrate deficiency may contribute to reduced vagal tone via vagal afferent sensitization
Combined butyrate + dysautonomia treatments may have synergistic effects

Mechanistic insight: Comparing butyrate dynamics in DAN vs ME/CFS may reveal:

Whether butyrate deficiency is common in dysautonomia across conditions
Whether butyrate supplementation can improve HRV in both DAN and ME/CFS
Which butyrate parameters (absorption, utilization, distribution) are impaired in dysautonomia

Cross-references: See Section [NO SECTION] for ODE model, Section [NO SECTION] for ME/CFS dysautonomia mechanisms, DAN pathophysiology and treatment evidence, and Section [NO SECTION] for butyrate distribution.

Certainty: 0.45 (butyrate-vagal coupling documented in animal studies; diabetic autonomic neuropathy mechanisms well-established; butyrate deficiency in ME/CFS not linked to dysautonomia; mechanistic comparison is hypothesis-generating)

Speculation: X5: Butyrate-Parkinson’s Constipation-Fatigue Bridge

Parkinson’s patients have constipation and fatigue; butyrate deficiency may contribute via similar gut-brain axis disruption. Exploratory cross-disease bridge.

Parkinson’s pathophysiology:

Constipation affects 70–90% of Parkinson’s patients, occurring years before motor symptoms
Gut microbiome dysbiosis: reduced butyrate producers (F. prausnitzii, R. hominis)
Fatigue common in Parkinson’s (fatigue prevalence 20–30%)
Gut-brain axis involvement: microbiota influence neuroinflammation and alpha-synuclein pathology

ME/CFS pathophysiology:

Constipation/GI dysfunction common (20–40% of ME/CFS patients)
Gut microbiome dysbiosis: reduced butyrate producers ((C. Guo et al. 2023))
Fatigue is cardinal feature of ME/CFS
Gut-brain axis involvement: serotonin-vagal pathways, inflammation

Cross-condition comparison:

Parkinson’s: Constipation precedes fatigue by years; fatigue is secondary to neurological progression
ME/CFS: Fatigue is primary; GI dysfunction and fatigue co-occur; unclear whether GI dysfunction precedes fatigue

Butyrate-ME/CFS link:

Butyrate deficiency reduces gut motility (butyrate stimulates intestinal peristalsis)
Butyrate deficiency reduces Treg differentiation → immune dysregulation
Butyrate deficiency contributes to fatigue via vagal dysfunction (similar to X4)

Butyrate-Parkinson’s link:

Butyrate deficiency may contribute to constipation via reduced gut motility
Butyrate deficiency may contribute to fatigue via gut-brain axis disruption and immune dysregulation
Cross-condition butyrate-targeted treatments could benefit both conditions

Mechanistic insight: Comparing butyrate dynamics in Parkinson’s vs ME/CFS may reveal:

Whether butyrate deficiency is common in Parkinson’s constipation
Whether butyrate deficiency contributes to fatigue in both conditions
Which butyrate parameters are impaired (absorption, utilization, distribution)

Cross-references: See Section [NO SECTION] for butyrate deficiency in ME/CFS, Parkinson’s gut-brain axis mechanisms, gut motility regulation, and Section [NO SECTION] for butyrate distribution.

Certainty: 0.40 (butyrate deficiency in Parkinson’s constipation documented; fatigue mechanisms in Parkinson’s not butyrate-linked; ME/CFS butyrate deficiency not linked to Parkinson’s; cross-condition bridge is exploratory)

Speculation: X6: Butyrate-MS Fatigue Overlap

Multiple sclerosis fatigue shows gut dysbiosis patterns; butyrate deficiency may contribute to fatigue via immune and metabolic mechanisms. Cross-disease exploration.

MS fatigue:

Fatigue prevalence 70–80% in MS (higher than ME/CFS fatigue prevalence of ~70% overall)
Fatigue worsens with disease progression, physical activity, heat
Gut microbiome dysbiosis in MS: reduced butyrate producers (F. prausnitzii, R. hominis)
Butyrate supplementation improves MS-related fatigue in pilot studies

ME/CFS fatigue:

Fatigue is cardinal feature, with PEM (24–72h delayed)
Gut microbiome dysbiosis: reduced butyrate producers ((C. Guo et al. 2023))
Fatigue mechanisms: uncertain, but includes immune dysregulation, metabolic dysfunction, CNS effort computation changes

Cross-condition comparison:

Fatigue severity: MS fatigue 70–80% prevalence; ME/CFS fatigue 70% prevalence (broad estimate)
Fatigue triggers: MS fatigue triggered by disease progression, activity, heat; ME/CFS fatigue triggered by exertion (PEM)
Gut microbiome: Both conditions show reduced butyrate producers; butyrate supplementation shows some benefit in MS fatigue

Butyrate-MS link:

Butyrate reduces neuroinflammation via HDAC inhibition and Treg differentiation
Butyrate improves mitochondrial function in immune cells
Butyrate deficiency contributes to fatigue via immune dysregulation and metabolic dysfunction

Butyrate-ME/CFS link:

ME/CFS butyrate deficiency documented in gut microbiome dysbiosis ((C. Guo et al. 2023))
Butyrate deficiency contributes to immune dysregulation via reduced Tregs and HDAC inhibition
Butyrate deficiency contributes to fatigue via vagal dysfunction and metabolic dysfunction

Mechanistic insight: Comparing butyrate dynamics in MS vs ME/CFS may reveal:

Whether butyrate deficiency is common in MS fatigue
Whether butyrate supplementation can improve ME/CFS fatigue
Which butyrate parameters are impaired (absorption, utilization, distribution)

Cross-references: See Section [NO SECTION] for butyrate deficiency in ME/CFS, MS gut microbiome dysbiosis, butyrate supplementation evidence in MS, and Section [NO SECTION] for immune modulation mechanisms.

Certainty: 0.45 (butyrate deficiency in MS gut dysbiosis documented; butyrate supplementation shows benefit in MS fatigue; ME/CFS butyrate deficiency documented; butyrate efficacy in ME/CFS not tested; cross-condition exploration is hypothesis-generating)

Speculation: Early-Onset Severity Reflects Developmental Neurovulnerability, Not Disease Duration

Early-onset ME/CFS is more severe (OR 2.15) independent of illness duration . Brain myelination and synaptic pruning continue into the mid-20s. Immune-mediated neuroinflammation during this developmental window may produce permanent structural changes (reduced prefrontal white matter integrity, altered default mode network connectivity) that are irreversible even if the immune trigger resolves. Late onset, hitting a mature brain, produces functional but not structural disruption. This predicts: (a) early-onset patients will show reduced fractional anisotropy on DTI in prefrontal regions compared to late-onset patients matched for duration and severity; (b) the severity difference will be mediated by white matter integrity metrics; (c) early-onset patients will show altered functional connectivity in networks still maturing at age 16 (default mode, salience network). Certainty: 0.30 (developmental neurobiology well-established; no ME/CFS neuroimaging data stratified by onset age; mediation analysis not performed).

Speculation: The Late Peak as Cumulative Mitochondrial Reserve Exhaustion

Age ~37 may represent the point where cumulative metabolic stress intersects with beginning age-related mitochondrial quality-control decline. Mitochondrial DNA deletions begin accumulating measurably in the 30s; mitophagy efficiency declines; NAD⁺ levels drop. Simultaneously, this is peak career stress (ages 35–44), peak childbearing years, and for some women, earliest perimenopausal hormonal shifts. Unlike early onset (a single catastrophic immune event), late onset may represent a “mitochondrial debt” model: years of subclinical metabolic inefficiency accumulate until a threshold event tips the system past a critical point. This predicts: (a) late-onset patients will show higher mitochondrial DNA deletion burden than age-matched controls but lower than early-onset patients; (b) metabolomic signatures of cumulative oxidative stress will correlate with pre-illness stress burden; (c) NAD⁺ precursor supplementation will preferentially benefit late-onset patients. Certainty: 0.25 (consistent with lower severity and lower infectious trigger rate in late onset; no direct evidence for mitochondrial reserve threshold at age ~37).

7 The Dysautonomia Spectrum

ME/CFS overlaps heavily with autonomic dysfunction syndromes:

Postural Orthostatic Tachycardia Syndrome (POTS).

Many ME/CFS patients meet POTS criteria
Both involve small fiber neuropathy in subsets
Both show autoantibodies to adrenergic receptors
Speculative link: POTS may represent ME/CFS with predominant autonomic lock; or both may be manifestations of autoimmune autonomic ganglionopathy spectrum

Limitation: POTS Diagnostic Validity: Threshold, Specificity, and Reproducibility Concerns

The assumed ME/CFS–POTS association deserves critical scrutiny . The defining POTS criterion—a $\geq$ 30 bpm heart rate increase on standing—has poor diagnostic specificity: 10–15% of healthy controls meet this threshold in population studies. Among ME/CFS patients with orthostatic intolerance symptoms, only 21% meet formal POTS criteria, indicating that the majority of orthostatic symptoms in ME/CFS are not captured by the POTS diagnosis. Symptom severity correlates poorly with heart rate increase magnitude, explaining only approximately 2% of variance—patients with modest tachycardia can be severely symptomatic, and vice versa. Perhaps most concerning, POTS test results show poor day-to-day reproducibility, with patients meeting criteria on one day but not the next. These limitations suggest that POTS as currently defined may be a poorly specified diagnostic category rather than a coherent pathophysiological entity, and that its apparent comorbidity with ME/CFS may partly reflect diagnostic threshold artefacts.

Inappropriate Sinus Tachycardia.

Elevated resting heart rate without clear cause
Often comorbid with POTS and ME/CFS
Speculative link: May reflect autoantibodies to cardiac $\beta$-receptors or sinoatrial node ion channels

Neurocardiogenic Syncope.

Vasovagal responses at inappropriate times
Common in ME/CFS population
Speculative link: Reflects vagal afferent sensitization combined with impaired compensatory responses

Open Question: Autonomic Autoimmunity Unifying Hypothesis

What if ME/CFS, POTS, and related dysautonomias all represent different manifestations of autoimmune attack on the autonomic nervous system? The specific antibody targets (muscarinic, adrenergic, ganglionic nicotinic, ion channels) might determine whether someone presents primarily as POTS, ME/CFS, or mixed. This “autoimmune autonomic spectrum” could be as common as rheumatoid arthritis but remains unrecognized because we don’t routinely test for the antibodies.

8 The Mast Cell Connection

Mast cell activation appears connected to ME/CFS:

Mast Cell Activation Syndrome (MCAS).

MCAS is the diagnosable subset of the broader “mast cell / histaminergic dysregulation” domain (Domain 6 in the multi-domain framework; Section Multi-Domain Co-Occurrence Model); histamine intolerance (HIT) due to DAO deficiency is a distinct, overlapping but mechanistically separate condition.

High comorbidity with ME/CFS
Explains chemical sensitivities, food reactions, flushing
Mast cells release histamine, prostaglandins, cytokines
Speculative link: MCAS may be both cause and effect—initial mast cell activation contributes to the trigger; ongoing activation maintains inflammation

Histamine Intolerance.

Many ME/CFS patients report histamine-related symptoms
May reflect DAO enzyme dysfunction or mast cell instability
Speculative link: Histamine is a circadian regulator; chronic histamine excess might contribute to circadian desynchronization

Mastocytosis.

Clonal mast cell disorders
More severe than MCAS but overlapping symptoms
Speculative link: Both conditions might involve mast cell progenitor dysregulation; ME/CFS could involve functional mastocytosis without clonal proliferation

Speculation: Na⁺/K⁺-ATPase Failure as Common Mechanism for PEM and MCAS Flares

Mast cell degranulation requires intracellular calcium elevation. The NCX reversal mechanism described in Chapter [NO CHAPTER] — Na⁺/K⁺-ATPase failure → intracellular sodium rise → NCX reverse mode → calcium influx—operates in all cell types, not only skeletal muscle. If $\beta_2$-adrenergic receptor dysfunction impairs pump function globally, then mast cells would experience the same calcium flooding that causes muscle necrosis during PEM. This would mean the same upstream mechanism simultaneously triggers PEM (via muscle calcium toxicity) and MCAS flares (via mast cell calcium-induced degranulation), explaining their frequent co-occurrence without requiring mast cells to be independently dysfunctional.

The prediction is clean: MDC002 (Na⁺/K⁺-ATPase stimulator, Chapter [NO CHAPTER]) should reduce both PEM severity and MCAS flare frequency if the shared calcium source is confirmed. Conversely, mast cell stabilizers (cromolyn, ketotifen) that partially regulate calcium channels may provide modest PEM benefit as a secondary effect.

Certainty: 0.30—mechanistically coherent but entirely untested; requires demonstration of elevated intracellular sodium and calcium in mast cells (not just muscle) of ME/CFS patients.

Open Question: Mast Cells as Central Orchestrators?

What if mast cells are the “hub” connecting multiple ME/CFS mechanisms? Mast cells:

Are activated by stress, infection, and multiple triggers
Release mediators affecting every organ system
Can maintain chronic inflammation
Are present at blood-brain barrier and affect CNS function
Are regulated by autonomic nervous system (which is dysfunctional)

The mast cell might be the cell type where multiple locks converge.

Open Question: Mast Cells as Neuro-Immune Signal Amplifiers?

The intimate physical proximity of mast cells to peripheral nerve endings ($<$ 20 nm in many tissues) may enable bidirectional signaling beyond currently recognized neuroimmune crosstalk. Could mast cells function as biological signal repeaters or gain modulators in the nervous system?

Proposed mechanism:

Mast cells detect neurotransmitter spillover and neuropeptide signals from nearby nerves
Release precisely timed micro-bursts of neurotransmitters (serotonin, histamine) and ions (Ca²⁺, K⁺)
Bridge gaps in neural signaling across regions of small fiber neuropathy
Modulate sensory sensitivity by adjusting nerve receptor thresholds via protease release (e.g., tryptase activation of PAR2)

This would explain:

Allodynia and hyperalgesia: Mast cells with lowered activation thresholds act as signal amplifiers, magnifying innocuous stimuli into pain signals
SFN-MCAS overlap: Small fiber neuropathy (non-length-dependent pattern documented in 34% of ME/CFS patients ) combined with mast cell hyperreactivity creates paradoxical hypersensitivity despite nerve damage
Variability of sensory symptoms: Mast cell activation state (influenced by histamine load, stress, inflammation) dynamically modulates sensory gain day-to-day
“Phantom” sensations: Mast cells broadcasting signals to multiple nerve fibers create diffuse sensory fields beyond the original stimulus location

Testable predictions:

Mast cell stabilizers should reduce allodynia severity
Quantitative sensory testing abnormalities should correlate with mast cell activation markers (tryptase, histamine)
Time-course studies: sensory thresholds should fluctuate with mast cell mediator levels
Electrophysiology: mast cell degranulation near nerve fibers should alter nerve conduction patterns

Supporting evidence: Mast cells form CADM1-mediated adhesion structures with sensory neurons that amplify degranulation ($\\times$ 2-fold) and IL-6 secretion ($\\times$ 3-fold) . Approximately 80% of mast cell disorder patients demonstrate small fiber neuropathy on objective testing , establishing the clinical overlap. Mast cell-nerve bidirectional signaling has been documented, though the specific role of tryptase-PAR2 interactions in ME/CFS sensory symptoms remains to be established.

Current evidence gaps: No direct studies demonstrate mast cells amplifying neural signals in real-time. However, the physical infrastructure exists (proximity, neurotransmitter release capability, bidirectional signaling via CADM1 ), and the clinical phenotype (SFN + MCAS + allodynia) suggests functional coupling.

Open Question: Mast Cells as Environmental Memory Keepers?

Mast cells are extraordinarily long-lived immune cells, persisting for years in the same tissue location at barrier surfaces (gut, skin, airways). Unlike B-cells that remember specific pathogens, could mast cells maintain an epigenetic archive of chronic environmental exposures?

Proposed mechanism:

Mast cells continuously sample the local chemical environment over years
Chronic exposures (pollutants, dietary patterns, stress hormones, microbiome metabolites) induce epigenetic modifications
These modifications adjust degranulation thresholds and mediator release patterns
Epigenetically modified mast cells maintain altered activation thresholds for their lifespan, creating persistent sensitization

This would explain:

Geographic remission: Why some chronic illness patients improve upon moving to different climates or environments—new location lacks the accumulated “environmental signature” archived in mast cells
Chemical sensitivity acquisition: Gradual sensitization to previously tolerated exposures as mast cells archive repeated low-level irritation
“Total load” phenomenon: Why symptoms worsen with cumulative exposure to multiple triggers—mast cells integrate exposures over time rather than responding to isolated events
Delayed recovery after trigger removal: Environmental changes require years to benefit because mast cells live for years and carry historical “memory”

Testable predictions:

Mast cells from patients in different environments should show distinct epigenetic signatures
Mast cell epigenetic profiles should correlate with lifetime environmental exposure history
Geographic relocation should gradually shift mast cell epigenetic patterns over 1–3 years (matching mast cell lifespan)
Tissue-resident mast cells should show different epigenetic profiles than circulating mast cell progenitors

Supporting evidence: Mast cells are exceptionally long-lived tissue residents (estimated months to years based on tissue turnover studies), maintaining themselves independently from bone marrow and accumulating tissue-specific programming. Epigenetic mechanisms (DNA methylation, histone acetylation) are known to control immune cell activation thresholds in general, and chronic immune activation can create lasting epigenetic signatures in other cell types. Environmental exposures (dietary factors, pollution) can alter immune cell function through epigenetic modifications. MCAS patients show persistent alterations in activation thresholds that may reflect long-term cellular reprogramming.

Current evidence gaps: While immune cell epigenetic memory is established and mast cell activation thresholds are known to be epigenetically controlled, no studies have directly examined whether mast cells archive general environmental exposures beyond standard antigen-specific immunity. Mast cell epigenetics in ME/CFS remain entirely unstudied.

Open Question: Circadian Mast Cell Regulation and Potential Temporal Learning

Mast cells possess intrinsic circadian clocks that regulate degranulation (established). But could they also develop learned temporal associations beyond the 24-hour circadian rhythm—anticipating specific triggers at arbitrary times based on repeated exposure patterns?

Proposed mechanism (speculative):

Established: Mast cells have circadian clocks that regulate Fc$\varepsilon$RI expression and degranulation sensitivity based on time-of-day
Speculative: With repeated exposure to triggers at consistent times (e.g., breakfast food at 8 AM daily), mast cells might develop learned temporal associations independent of circadian phase
Granules could undergo partial “pre-thaw” 15–30 minutes before expected trigger time
If trigger arrives on schedule, full degranulation occurs rapidly with amplified response
If trigger is absent, partial priming gradually reverses

This would explain:

Time-of-day variability: Why patients tolerate certain foods/medications better at different times—mast cells are or aren’t pre-primed
Nocturnal symptom flares: If evening routines consistently trigger mild mast cell activation, circadian priming might amplify nighttime symptoms
Elimination diet inconsistency: Removing a food might fail if mast cells remain circadian-primed for weeks, causing reactions to “safe” foods eaten at the same time
“Spontaneous” reactions: Circadian mast cell priming without actual trigger exposure could cause symptoms at predictable times
Vacation effect: Disrupted routines break circadian priming patterns, temporarily reducing reactivity

Testable predictions:

Mast cell mediators (histamine, tryptase) should show circadian oscillations correlating with habitual trigger exposure times
Time-series sampling: baseline mediator levels should rise 15–30 min before scheduled triggers
Experimental circadian disruption (shift work simulation) should temporarily reduce food/medication reactions
Re-timing trigger exposure (breakfast foods eaten at dinner) should shift circadian mediator patterns within 1–2 weeks

Supporting evidence: Mast cells possess intrinsic molecular clocks that temporally regulate degranulation through circadian oscillation of Fc$\varepsilon$RI receptor expression and downstream signaling components . The mast cell clock is entrained by humoral factors (adrenal hormones) and can be modulated by environmental stressors . Circadian disruption eliminates temporal gating of mast cell activation, resulting in sustained hyperreactivity throughout the day . The immune system exhibits anticipatory responses to predictable environmental threats as a fundamental circadian function.

Current evidence gaps: While circadian immune regulation is established (cortisol awakening response, circadian cytokine patterns) and mast cells demonstrably have circadian clocks , this hypothesis proposes a distinct phenomenon: learned temporal associations beyond the 24-hour circadian rhythm. Nakamura et al. demonstrated that mast cells respond to endogenous circadian cues (hormones, light-dark cycles), not learned associations with specific environmental triggers at arbitrary times. No studies have examined whether mast cells can develop anticipatory priming to non-circadian temporal patterns (e.g., “breakfast at 8 AM” vs. “breakfast at 10 AM”). This would require a form of Pavlovian temporal conditioning not yet demonstrated in immune cells. The “predictive brain” framework is well-developed in neuroscience but hasn’t been applied to mast cell biology.

9 Food-Triggered Mast Cell Activation: The Silent Pathway

A subset of ME/CFS patients report food-related symptom exacerbation—fatigue flares, brain fog, tachycardia, gastrointestinal distress—yet standard allergy testing (skin prick, specific IgE) returns negative. Food intolerances are present in 68–70% of ME/CFS patients across immunological subgroups , and MCAS prevalence in ME/CFS is estimated at 16.7–25.3% depending on diagnostic stringency . The question is whether these food reactions reflect genuine mast cell activation through pathways invisible to conventional allergy diagnostics.

9.1 Non-IgE Pathways of Food-Triggered Mast Cell Degranulation

At least four mechanistically distinct pathways can trigger mast cell degranulation in response to food without involving allergen-specific IgE. These pathways operate in parallel; an individual patient may be affected by one or several simultaneously.

MRGPRX2-mediated activation. MRGPRX2 (Mas-Related G Protein-Coupled Receptor Member X2) is expressed exclusively on connective tissue mast cells and mediates IgE-independent degranulation via G$\alpha$i/G$\alpha$q signaling, intracellular calcium mobilization, and ERK1/2 activation . Its ligands include substance P, hemokinin-1, host-defense antimicrobial peptides (LL-37, $\beta$-defensins), and bacterial quorum-sensing peptides from the gut microbiota . No food-derived peptide has been directly demonstrated to activate MRGPRX2 in published human studies, but the pathway is plausibly engaged indirectly: food ingestion stimulates gut sensory neurons to release substance P, particularly in the context of pre-existing intestinal inflammation or barrier dysfunction. MRGPRX2 has naturally occurring gain-of-function variants that lower degranulation thresholds for substance P, suggesting variable susceptibility across individuals .

Complement-mediated activation (C3a/C5a). The complement fragments C3a and C5a (“anaphylatoxins”) bind cognate receptors on mast cells and trigger degranulation via PLC$\beta$-mediated calcium mobilization, PKC, and PI3K . Food-relevant complement activation occurs through two routes: dietary lectins and plant-derived polysaccharides activate the lectin pathway via mannose-binding lectin (MBL), generating C3a/C5a locally in gut mucosa; and IgG food antibodies (distinct from IgE) fix complement via the classical pathway. Neither route is detected by standard food allergy testing. Mast cells are themselves sources of complement proteins, forming a positive feedback amplification loop .

Lectin–IgE glycan cross-linking. Food lectins—carbohydrate-binding proteins abundant in legumes, nightshades, and grains—can activate mast cells by binding to the glycan chains of cell-surface IgE. IgE contains 10–12% carbohydrate by weight; lectins cross-link these glycan cores on adjacent IgE molecules, triggering Fc$\varepsilon$RI clustering and degranulation even in the complete absence of allergen-specific IgE. This mechanism is most pronounced in atopic individuals with elevated total IgE and predicts that legume-heavy or nightshade-heavy diets could provoke symptoms in susceptible individuals without any detectable food-specific sensitization.

Food additives and oxidative pathways. Sodium sulfite (present in wine, dried fruit, processed foods) triggers mast cell activation through NADPH oxidase-mediated intracellular oxidative stress, NLRP3 inflammasome activation, and pyroptosis-associated degranulation—a pathway independent of both IgE and extracellular calcium . Natural colorants (carmine, annatto) have stronger evidence for genuine hypersensitivity than synthetic dyes, while the “Chinese Restaurant Syndrome” attributed to MSG has not been confirmed in placebo-controlled trials . The clinical reality is that perceived food additive sensitivities are common in MCAS patients, but the fraction attributable to direct mast cell activation versus other mechanisms (nocebo, altered gut motility, dysbiosis) remains uncertain.

9.2 Amplifying Factors in ME/CFS

Several features of ME/CFS pathophysiology may lower the threshold for food-triggered mast cell activation, creating a vicious cycle not present in healthy individuals.

Intestinal barrier dysfunction is well-documented in ME/CFS. Rohrhofer et al. found elevated lipopolysaccharide-binding protein (LBP) in ME/CFS patients without immunodeficiency ($n = 39$), indicating bacterial endotoxin translocation across a leaky gut. This barrier compromise allows dietary components—food antigens, bacterial products, lectins—to access submucosal mast cells at higher concentrations than in healthy gut. LPS itself is a TLR4 agonist that primes mast cells for lower-threshold degranulation on subsequent challenge.

Concurrently, the ME/CFS gut microbiome shows reduced diversity and altered short-chain fatty acid (SCFA) production. Since butyrate and propionate actively inhibit mast cell degranulation via JNK suppression and HDAC inhibition , reduced SCFA production removes a natural brake on intestinal mast cell reactivity. Patients who self-restrict fiber intake due to gastrointestinal sensitivity may paradoxically worsen their mast cell activation burden.

9.3 Mast Cell Mediators and ME/CFS Symptom Mapping

The mediators released during mast cell degranulation map directly onto core ME/CFS symptoms. Cytokines (IL-1$\beta$, IL-6, TNF-$\alpha$) are canonical inducers of “sickness behavior”—centrally mediated fatigue, malaise, and somnolence. Histamine acts at H3 receptors in the CNS to suppress acetylcholine, serotonin, and norepinephrine release, disrupting arousal, attention, and memory—the neurochemical substrate of “brain fog.” Prostaglandins and histamine cause vasodilation and tachycardia, driving orthostatic intolerance. Kohno et al. found that 42% of POTS patients ($n = 69$) exhibited elevated mast cell mediators, with histamine/methylhistamine elevated in 52% and prostaglandin D₂ in 36%—but tryptase in only 9%, demonstrating that tryptase is a poor indicator of mast-cell-mediated autonomic dysfunction. Rohrhofer et al. confirmed this association in ME/CFS: patients with mast cell activation had significantly higher orthostatic intolerance rates (89% vs. 72%; $p < 0.0001$) and a trend toward higher POTS prevalence (50.8% vs. 36.5%).

At the CNS level, mast cells in the median eminence and hypothalamus are positioned adjacent to CRH neurons. Mast cell-released tryptase, neurotensin, and CXCL8 activate microglia, propagating neuroinflammation to prefrontal and hippocampal circuits . Tsilioni et al. provided mechanistic support: exosome-associated mitochondrial DNA is elevated in ME/CFS patient serum after exercise and stimulates cultured human microglia to release IL-1$\beta$—connecting exertion-triggered immune activation to central neuroinflammation. Weinstock et al. documented resolution of neuropsychiatric symptoms including cognitive dysfunction in all eight MCAS patients treated with mast cell-directed therapy, though the case series design limits generalizability.

Speculation: Food-Triggered Non-IgE Mast Cell Activation as Contributor to ME/CFS Symptom Burden

In the subset of ME/CFS patients with comorbid MCAS (estimated 17–25% ), food ingestion triggers mast cell degranulation through non-IgE pathways (MRGPRX2 via gut neuropeptides, complement C3a/C5a via lectins, lectin–IgE glycan cross-linking) that are invisible to standard allergy testing. The resulting mediator release—histamine, prostaglandins, cytokines—contributes to post-meal fatigue flares, brain fog, tachycardia, and gastrointestinal distress. This mechanism is amplified in ME/CFS by intestinal barrier dysfunction and reduced SCFA-mediated mast cell inhibition , creating a self-reinforcing cycle where food-triggered activation worsens gut barrier integrity, further lowering activation thresholds (certainty: 0.40; multiple plausible mechanisms with indirect clinical support, but no prospective mediator study during food challenges in ME/CFS patients).

Testable predictions:

Urinary mast cell mediators (NMH, LTE4, PGD₂ metabolite) should rise after controlled food challenges in ME/CFS patients with clinical food reactivity, even when specific IgE testing is negative
MRGPRX2 expression on intestinal mast cells should be elevated in food-reactive ME/CFS patients compared to non-reactive ME/CFS patients and healthy controls
Elimination diet responders should show mediator normalization on paired urinary testing during elimination versus rechallenge phases
Gut permeability markers (LBP, zonulin) should correlate with food-triggered mediator elevations

Limitations: No prospective study has measured mast cell mediators before and after controlled food challenges in ME/CFS patients. The prevalence figures for MCAS in ME/CFS (Rohrhofer et al. 2025) are based primarily on symptom pattern and treatment response rather than laboratory mediator confirmation—mediator testing was performed in only 3.1% of the cohort. The non-IgE activation pathways are established in vitro but their contribution to clinical food reactions in MCAS remains unquantified.

Speculation: Post-Exertional Malaise Partially Mediated by Food–Exercise Mast Cell Co-Triggering

Exercise releases substance P and neuropeptides from sensory fibers, activating mast cells via MRGPRX2 . In ME/CFS, the threshold for this cascade may be lowered by prior food-triggered mast cell priming—analogous to food-dependent exercise-induced anaphylaxis (FDEIA), where food antigens alone do not trigger reactions but the combination of food ingestion and exercise does, because exercise increases intestinal permeability and transglutaminase-mediated antigen cross-linking. A similar co-trigger mechanism might operate at subthreshold levels: meals consumed close to physical activity could lower the exertion threshold for PEM in patients with concurrent MCAS (certainty: 0.25; coherent mechanistic hypothesis built from adjacent evidence in FDEIA, MRGPRX2 biology, and ME/CFS exercise physiology, but no direct experimental evidence).

Testable predictions:

PEM severity should correlate with proximity of food intake to exertion in MCAS+ME/CFS patients
Fasted exercise should produce less severe PEM than fed exercise in this subgroup
Pre-treatment with mast cell stabilizers before food+exercise challenges should attenuate PEM
Urinary mediators should show greater post-exercise elevation when exercise follows a meal versus fasting

Limitations: No published study has measured mast cell mediators before and after exertion challenges in ME/CFS. The food-dependent exercise-induced anaphylaxis model involves IgE-mediated reactions at suprathreshold levels, while this hypothesis proposes subthreshold non-IgE activation—an untested extrapolation.

Warning: MCAS Diagnostic Precision Required

The international consensus criteria for MCAS require all three of: (1) recurrent symptoms across $\geq$ 2 organ systems, (2) serum tryptase increase of $>$ 20% above baseline plus 2 ng/mL during an acute episode, and (3) response to mast cell mediator-targeting therapy . Under strict criteria, idiopathic MCAS prevalence is approximately 4.4% ; broader clinical definitions that omit laboratory confirmation yield much higher prevalence estimates but at the cost of diagnostic specificity. Fewer than 5% of patients suspected of idiopathic MCAS meet strict consensus criteria . Not all ME/CFS symptom burden should be attributed to mast cells; this mechanism applies to a subset. Hereditary alpha-tryptasemia (extra copies of TPSAB1, affecting 5–6% of the population) causes chronically elevated baseline tryptase and must be excluded before interpreting tryptase-based MCAS criteria.

Open Question: The SCFA–Mast Cell–Dysbiosis Triangle in ME/CFS

If ME/CFS gut dysbiosis reduces SCFA production, and SCFAs are natural inhibitors of mast cell degranulation , then gut microbiome composition may directly modulate mast cell activation thresholds. Could targeted prebiotic or SCFA supplementation reduce mast cell reactivity to food in ME/CFS patients with comorbid MCAS? This would represent a dietary intervention acting not on the trigger (food) but on the brake (SCFA-mediated inhibition). However, the SCFA–mast cell connection has been demonstrated only in vitro and animal models; clinical translation to MCAS or ME/CFS patients is entirely untested.

10 The Ehlers-Danlos Connection

The high comorbidity of ME/CFS with hypermobile Ehlers-Danlos Syndrome (hEDS) is striking and demands mechanistic explanation. Registry data from 815 ME/CFS patients found 15.5% were joint hypermobility positive, with this subgroup showing significantly worse quality of life, more autonomic symptoms, and higher rates of both POTS (33% vs. 20%) and formal EDS diagnosis (29% vs. 3%) . This represents a distinct clinical phenotype within ME/CFS.

10.1 Epidemiological Evidence

Comorbidity Rates. Joint hypermobility prevalence varies across conditions:

General population: 10–20%
ME/CFS: 15.5–57% (varies by study and criteria)
POTS: up to 57%
Long COVID: approximately 30%
Fibromyalgia: approximately 27%

The enrichment of hypermobility in ME/CFS and related conditions is statistically significant, though its biological significance is debated (see limitation below).

Limitation: hEDS Diagnostic Validity: No Identified Connective Tissue Defect

The assumed ME/CFS–hEDS association must be interpreted in light of fundamental problems with the hEDS diagnosis itself . Unlike other Ehlers-Danlos subtypes (classical, vascular, kyphoscoliotic), which have identified genetic and structural defects, hEDS has no identified connective tissue defect despite extensive investigation. The diagnostic features used to identify hEDS—joint hypermobility, skin hyperextensibility, and associated symptoms—are extremely common in the general population (10–20% prevalence for hypermobility alone), raising questions about whether hEDS represents a distinct pathological entity or a statistical tail of normal variation.

Further challenging the ME/CFS–hEDS–POTS triad, the autonomic profile of hEDS patients resembles fibromyalgia more closely than it resembles other EDS types . If hEDS shared the same structural connective tissue mechanism as classical or vascular EDS, one would expect similar autonomic phenotypes across EDS types—but this is not observed. The hEDS-specific autonomic features may reflect central sensitization or functional nervous system changes rather than peripheral structural defects.

These observations do not negate the clinical reality that patients with hypermobility and ME/CFS form a distinct, more severely affected phenotype (). However, they suggest that the mechanistic pathways proposed in this section (Pathways 1–5) should be treated as more speculative than the clinical epidemiology alone would indicate, because the upstream assumption—that hEDS reflects a specific connective tissue defect—remains unvalidated.

Clinical Phenotype Differences. ME/CFS patients with joint hypermobility (JH+) compared to those without (JH$-$) show :

Worse physical functioning and pain scores
Higher burden of autonomic, neurocognitive, and musculoskeletal symptoms
More frequent headaches and gastrointestinal symptoms
Family history of EDS more common

This suggests JH+ ME/CFS may represent a mechanistically distinct subtype.

10.2 Mechanistic Pathways: From Connective Tissue to Systemic Dysfunction

The question is not merely whether EDS and ME/CFS are associated, but why. Several mechanistic pathways have varying levels of evidence.

Pathway 1: Vascular Laxity → Autonomic Dysfunction (HIGH EVIDENCE).

This is the best-supported mechanistic link. Defective connective tissue directly affects blood vessel structure and function:

Increased arterial compliance: EDS patients show significantly lower central pulse wave velocity (4.73 m/s vs. controls), indicating excessive arterial elasticity . This impairs baroreceptor signaling—stretch receptors in vessel walls cannot accurately detect blood pressure changes when the walls are too compliant.
Excessive venous pooling: Abnormal connective tissue in veins causes excessive distension under normal hydrostatic pressures cardiovascular. Blood pools in lower extremities upon standing, reducing venous return and cardiac preload.
Compensatory tachycardia: The heart races to maintain cardiac output despite reduced preload, producing POTS. Up to 70% of hEDS patients report dysautonomia symptoms, and up to 40% meet formal POTS criteria .
Cerebral hypoperfusion: Inadequate blood pressure regulation leads to reduced cerebral blood flow, particularly upon standing, causing cognitive symptoms, lightheadedness, and fatigue.

This pathway explains why POTS is so prevalent in both hEDS and ME/CFS—the autonomic dysfunction in hEDS is a direct, structural consequence of connective tissue abnormality rather than a secondary phenomenon.

Pathway 2: Craniocervical Instability → Brainstem Dysfunction (MODERATE EVIDENCE).

Ligamentous laxity at the craniocervical junction (C0–C2) can cause structural instability with neurological consequences:

Brainstem compression: The brainstem controls autonomic functions. Instability at the skull-spine junction can cause intermittent compression or stretching of brainstem structures .
CSF flow obstruction: Craniocervical instability can obstruct cerebrospinal fluid flow at the craniocervical junction, potentially causing increased intracranial pressure and impairing glymphatic waste clearance.
Vertebral artery effects: Cervical instability may affect vertebral artery flow, contributing to posterior circulation insufficiency.

A systematic review of 16 studies (695 EDS patients) found significant heterogeneity in diagnostic criteria for craniocervical instability, with no standardized thresholds cci. Dynamic imaging (upright MRI, flexion-extension views) provides superior diagnostic information compared to static supine imaging. Some ME/CFS patients with craniocervical instability report improvement after surgical stabilization, though controlled outcome data remain limited.

Warning: Craniocervical Instability: Evidence Limitations

While biologically plausible, the CCI-ME/CFS connection remains largely anecdotal. No controlled studies have established:

True prevalence of CCI in ME/CFS populations
Whether CCI causes ME/CFS symptoms vs. co-occurring conditions
Long-term surgical outcomes in ME/CFS patients with CCI

Screening for CCI may be appropriate in ME/CFS patients with hypermobility and progressive neurological symptoms, but surgery should be approached cautiously given the limited evidence base.

Pathway 3: Extracellular Matrix → Mast Cell Dysregulation (LOW EVIDENCE).

The “EDS-MCAS-POTS triad” is frequently discussed clinically, but a critical review found that “an evidence-based, common pathophysiologic mechanism between any of the two, much less all three conditions, has yet to be described” . The proposed mechanisms remain speculative:

ECM-mast cell interactions: Mast cells anchor to extracellular matrix proteins (fibronectin, vitronectin) via integrins. Bidirectional signaling means abnormal ECM composition could theoretically alter mast cell activation thresholds and mediator release patterns.
Abnormal tissue remodeling: Mast cell proteases contribute to ECM remodeling. A vicious cycle might develop where abnormal ECM triggers mast cell activation, which causes further ECM abnormalities.
Epidemiological association: Approximately 31% of patients with both POTS and EDS also have MCAS, compared to 2% of those without EDS. However, diagnostic criteria heterogeneity limits interpretation.

While the clinical co-occurrence is real, the mechanistic explanation remains a hypothesis rather than established science.

Pathway 4: Tissue Fragility → Purinergic Signaling (SPECULATIVE).

This pathway connects EDS tissue fragility to the cell danger response hypothesis of ME/CFS :

Microtrauma from daily activities: EDS patients experience more joint subluxations, soft tissue injuries, and tissue stress from normal activities due to structural fragility.
ATP release: Damaged and stressed cells release ATP into the extracellular space. This is a universal cellular alarm signal.
Purinergic receptor activation: Extracellular ATP activates P2X and P2Y receptors, triggering the cell danger response—a metabolic shift toward a protective but hypometabolic state.
Chronic activation: If tissue fragility causes ongoing microtrauma, the purinergic alarm system might never fully reset, maintaining the hypometabolic state characteristic of ME/CFS.

This pathway is mechanistically plausible but entirely unvalidated. No studies have measured extracellular ATP levels or purinergic receptor activation in EDS patients.

Pathway 5: Small Fiber Neuropathy as Common Downstream Pathway (MODERATE EVIDENCE).

Small fiber neuropathy (SFN) may represent a convergent mechanism linking EDS structural pathology to ME/CFS-like symptoms:

Universal SFN in EDS: All 24 EDS patients in one study showed decreased intraepidermal nerve fiber density consistent with SFN, with 95% meeting criteria for neuropathic pain .
SFN in ME/CFS: ME/CFS patients show evidence of C-fiber denervation on quantitative sensory testing, with 31% meeting POTS criteria and 34% showing non-length-dependent SFN patterns .
Autonomic small fibers: SFN affects not only sensory nerves but also autonomic small fibers controlling heart rate, blood pressure, digestion, sweating, and temperature regulation—explaining the widespread autonomic dysfunction in both conditions.

SFN may be where the EDS structural abnormality and the ME/CFS functional abnormality converge, though whether SFN in EDS has the same etiology as SFN in ME/CFS remains unknown.

Speculation: Defective Extracellular Matrix Alters Lymph Node Architecture and Viral Clearance

Certainty: 0.25. Highly speculative—no direct evidence links hEDS connective tissue defects to lymph node microenvironment dysfunction. The hypothesis bridges two independently supported observations (hEDS susceptibility to ME/CFS and viral persistence in lymph nodes) with a plausible but untested mechanistic bridge.

Lymph nodes are structurally dependent on a fibroblastic reticular cell (FRC) network embedded in extracellular matrix (ECM). This collagen- and fibronectin-rich scaffold organizes immune cell trafficking, antigen presentation, and pathogen clearance. In hEDS, systemic collagen defects alter ECM composition throughout the body. We speculate that this extends to lymphoid tissue, creating altered lymph node microenvironments where viral clearance is impaired.

Proposed Mechanism. Defective ECM in hEDS lymph nodes may produce:

Looser FRC networks: Altered stromal architecture could change the spatial organization of T cell zones and B cell follicles, reducing the efficiency of antigen presentation and immune synapse formation.
Abnormal conduit systems: The FRC conduit system filters soluble antigens by size; altered collagen composition could change filtration properties, affecting which antigens reach dendritic cells.
Permissive niches for viral persistence: If the Kol et al. FIP model applies to humans, virus-harbouring lymphocytes may find sanctuary in structurally abnormal lymph node regions where immune surveillance is spatially disorganized.

This would explain why hEDS patients are disproportionately susceptible to post-viral ME/CFS: the same connective tissue defect that causes joint hypermobility also compromises the structural infrastructure of immune defense.

Testable Predictions.

Lymph node biopsies or fine-needle aspirates from hEDS patients will show altered FRC network density and collagen composition compared to non-hEDS controls.
Advanced lymph node imaging (contrast-enhanced ultrasound or MR lymphangiography) will reveal structural differences in hEDS patients.
Post-COVID hEDS patients will show higher rates of persistent viral antigen in lymphoid tissue than non-hEDS post-COVID patients matched for illness duration and severity.
In vitro, culturing immune cells on hEDS-derived versus normal ECM scaffolds will show reduced viral clearance efficiency.

Falsifiability. This hypothesis would be falsified if lymph node architecture in hEDS patients is structurally normal on biopsy and imaging, or if viral persistence rates are equal in hEDS and non-hEDS ME/CFS patients. It would also be weakened if the hEDS–ME/CFS association is fully explained by autonomic dysfunction and MCAS without any lymphoid tissue involvement.

Limitations. No studies have examined lymph node histology in hEDS. The ECM defect in hEDS is heterogeneous and incompletely characterized; its effects on lymphoid tissue are entirely extrapolated. Lymph node biopsy is invasive, limiting feasibility. The hypothesis cannot explain ME/CFS in patients without connective tissue disorders.

10.3 The Deconditioning Spiral

A vicious cycle may amplify the initial pathology. In hEDS patients with dysautonomia :

78% report exercise intolerance as a primary symptom
Sedentary behavior increased from 44% to 85% after symptom onset
Dysautonomic patients showed smaller cardiac chamber sizes and reduced left ventricular end-diastolic volume—cardiac atrophy from deconditioning

The proposed cycle:

Connective tissue abnormality → orthostatic intolerance
Orthostatic intolerance → exercise avoidance
Exercise avoidance → cardiovascular deconditioning
Deconditioning → reduced blood volume, cardiac atrophy
Reduced cardiovascular capacity → worsened orthostatic intolerance

This spiral is similar to—but distinct from—ME/CFS, where post-exertional malaise adds an additional constraint. In pure hEDS without ME/CFS, carefully graded exercise may help break the cycle. In ME/CFS with hEDS, the PEM constraint means standard exercise approaches are contraindicated (see Section Graded Exercise Therapy is Harmful).

10.4 Synthesis: EDS as Susceptibility Factor

Hypothesis: Connective Tissue Disorders as ME/CFS Susceptibility Factors

Hypermobility spectrum disorders do not cause ME/CFS directly but dramatically increase susceptibility through multiple mechanisms:

Lower trigger threshold: Pre-existing autonomic dysfunction means less physiological reserve. A viral infection that a person with normal connective tissue might recover from could tip an hEDS patient into chronic illness.
Additional perpetuating mechanisms: Craniocervical instability, vascular dysfunction, and mast cell activation provide additional “locks” that maintain the disease state once triggered.
Impaired recovery capacity: Tissue repair mechanisms are compromised. The body cannot fully restore homeostasis after an acute insult.
Diagnostic confusion: Symptom overlap delays ME/CFS diagnosis and appropriate management. Patients may be told their symptoms are “just EDS” when they actually have both conditions.

This model explains the high comorbidity without requiring that EDS directly causes ME/CFS. Instead, EDS removes the physiological buffer that would normally allow recovery from acute triggers.

Open Question: Research Priorities for EDS-ME/CFS Connection

Critical unanswered questions include:

Does ME/CFS in hEDS patients have the same pathophysiology as ME/CFS in non-hypermobile patients, or are these distinct conditions with overlapping symptoms?
Can early, aggressive management of dysautonomia in hEDS patients prevent progression to ME/CFS after viral triggers?
What is the true prevalence of craniocervical instability in ME/CFS, and does surgical correction improve ME/CFS-specific outcomes?
Do hEDS patients show elevated extracellular ATP or purinergic activation compared to controls?
Is small fiber neuropathy in EDS mechanistically related to SFN in ME/CFS?

Answering these questions could identify preventive strategies and targeted treatments for this high-risk subgroup.

Speculation: Acquired Progressive Hypermobility via Mast Cell Mediator-Mediated Connective Tissue Degradation

(Certainty: 0.35 — individual mechanistic steps well-supported; causal chain from MC activation to clinical hypermobility progression is assembled from separate literatures, not demonstrated in a single study. Origin: brainstorm.)

The premise that hypermobility is exclusively genetic is questioned by the failure to identify a molecular defect in most hEDS cases despite extensive genetic search (Martin 2019) — though this absence may reflect GWAS technical limitations (rare variants, non-coding regions) rather than confirmation of acquired mechanisms — and by the clinical observation that some patients report new or worsening hypermobility following infections, new allergies, or MCAS onset. Mast cell mediators provide a plausible mechanism for acquired connective tissue degradation:

Mechanistic chain (component steps individually supported, chain unvalidated). The full chain invokes many intermediate nodes, but the critical testable link is MC activation → MMP/histaminylation → measurable CT degradation. If blocking MC activation does not reduce CT degradation, the intermediate detail is irrelevant regardless of how many component steps are individually validated.

Mast cell tryptase activates MMP-3 and MMP-13 zymogens (certainty 0.70) (Magarinos et al. 2013)
Mast cell chymase directly activates procollagenase MMP-1 via Leu83-Thr84 cleavage (certainty 0.75) (Saarinen et al. 1994)
Mast cell mediators regulate MMP-mediated collagen degradation in tissue ECM (certainty 0.70) (Janicki et al. 2006)
Histaminylation independently alters collagen matrix mechanics via covalent histamine modification (certainty 0.70) (Zhu et al. 2026)
Chymase negatively impacts bone ECM via osteoblast effects (certainty 0.65) (Lind et al. 2022)

Proposed causal cascade. Triggering event (infection, newly acquired allergy, persistent pathogen) → mast cell activation → sustained tryptase/chymase release + histaminylation → cumulative collagen/ECM degradation → progressive ligament/joint capsule laxity → clinically significant increase in hypermobility (Beighton score increase, new joint instability, worsening POTS from vascular CT changes). The transition from acute to chronic mast cell activation — the critical step that distinguishes a transient post-infectious phenomenon from a progressive CT disorder — is not mechanistically specified here and represents the central unresolved question shared by all post-infectious chronic illness models.

Gates. This hypothesis is constrained by three null findings: (a) hereditary alpha-tryptasemia (elevated tryptase genetics) does NOT concentrate in hypermobility disorders (Vazquez et al. 2022) — genetic tryptase elevation alone is insufficient; (b) random clinical tryptase and urinary mast cell markers are NOT elevated in high-MC-score hEDS patients (Wilson et al. 2026) — dynamic, event-associated measurement is required; (c) no molecular defect has been identified for most hEDS (Martin 2019) — the field is open to non-genetic mechanisms but has not yet demonstrated them.

Clinical prediction. Patients presenting with progressive or new-onset hypermobility after an infectious trigger or after developing new allergies/atopic disease will show elevated mast cell activation markers (provoked, not random tryptase; urinary N-methylhistamine/LTE4/PGD2 metabolite) and elevated serum MMP-3/MMP-9 relative to patients with stable, lifelong hypermobility. Mast cell stabilization (cromolyn, ketotifen) or anti-IgE therapy (omalizumab, if IgE-mediated) should slow or arrest hypermobility progression in responsive patients — a prediction not yet tested in any clinical trial.

Falsifiable prediction. In a prospective study of post-infectious hypermobility patients (n≥30), those randomized to mast cell stabilizer + antihistamine therapy will show no significant Beighton score increase over 12 months, while untreated controls will show ≥1-point increase. Falsified if mast cell stabilizer treatment does not alter hypermobility progression rate. A stronger falsification: if provoked mast cell mediator levels (tryptase >20%+2 ng/mL above baseline during symptoms) are normal in patients with progressive hypermobility, the MC→CT chain is incorrect for that patient.

Limitations. The full causal chain is assembled from separate literatures and has not been demonstrated end-to-end. Progressive hypermobility may reflect cumulative mechanical joint damage from MCAS-driven proprioceptive dysfunction rather than direct CT degradation — an alternative explanation that is equally consistent with the clinical observation. The Wilson 2026 prevalence data (25% high MC score in hEDS/HSD) suggest MCAS is present in a subset, not universal. Beighton score ceiling effects may mask progression in already-highly-hypermobile patients. Competing mechanism: TNXB haploinsufficiency (tenascin-X deficiency) causes hypermobility in ~5–10% of cases via a non-MC mechanism (Imanaka et al. 2026).

Clinical Finding: Clinical Algorithm: Evaluating Progressive Hypermobility for MC-Driven Causes

Evidence type: mechanistic synthesis; no interventional data. Severity applicability: all levels where progressive hypermobility is reported.

In patients with hypermobility spectrum disorders or hEDS who report progressive worsening of joint laxity, subluxations, or POTS severity — particularly following a known infectious trigger, mould exposure, or development of new allergies — a targeted evaluation for mast-cell-driven connective tissue degradation may identify reversible factors:

Step 1 — Identify triggers:

Tickborne disease serology (Lyme, Bartonella — note diagnostic limitations of seronegative disease)
Recent or prior COVID-19 infection (temporal relationship to hypermobility onset/worsening)
Mould/mycotoxin exposure history (water-damaged building, visible mould)
Other infections: EBV reactivation, persistent viral infections
Newly acquired allergies (including food, environmental — may postdate hypermobility onset)
New autoimmune disease onset

Step 2 — Evaluate mast cell activation:

Provoked serum tryptase during symptoms (baseline + 1–4h post-symptom-onset; >20%+2 ng/mL above baseline is the international consensus threshold). Practically, this requires patients to have a standing lab order and present for phlebotomy during a symptomatic episode — logistically challenging outside dedicated research protocols. A pragmatic alternative is to collect a baseline during an asymptomatic period, then instruct the patient to present for a second draw during their next spontaneous episode.
24-hour urinary N-methylhistamine, LTE4, PGD2 metabolite (paired baseline-vs-symptomatic day if feasible)
Screen for hereditary alpha-tryptasemia (TPSAB1 copy number) — a genetic confound with chronically elevated baseline tryptase
Evaluate for clinical MCAS features: flushing, urticaria, dermatographism, food reactions, autonomic instability, GI symptoms
Note: random/single-point tryptase is frequently normal even in patients with clinically significant MCAS; negative random labs do not exclude MCAS

Step 3 — If MC activation confirmed or clinically suspected:

H1+H2 antihistamine combination (e.g., rupatadine 10mg + famotidine 20mg BID) as first-line, low-risk trial
If refractory or IgE-elevated (total IgE >100 IU/mL): consider mast cell stabilizer (cromolyn, ketotifen) or anti-IgE therapy (omalizumab — specialist prescribing only; access is limited by cost of $15,000–$30,000 per year, prior authorization requirements, and documentation of a specific IgE-mediated indication; not currently approved for hypermobility or MCAS by most insurers)
Track hypermobility metrics (Beighton score, subluxation frequency, POTS orthostatic tolerance) at baseline and every 3 months. Note: Beighton score is a screening tool, not a validated treatment monitoring instrument; no validated metric exists for tracking hypermobility progression in clinical practice, and monitoring relies on clinical judgment.
Monitor for treatment response: stabilisation of previously progressive hypermobility is the target; reversal is not expected

Step 4 — Stopping criteria:

No improvement in hypermobility metrics or MCAS symptom burden after 6 months of optimal MC-targeted therapy → discontinue; evaluate alternative mechanisms (TNXB deficiency, cumulative mechanical damage, proprioceptive dysfunction)
Treatment-emergent adverse effects → discontinue per standard clinical judgment

Limitations. This algorithm is based on mechanistic plausibility, not clinical trial evidence. No RCT has tested whether mast-cell-directed therapy alters hypermobility progression. Individual steps (provoked tryptase testing, urinary mediator panels) require access to specialised laboratory services that may not be available in all settings. The Wilson 2026 finding that random MC markers are normal in most hEDS patients means the algorithm depends on provoked/dynamic testing that is logistically demanding. Stopping rule at 6 months is pragmatic consensus, not evidence-based. Risk of over-investigation: most progressive hypermobility may reflect cumulative mechanical joint damage from proprioceptive dysfunction, not active MC-driven degradation — applying this algorithm broadly risks inappropriate treatment.

Limitation: Hereditary Alpha-Tryptasemia Does Not Explain MCAS-Hypermobility Link

Hereditary alpha-tryptasemia (HaT; extra copies of TPSAB1, ~5–6% of the general population) causes chronically elevated baseline tryptase and an MCAS-like clinical picture. However, HaT does NOT concentrate in hypermobility disorders (Vazquez et al. 2022) (certainty 0.70). This is an important constraint on the MCAS→hypermobility hypothesis: constitutive, genetically elevated tryptase is insufficient to produce hypermobility — episodic, high-amplitude degranulation (i.e., acquired mast cell activation) may be required. HaT should be excluded in patients being evaluated for MCAS-hypermobility overlap, but the absence of HaT enrichment in hEDS does not negate the role of acquired MC activation, as these represent different physiological states (chronic low-grade elevation vs episodic high-amplitude degranulation with different downstream mediator profiles).

10.5 Cross-Disease Models for IgE-Mediated Connective Tissue Degradation

The IgE → mast cell → MMP → connective tissue degradation chain described in Section Acquired Progressive Hypermobility via Mast Cell Mediator-Mediated Connective Tissue Degradation has parallels in other disease contexts that illustrate mechanistic plausibility without requiring ME/CFS-specific validation:

Speculation: Cardiac Extracellular Matrix Remodelling as Mast Cell-MMP Degradation Prototype

(Certainty: 0.60 — direct mechanistic analogy from replicated cardiac ECM model; no ligament-specific data in ME/CFS. Origin: brainstorm.)

Janicki 2006 established that cardiac mast cells regulate MMP-mediated collagen degradation in ventricular remodelling — the same mediators (tryptase, chymase, TNF-α) that degrade cardiac collagen are the ones proposed to degrade ligamentous ECM in MCAS-associated hypermobility (Janicki et al. 2006). Cardiac ECM and ligamentous ECM share conserved substrate proteins (collagen I/III, fibronectin, proteoglycans), and mast cell protease targets are conserved across tissues. This is the most directly analogous established pathway: tissue ECM is degraded by local mast cells via MMP activation in a well-replicated disease model.

The cardiac model establishes that chronic mast cell activation is sufficient to produce measurable ECM degradation in vivo, and that the temporal scale of degradation (weeks to months in animal models) is consistent with the clinical observation of progressive hypermobility developing over months to years in MCAS-hEDS patients. Translating this mechanism to ligamentous and vascular CT requires demonstration of the same MC→MMP→ECM chain in those tissues, which has been initiated in osteoblast (Lind 2022 (Lind et al. 2022)) and synovial models (Guo 2021 (Y. Guo et al. 2021)) but not in ligament or joint capsule specifically.

Falsifiable prediction: Serum MMP-3, MMP-9, and TIMP levels in MCAS-hEDS patients should parallel the profiles observed in cardiac MC-activation models (adjusted for age and sex). Ligament biopsy MMP expression should mirror cardiac MMP profiles from heart failure studies. If MCAS-hEDS patients show fundamentally different MMP profiles from cardiac MC activation models, tissue-specific mediators must be invoked.

Limitations: Cardiac ECM physiology differs from ligamentous ECM in mechanical loading, repair rate, and resident cell populations. Cardiac mast cells are studied primarily in the context of TNF-α-driven remodelling, not IgE-driven degranulation specifically — the upstream trigger differs fundamentally even if downstream MMP targets overlap. Translation requires tissue-specific validation.

Speculation: Periodontitis as Human Model of IgE-Mediated Connective Tissue Degradation

(Certainty: 0.55 — strongest human precedent for IgE→MC→MMP→CT degradation with measurable clinical outcomes. Origin: brainstorm.)

Periodontitis is the best-established human model of IgE-mediated connective tissue degradation. IgE against oral bacteria (e.g., Porphyromonas gingivalis) drives local mast cell activation → MMP-mediated degradation of the periodontal ligament and alveolar bone → measurable attachment loss and eventual tooth loss. The causal chain — IgE → MC activation → MMP release → CT degradation → clinical outcome (attachment loss) — is identical to the proposed mechanism for MCAS-associated hypermobility. Periodontitis treatment targeting this axis (subantimicrobial doxycycline 20 mg BID, which inhibits MMP-2/9 without antimicrobial effects) has been in clinical use for decades and preserves periodontal attachment in controlled trials. Note that this analogy is a cross-disease inference — the periodontitis literature establishing the IgE→MC→MMP chain is well-developed but was not part of the Phase 1 literature search for this topic; the analogy is presented as mechanistic plausibility, not as direct evidence from Phase 1 papers.

The periodontitis parallel establishes three key principles: (a) that local IgE-mediated mast cell activation can produce clinically significant CT degradation in humans; (b) that MMP inhibition can interrupt this degradation; and (c) that subantimicrobial doxycycline dosing is safe for chronic use. The critical open question is whether the same principles apply to ligament and joint capsule CT in hEDS/MCAS patients.

Falsifiable prediction: MCAS-hEDS patients will show elevated IgE to common periodontal pathogens compared to non-MCAS hEDS patients. If confirmed, the model predicts that treatments reducing MC activation or MMP activity (subantimicrobial doxycycline, ketotifen, omalizumab) should slow Beighton score progression rate by $\geq$ 50% at 12 months relative to pre-treatment trajectory, analogous to periodontal attachment-loss preservation. Falsified if IgE to periodontal pathogens is not elevated (OR $\leq$ 1.5 vs non-MCAS hEDS), or if Beighton score progression rate is reduced by $<$ 30% in treated patients despite adequate MC suppression.

Limitations: Periodontal and ligamentous CT have different mechanical environments, turnover rates, and inflammatory cell populations. IgE against oral bacteria is a luminal exposure; IgE in MCAS-hEDS may target systemic or self-antigens, not oral microbes. Periodontal MMP inhibition trials used doxycycline; the MC stabiliser component of the proposed dual therapy lacks human CT-outcome data.

Open Question: Can Omalizumab Serve as a Therapeutic Probe for the IgE→MC→CT Hypothesis?

Omalizumab (anti-IgE monoclonal antibody, approved for allergic asthma and chronic urticaria) is safe and effective for refractory MCAS (Matheny 2025 systematic review, 28 patients, 61% partial and 18% complete response (Matheny, Craig, and Al-Shaikhly 2025)), but no study has examined its effect on connective tissue integrity or hypermobility outcomes. The drug’s mechanism — IgE sequestration → reduced FcεRI activation → reduced mast cell degranulation — is specific to the IgE pathway, making it a useful therapeutic probe for this specific pathway: if omalizumab reduces mast cell activation and slows hypermobility progression, the IgE→MC→CT mechanism is supported; if mast cell activation decreases without affecting CT outcomes, the IgE→MC link exists but the MC→CT degradation step is wrong or the damage is cumulative and irreversible by the time of treatment; if neither MC activation nor CT outcomes change, the IgE pathway is not the dominant degranulation route in that patient.

The Wilson 2026 finding of elevated IgE in high-MC-score hEDS/HSD patients (n=2141, p=0.0004, certainty 0.45 (Wilson et al. 2026)) — combined with the omalizumab safety database — makes this a feasible proof-of-mechanism trial. However, a negative result would only falsify the IgE-specific route, not the broader MC→CT chain: if the dominant degranulation pathway is MRGPRX2, complement, or TLR-mediated (see Limitation The Dominant Mast Cell Activation Pathway in ME/CFS May Be Non-IgE), omalizumab would have no effect regardless of whether MC-driven CT degradation is correct. A 12-month open-label study in IgE-elevated hEDS/MCAS patients measuring provoked tryptase, serum MMP-3/9, histaminylation markers, and Beighton score trajectory could test the IgE pathway specifically. A positive result would justify an RCT; a negative result would redirect research toward non-IgE MC activation routes, leaving the MC→CT chain itself untested.

(Origin: brainstorm.)

10.6 Competing Explanations and Evidence Limitations

Limitation: Acquired Connective Tissue Degradation May Be Entirely Non-MC Mediated in ME/CFS

The connective tissue changes observed in ME/CFS patients (increased joint laxity, progressive hypermobility, skin extensibility) have plausible non-MC explanations that do not require invoking mast cells: (a) reduced collagen synthesis from metabolic insufficiency — mitochondrial dysfunction in ME/CFS (certainty 0.70) reduces ATP availability for prolyl/lysyl hydroxylase-dependent collagen crosslinking; (b) systemic inflammation activating MMPs independently of mast cells — neutrophil-derived MMP-8/9 and macrophage-derived MMP-12 are elevated in inflammatory states and can degrade collagen without mast cell involvement; (c) HPA axis dysfunction — chronically elevated or dysregulated cortisol promotes CT catabolism (skin thinning, bone loss, tendon weakening); (d) mechanical unloading in bedbound patients — prolonged immobility causes CT atrophy and reduced collagen content independent of inflammation. The skeletal asymmetry hypothesis (Skeletal Asymmetry as a Primary Mechanical Trigger of the ME/CFS Cascade) identifies a specific instance of mechanical unloading with a distinct direction: rather than immobility causing CT atrophy, asymmetric mechanical loading creates a chronic compensatory demand that independently drives metabolic and autonomic dysfunction through four mechanistically explicit pathways. All of these are documented in ME/CFS to varying degrees. The MC→CT degradation mechanism may be one contributor among several non-MC factors, not the dominant driver. No study has partitioned the relative contribution of MC-mediated versus non-MC-mediated CT degradation in ME/CFS or hEDS. (Origin: brainstorm.)

Limitation: The Dominant Mast Cell Activation Pathway in ME/CFS May Be Non-IgE

Mast cell degranulation can occur via at least five pathways: IgE-FcεRI, MRGPRX2 (neuropeptide/drug-activated), complement C3a/C5a, TLR (pathogen-associated), and mechanical stimulation. Wilson 2026 found elevated IgE in the high-MC-score hEDS/HSD subset (p=0.0004, n=2141), but the effect size is modest and IgE elevation may be a general marker of type 2 immune skewing (atopy, allergy) rather than a specific causal driver of CT degradation (Wilson et al. 2026). The dominant MC activation triggers in ME/CFS — neuropeptides (substance P, VIP via MRGPRX2), viral RNA/LPS (via TLRs), or mechanical stress (via integrins/CADM1) — may all operate independently of IgE. If the actual CT-degrading mediators (tryptase, chymase, histamine) are released by non-IgE triggers, then anti-IgE therapy (omalizumab) would be ineffective even if the MC→CT chain is correct. This is a critical constraint on the therapeutic probe design described in Open Question Can Omalizumab Serve as a Therapeutic Probe for the IgE→MC→CT Hypothesis?: the probe only tests the IgE pathway, not the MC→CT chain itself. (Origin: brainstorm.)

Limitation: Key Evidence Is from Retrospective Preprints and Cross-Disease Analogies

The largest clinical dataset supporting the MCAS-hEDS IgE association (Wilson 2026, n=2141) is a preprint — peer review is pending, and findings may change (Wilson et al. 2026). The strongest mechanistic evidence for MC→CT degradation (Magarinos2013 (Magarinos et al. 2013), Saarinen1994 (Saarinen et al. 1994), Janicki2006 (Janicki et al. 2006)) is from non-hypermobility, non-ME/CFS tissue contexts (cartilage, skin, cardiac). The proposed causal chain (IgE → MC → CT degradation → progressive hypermobility) is assembled from separate literatures; no single study has demonstrated all nodes simultaneously. The histaminylation mechanism (Zhu2026 (Zhu et al. 2026)) has been demonstrated in vitro but not in human tissue, and its quantitative contribution to clinical hypermobility is entirely unknown. If any study in this chain fails to replicate in a human hypermobility context, the entire assembly would need revision. (Origin: brainstorm.)

Synthesis: Mast Cell IgE Pathway as a Modifiable Contributor to Progressive Hypermobility

Taken together, these findings converge on a testable model: in a subset of hEDS/HSD patients — particularly those with elevated IgE, post-infectious onset, and progressive rather than lifelong hypermobility — chronic mast cell activation may contribute to ongoing connective tissue degradation via two parallel mechanisms (MMP activation and histaminylation). The strongest constraints are that HaT does not cause hypermobility (genetic tryptase elevation is insufficient (Vazquez et al. 2022)), the dominant MC activation pathway in ME/CFS is likely non-IgE, and competing non-MC degradation mechanisms (metabolic insufficiency, systemic-inflammation-driven MMPs, HPA-mediated catabolism, mechanical unloading) have not been excluded. The omalizumab therapeutic probe Can Omalizumab Serve as a Therapeutic Probe for the IgE→MC→CT Hypothesis? is the most informative near-term experiment for resolving the IgE-specific arm, though a negative result would leave the broader MC→CT chain untested. The clinical algorithm Clinical Algorithm: Evaluating Progressive Hypermobility for MC-Driven Causes is offered as a pragmatic framework pending trial data, not as validated guidance. Three fundamental questions remain unanswered: (1) what fraction of progressive hypermobility is MC-mediated versus non-MC? (2) what is the dominant MC activation pathway in the hEDS/MCAS subset? (3) is the causal direction MC→CT, CT weakness→MC, or bidirectional with different primary entry points by patient?

Synthesis: Connective Tissue Degradation in ME/CFS: Two Convergent Mechanisms, One Structural Vulnerability

Connective tissue pathology in ME/CFS is addressed by three major integration cycles spanning 110+ hypotheses: the full connective tissue disorders plan (64 ideas across pathophysiology, treatment, and biomarkers), the HIF-2α endothelial postviral integration, and the mast cell IgE→CT degradation analysis Mast Cell IgE Pathway as a Modifiable Contributor to Progressive Hypermobility. These converge on two independent degradation mechanisms — mast cell mediator-mediated MMP/histaminylation (via tryptase, chymase, IgE pathway, mechanical stress Acquired Progressive Hypermobility via Mast Cell Mediator-Mediated Connective Tissue Degradation) and HIF-1α-driven ECM remodeling (via MMP-3 upregulation, collagen synthesis suppression, and basement membrane thickening Cardiac Extracellular Matrix Remodelling as Mast Cell-MMP Degradation Prototype) — plus one structural vulnerability: the permissive matrix hypothesis Connective Tissue Matrix Permissivity as the Unifying Mechanism of the hEDS-POTS-MCAS-ME/CFS Cluster posits that the same ECM defect causing hypermobility also enables pathological mast cell tissue infiltration. HIF-2α endothelial dysfunction may compound both mechanisms by impairing vascular delivery of collagen precursors and antioxidant defenses to CT repair sites. The two degradation routes are potentially addressable with existing drugs (MC stabilizers, subantimicrobial doxycycline, belzutifan for HIF-2α), but no study has tested whether intervening in either mechanism alters CT integrity or hypermobility progression. The central unresolved question is whether CT degradation in ME/CFS is primarily MC-mediated, HIF-mediated, both synergistically, or neither — competing non-MC/non-HIF mechanisms (metabolic insufficiency, HPA catabolism, mechanical unloading Acquired Connective Tissue Degradation May Be Entirely Non-MC Mediated in ME/CFS) cannot be excluded without partitioning studies.

11 Skeletal Asymmetry as a Distinct Mechanical Driver

Hypothesis origin: Colette Marie Gerlier (2026-07-21, personal communication) — independent researcher (BTS Biochimie, former neurology EEG technician) — proposed that fixed skeletal asymmetries (rotoscoliosis, DAMI — Des Axes Morphostatiques Insuffisants, ILMI — Inégalité de Longueur des Membres Inférieurs) may act as a primary mechanical upstream driver triggering cascading autonomic, metabolic, and neurological dysfunction in a subset of ME/CFS patients, distinct from the hypermobility/hEDS/CCI mechanisms already discussed.

Speculation: Skeletal Asymmetry as a Primary Mechanical Trigger of the ME/CFS Cascade

Certainty: 0.10. Skeletal asymmetry produces chronic asymmetric mechanical loading that drives metabolic and autonomic dysfunction through four pathways: (1) sustained compensatory muscle overuse on the convexity/longer-limb side → premature anaerobic metabolism, proton accumulation, and the Na⁺/Ca²⁺ overload cascade described in Section WASF3–DRP1 Convergence: Two Routes to the Same Mitochondrial Failure; (2) mechanical compression or irritation of the paravertebral sympathetic chain (T1–L2) by vertebral rotation or osteophytic encroachment → multi-system dysautonomia affecting digestive, respiratory, hepatic, and renal function; (3) asymmetric fascial tension transmitting mechanical stress to sensory nerve trunks (sciatic, pudendal, lateral femoral cutaneous) → peripheral sensitization and allodynia; (4) cervical arthrosis progression with age → worsening craniocervical instability → brainstem compression and vestibular dysfunction, creating a mechanical aging spiral superimposed on the metabolic ratchet (Disease Progression Models).

Origin: Gerlier 2026-07-21, personal communication.

Evidence base. This hypothesis rests on indirect evidence and clinical reasoning, not on direct ME/CFS studies. The relevant literature is sparse:

Sympathetic chain vulnerability. Schulte et al. (2010, (Schulte et al. 2010)) documented sympathetic chain dysfunction (confirmed by sudomotor testing) in 12/31 patients following anterior scoliosis surgery — establishing that the thoracic sympathetic chain is anatomically vulnerable to vertebral pathology. Whether idiopathic spinal asymmetry produces similar autonomic deficits without surgical trauma is entirely untested.
Scoliosis-autonomic co-occurrence. In Rett syndrome (n=913), severe scoliosis (>40°) occurs in 27% of patients alongside autonomic dysfunction driven by MECP2 mutations (Killian et al. 2017). This provides a disease model where scoliosis and dysautonomia co-occur, but does not establish scoliosis as the cause.
Compensatory-overuse precedent. Post-polio syndrome fatigue is driven by metabolic exhaustion of surviving motor neurons compensating for lost ones (Jubelt 2004) — establishing a precedent for compensatory overuse as a fatigue mechanism, though the lesion (motor neuron loss) differs fundamentally from skeletal asymmetry.
Bidirectional ANS-bone regulation. The SNS regulates bone remodeling via $\beta$₂-adrenergic receptors on osteoblasts (He, Zheng, and Jiang 2014), establishing that ANS and skeletal structure interact — but the direction tested is ANS→bone, not bone→ANS.
LLD treatment gap. A systematic review of shoe lifts for leg length discrepancy found they reduce low back pain (1 RCT only, low-quality evidence) but no study assessed fatigue as an outcome (Campbell et al. 2019).
Null signal. Leg strength asymmetry was NOT associated with fatigue or fatigability in multiple sclerosis (n=15) (Proessl, Ketelhut, and Rudroff 2018) — the only study to test an asymmetry→fatigue pathway in a neurological disease found no association, suggesting additional factors are required.

Four-pathway mechanistic model (provisional, untested):

Pathway 1 — Compensatory muscle overuse. A pelvic obliquity of 10–15 mm (common in mild LLD) requires the ipsilateral hip abductors, contralateral lumbar extensors, and cervical stabilizers to work continuously to maintain head and eye level. In healthy individuals, this imposes a ~5–10% increase in basal energy expenditure during standing and walking (biomechanical modeling, no ME/CFS data). In a patient whose cellular energy metabolism is already compromised — whether by WASF3-mediated supercomplex disruption (P. Wang et al. 2023), PDH inhibition, or AMPK suppression — this constant additional demand may be sufficient to breach the critical ATP threshold (WASF3–DRP1 Convergence: Two Routes to the Same Mitochondrial Failure), triggering PEM from trivial activity. This provides a mechanical explanation for why some patients crash from mere standing or slow walking: their baseline energy consumption is elevated by postural compensation before any voluntary activity begins.

Pathway 2 — Sympathetic chain irritation. The paravertebral sympathetic chain runs along the anterolateral aspect of vertebral bodies from T1 to L2. In rotoscoliosis, vertebral rotation displaces the chain asymmetrically; osteophytic encroachment from degenerative changes may directly compress ganglia. Schulte et al. (Schulte et al. 2010) demonstrated that surgical manipulation of this region causes measurable sudomotor dysfunction — a proof of principle that mechanical disruption of the sympathetic chain alters autonomic output. Spontaneous mechanical irritation (rather than surgical transection) would produce a more complex picture: intermittent compression → irritative firing (sympathetic hyperactivity, hyperadrenergic POTS-like state); chronic compression → conduction block → compensatory denervation hypersensitivity. The multi-organ distribution follows sympathetic anatomy: T5–T9 → splanchnic nerves → digestive, hepatic; T10–L1 → renal; T1–T4 → cardiac, pulmonary.

Pathway 3 — Mechanical nerve compression and sensitization. Asymmetric pelvic tilt produces asymmetric fascial tension throughout the myofascial chains. Fascial thickening and loss of glide planes (documented in chronic pain via ultrasound elastography) can entrap peripheral nerves against bony structures. The lateral femoral cutaneous nerve (meralgia paresthetica), sciatic nerve (piriformis syndrome), and pudendal nerve (Alcock canal syndrome) are mechanically vulnerable to pelvic asymmetry. Chronic low-grade compression produces a distinct clinical picture from acute entrapment: fluctuating paresthesia, positional allodynia, and a sensation of “wrongness” rather than lancinating pain. Central sensitization (Central Sensitization and Nociplastic Pain) amplifies these peripheral signals, producing the diffuse hypersensitivity pattern characteristic of ME/CFS.

Pathway 4 — Cervical aging spiral. Skeletal asymmetry below the cervical spine alters head position: a pelvic obliquity is compensated by lumbar scoliosis → thoracic countercurve → cervical tilt to maintain horizontal gaze. This imposes asymmetric loading on cervical facet joints and uncovertebral joints, accelerating osteoarthritic degeneration. With age, osteophyte formation narrows the intervertebral foramina and the craniocervical space, progressively worsening any pre-existing CCI or creating it de novo. Unlike the acute CCI from ligamentous laxity (Cervical Sympathetic Ganglion Compression in CCI), this is a slow mechanical progression — a “wear and tear” CCI that worsens with age, explaining the clinical observation that some ME/CFS patients deteriorate in their 40s–60s despite stable disease activity.

How does the asymmetry originate? Competing hypotheses. The four pathways describe the cascade once skeletal asymmetry is established, but its origin is an open question with multiple plausible explanations:

Developmental/traumatic (Gerlier hypothesis). Gerlier (2026-07-21, personal communication) suggests an early cervical trauma — often at birth (subluxation during delivery) — initiates a cascade of asymmetric muscle tonus, arched posture, and progressive axial rotation. The body compensates via spiralization (rotoscoliosis) to cancel the lateral arc and maintain a horizontal skull base. These rotoscolioses typically have a low Cobb angle, become structural early, are minimally progressive, and produce a “flat back” appearance assumed benign. Gerlier notes digestive disturbances may appear by day 21 of life — consistent with vagal/sympathetic compression at C1–C2 before the scoliosis is radiologically detectable. The intervention window would be neonatal, not adult.
Congenital/genetic. Idiopathic scoliosis has a heritable component (twin concordance 73% monozygotic vs 36% dizygotic), and congenital vertebral anomalies (hemivertebrae, butterfly vertebrae) produce fixed asymmetry from birth. In this model, the asymmetry is genetically programmed; the mechanical cascade is an inevitable downstream consequence activated when metabolic reserve declines with age or illness.
Acquired postural. Prolonged asymmetric loading (occupational, sport-specific, post-traumatic) can produce functional scoliosis that becomes structural over time through Wolff’s law remodeling. Deconditioning from severe ME/CFS — prolonged bedrest, muscle wasting — predictably produces acquired pelvic obliquity and antalgic scoliosis, reversing the proposed causal direction.
Idiopathic. As with most scoliosis (80%+ are classified “idiopathic”), the origin may be multifactorial and unknown — a combination of subtle genetic predisposition, neuromuscular coordination variants, and biomechanical chance. The four-pathway cascade downstream is testable regardless of which upstream origin applies.

Distinguishing these competing origin hypotheses requires pre-illness imaging or longitudinal birth cohorts — neither exists for ME/CFS. The four-pathway cascade is agnostic to origin: any fixed skeletal asymmetry, however acquired, would impose the same compensatory load.

Testable predictions:

ME/CFS patients with >10 mm leg length discrepancy (measured by standing pelvic radiograph) will show higher standing norepinephrine and more severe POTS than LLD-matched controls without ME/CFS
EOS full-spine radiography in a cohort of n=200 ME/CFS patients will reveal a bimodal distribution: one cluster with normal sagittal/coronal alignment, and a second cluster with rotoscoliosis >10°, pelvic obliquity >10 mm, or both — and the second cluster will have higher COMPASS-31 autonomic symptom scores
Shoe lift correction of LLD >10 mm in ME/CFS patients will produce measurable reduction in 24h heart rate and standing norepinephrine within 4 weeks (n-of-1 trial design acceptable for initial test; RCT required for generalizability)
Cervical MRI in ME/CFS patients aged >50 will show greater uncovertebral osteophytosis and foraminal stenosis than age-matched sedentary controls, independent of hEDS status

Treatment implications (if confirmed): Postural correction (shoe lifts, custom orthotics, targeted physiotherapy for pelvic realignment) would address the mechanical root cause rather than downstream metabolic consequences. Cervical decompression surgery would have a stronger rationale in patients with documented progressive foraminal stenosis. However, with current evidence (certainty 0.10), no treatment recommendation can be made. Osteopathic manipulative treatment is not endorsed as a diagnostic probe — unblinded manual therapy where “improvement confirms the mechanism” is confirmation bias by design, and OMT has no validated mechanism for sympathetic chain modulation.

Consequence: If skeletal asymmetry contributes to ME/CFS in even a small subset of patients, a simple, low-cost intervention — shoe lift or postural physiotherapy — could be disease-modifying for that subgroup. But the hypothesis is currently untested at every link in the chain, and the one study testing an asymmetry→fatigue pathway in a related disease (Proessl 2018, MS) found no association. This is a high-impact/low-probability hypothesis that requires systematic biomechanical and autonomic phenotyping before clinical translation.

Clinical safety note. The neonatal subluxation origin hypothesis described above should not be interpreted as endorsing cervical manipulation in infants — no evidence supports such intervention, and high-velocity cervical manipulation in infants carries documented risk of severe iatrogenic injury. This is a research hypothesis about etiology, not a clinical recommendation for any age group.

Alternative formulation: the two-hit model. A more defensible framing recasts skeletal asymmetry not as a primary driver but as a permissive substrate that lowers the threshold for developing ME/CFS when a second hit occurs (infection, trauma, hormonal change). In this model, asymmetry alone rarely causes disease — explaining why most people with scoliosis (2–3% of the general population) never develop ME/CFS — but it reduces reserve capacity across autonomic, metabolic, and pain-processing systems such that a trigger that would be subclinical in a symmetric person becomes disease-initiating in an asymmetric person. Positioning this as a “risk factor” model does not immunize it from falsification: in a prospective post-infectious cohort, pre-existing LLD >10 mm would be expected to increase the odds of developing ME/CFS at 12 months by ≥2× compared to symmetric individuals; if OR ≈ 1.0, the claim is false regardless of framing.

Missing mechanistic link: mechanotransduction. The four-pathway model posits that compensatory muscle overuse drives anaerobic metabolism, but does not specify how sustained mechanical load becomes mitochondrial failure. Muscle cells express mechanosensitive ion channels (TRPV4, Piezo1) that open under sustained stretch, admitting Ca²⁺. In ME/CFS muscle where baseline Ca²⁺ handling is already dysregulated, sustained stretch-induced Ca²⁺ influx may convert a normally adaptive signal (CaMKII→PGC-1α→mitochondrial biogenesis) into the pathological Na⁺/Ca²⁺ overload cascade. This hypothesis would be falsified if TRPV4/Piezo1 expression and stretch-induced Ca²⁺ responses are normal in ME/CFS muscle biopsies. (Origin: brainstorm.)

Limitation: Skeletal Asymmetry Hypothesis: Evidence Base is Near-Zero

The causal chain proposed in Skeletal Asymmetry as a Primary Mechanical Trigger of the ME/CFS Cascade has no direct ME/CFS evidence at any link. The literature base consists of: one surgical complication study (n=31, thoracic approach, not idiopathic scoliosis) (Schulte et al. 2010); one genetic disease model (Rett syndrome, MECP2-driven, n=913) (Killian et al. 2017); one review establishing the concept of compensatory-overuse fatigue (post-polio, motor neuron lesion, not skeletal asymmetry) (Jubelt 2004); one systematic review of shoe lifts for LLD (low-quality, fatigue not assessed) (Campbell et al. 2019); and one null study (asymmetry not associated with fatigue in MS, n=15) (Proessl, Ketelhut, and Rudroff 2018). None of these directly support the hypothesis. Furthermore, the clinical entities DAMI and ILMI yield zero PubMed results — they are observational concepts from French clinical practice, not validated measurement constructs. This hypothesis should be considered a structured clinical intuition meriting systematic investigation, not a mechanism with evidentiary support. The 0.10 certainty reflects the near-total absence of direct evidence, not a negative assessment of the hypothesis’s plausibility.

Consequence: Clinicians should not use postural correction or osteopathic manipulation as ME/CFS treatments based on current evidence. For researchers, the structured four-pathway model provides testable predictions that could be investigated with existing technology (EOS radiography, COMPASS-31, n-of-1 shoe lift trials) at modest cost.

Additional caveats.

Prevalence mismatch. Scoliosis (Cobb angle >10°) affects 2–3% of the general population, yet ME/CFS prevalence is 0.2–0.4%. If skeletal asymmetry were a strong driver, ME/CFS prevalence should be closer to scoliosis prevalence — it is not, by an order of magnitude. This is not fatal to the hypothesis (most people with a risk factor do not develop the disease), but it forces any viable model into a two-hit or permissive-substrate framing.

Directionality reversal. The entire four-pathway model assumes skeletal asymmetry precedes ME/CFS. But deconditioning from severe ME/CFS — prolonged bedrest, muscle wasting, loss of postural tone — predictably produces acquired functional asymmetry (pelvic tilt from muscle imbalance, antalgic scoliosis from asymmetric pain). The Proessl 2018 null result in MS (Proessl, Ketelhut, and Rudroff 2018) further suggests that asymmetry alone, even in a neurologically compromised patient, does not cause fatigue. Distinguishing congenital/developmental asymmetry from acquired post-morbid asymmetry requires pre-illness radiographs or longitudinal post-infectious cohorts — neither exists.

DAMI/ILMI absence from the validated measurement literature. The clinical concepts DAMI (Des Axes Morphostatiques Insuffisants) and ILMI (Inégalité de Longueur des Membres Inférieurs) yield zero PubMed results. They originate from French clinical observation and have no validated measurement protocols, no inter-rater reliability studies, and no established norms. A hypothesis whose core constructs lack operational definitions cannot be falsified until those definitions exist — a measurement gap that must be addressed before the mechanistic hypothesis can be tested.

Rett syndrome model is a common-cause not a causal-chain model. The co-occurrence of scoliosis and autonomic dysfunction in Rett syndrome (Killian et al. 2017) is driven by a shared genetic cause (MECP2), not by scoliosis causing dysautonomia. The Rett model supports that skeletal and autonomic abnormalities can co-occur in a single disease, but does not support the directional claim that skeletal asymmetry drives autonomic dysfunction.

Schulte2010 is a surgical lesion, not a spontaneous asymmetry model. The documented sympathetic chain dysfunction (Schulte et al. 2010) occurred after anterior surgical approach to the thoracic spine — a controlled mechanical disruption qualitatively different from the low-grade, chronic, intermittent compression hypothesised for idiopathic skeletal asymmetry. Surgical transection is to chronic compression what amputation is to neuropathy: an extreme case that demonstrates vulnerability but does not model the natural disease.

Speculation: Skeletal Asymmetry as a Convergent Mechanical Framework: Synthesis

Certainty: 0.10. The four-pathway model (Skeletal Asymmetry as a Primary Mechanical Trigger of the ME/CFS Cascade), the thoracolumbar sympathetic extension (Thoracolumbar Sympathetic Chain Irritation from Skeletal Asymmetry), the mechanical/postural subgroup definition (Skeletal Asymmetry as a Subgroup-Defining Feature), and the mechanical nerve compression mechanism (Chapter Symptom-Producing Mechanisms in ME/CFS) collectively propose that fixed skeletal asymmetry — regardless of whether its origin is developmental (Gerlier neonatal subluxation hypothesis), congenital, acquired, or idiopathic — may act as a distinct upstream driver of ME/CFS pathophysiology that operates through mechanical rather than biochemical or immunological routes. This framework is structurally independent of the connective tissue/hEDS/CCI model: it requires normal collagen but abnormal bone geometry. The two are not mutually exclusive — a patient may have both hypermobile joints and asymmetric bones, compounding the mechanical burden. The framework’s primary weakness is the near-total absence of direct ME/CFS evidence (Skeletal Asymmetry Hypothesis: Evidence Base is Near-Zero); its primary strength is that it generates specific, low-cost, falsifiable predictions — EOS full-spine radiography + COMPASS-31 in n=200 ME/CFS patients, n-of-1 shoe lift autonomic trials, and prospective post-infectious cohorts with pre-existing asymmetry measurement — using technology that already exists. If any of these predictions is confirmed, a structural, non-pharmacological treatment pathway (postural correction, targeted physiotherapy, orthotics) would become testable for a subgroup currently unrecognized by any existing ME/CFS model. (Origin: synthesis — Gerlier 2026-07-21 core hypothesis, refined by brainstorm.)

12 The Fibromyalgia Overlap

ME/CFS and fibromyalgia share extensive clinical and biological overlap. A systematic review and meta-analysis of 21 publications found that the two diagnoses co-occur in 47.3% of cases (95% CI: 45.97–48.63), though high heterogeneity ($I^2=98%$) reflects varying diagnostic criteria across studies . The 2016 Wolfe fibromyalgia criteria, which incorporate fatigue, unrefreshed sleep, and cognitive dysfunction, likely inflate this figure by absorbing core ME/CFS features into the FM definition. Community-based estimates place co-occurrence closer to 22–35%.

At the cerebrospinal fluid level, the two conditions may be even more closely related than clinical phenotypes suggest. Mass spectrometry analysis of 2,083 CSF proteins found no significant differences between ME/CFS patients with and without comorbid fibromyalgia, with machine learning achieving only AUC 0.67—barely above chance . The authors concluded that the two conditions “fall along a common illness spectrum,” challenging the hypothesis that they have distinct central nervous system pathophysiology.

Neuroinflammation provides further convergence. Multi-site PET imaging with [¹¹C]PBR28 (a TSPO radioligand marking activated microglia and astrocytes) revealed significant microglial activation in the thalamus, somatosensory cortex, and prefrontal cortex of fibromyalgia patients compared to controls . Parallel findings of microglial activation in ME/CFS suggest a shared neuroinflammatory substrate.

Autoantibody signatures add a further layer of convergence. In a study comparing ME/CFS patients with and without comorbid FM and healthy controls, 54.5% of the ME/CFS+FM group and 45.5% of the ME/CFS-only group showed abnormal GABA receptor immunoreactivity, versus 0% of controls. Altered $\beta$-endorphin immunoreactivity correlated with fatigue severity in both groups, suggesting dysregulated natural autoimmunity rather than classical autoimmune pathology .

Post-Exertional Malaise: Shared but Asymmetric.

A three-level meta-analysis of 45 effects from 15 studies confirmed that both ME/CFS and fibromyalgia patients show small-to-moderate increases in pain severity following exercise, establishing pain-related PEM as a shared feature . However, PEM after physical exertion remains the cardinal distinguishing criterion: present in $>$ 90% of ME/CFS patients but absent or inconsistent in pure fibromyalgia. When FM is comorbid, PEM severity is amplified (scores 79.63 vs. 71.18, $p < 0.001$).

Key Differences.

Despite the overlap, clinically important distinctions remain. A recent comprehensive review summarized the distinguishing features: the core symptom in ME/CFS is profound fatigue with obligatory PEM after physical exertion, while FM is defined by widespread chronic pain. NK cell reduction is consistent and pronounced in ME/CFS but variable in FM. Mitochondrial impairment is severe in ME/CFS versus moderate in FM. GPCR autoantibody evidence is stronger in ME/CFS, though emerging in FM.

Open Question: Same Disease, Different Locks?

What if ME/CFS and fibromyalgia represent the same underlying pathophysiology with different predominant locks?

ME/CFS-predominant: Stronger metabolic/immune locks, less central sensitization
Fibromyalgia-predominant: Stronger central sensitization lock, less metabolic involvement
Mixed: Both lock types active

The CSF proteome data and shared microglial activation patterns support this spectrum model. If confirmed, this would explain why they so often co-occur and why treatments for one sometimes help the other. The key clinical implication: clinicians seeing fibromyalgia should systematically screen for PEM, as the high co-occurrence (47%) means nearly half of FM patients may also meet ME/CFS criteria .

Speculation: Shared Polygenic Architecture, Different Regulatory Mechanisms

(Certainty: 0.55; 0.50→0.55: reinforced by Kerrebijn 2026 fibromyalgia GWAS, which confirmed exclusive brain-enriched heritability in fibromyalgia and reported fibromyalgia risk regions overlapping with ME/CFS at OLFM4* and RABGAP1L — convergent evidence, since the direct fibromyalgia × ME/CFS genetic correlation is untested; see Section Three-Line Genetic Convergence on Neuronal Biology for the cross-reference.)* DecodeME + Steen 2026 twin-sibling study + Kendler 2023 family scores collectively support shared polygenic architecture across ME/CFS, fibromyalgia, and IBS — the genetic correlation of ME/CFS with IBS ($r_g = 0.75$, strongest in DecodeME) and with fibromyalgia (by twin-sibling heritability partitioning) supports a common genetic liability spectrum (Maccallini 2026) (DecodeME Consortium, Ponting, et al. 2025) (Steen et al. 2026) (Kendler et al. 2023) (Kerrebijn et al. 2026). However, Hirsch 2025 comparative GWAS found that while the same genetic loci are involved, the specific DNA signals at shared loci differ between conditions — suggesting same genes, different regulatory mechanisms (Hirsch et al. 2025).

This pattern — shared polygenic architecture with divergent causal variants at shared loci — is most parsimoniously explained by tissue-specific regulatory variation: the same glutamatergic and neuronal genes are under different cis-regulatory control in different tissues or developmental windows, producing condition-specific expression patterns from shared genetic variation (Wirth and Scheibenbogen 2026). This is distinct from pleiotropy (one variant affecting multiple traits) and instead represents allelic heterogeneity within shared biological pathways.

Evidence base. Steen 2026 (twin-sibling, shared heritability CFS/FM/IBS); Kendler 2023 (family scores, cross-condition genetic overlap); Hirsch 2025 (comparative GWAS, different signal at same loci); DecodeME ($r_g = 0.75$ with IBS) (DecodeME Consortium, Ponting, et al. 2025).

Falsifiable prediction. Fine-mapping of the top DecodeME loci in ME/CFS vs. IBS vs. FM will show non-overlapping credible SNP sets (Jaccard similarity < 0.3) despite the same genes being implicated, consistent with distinct causal variants at shared loci.

The largest fibromyalgia GWAS to date (Kerrebijn et al. 2026, Nature Medicine, 2.56 million individuals) now provides the strongest direct genome-wide evidence relevant to this shared-architecture model: it identifies 26 fibromyalgia risk loci and finds heritability exclusively enriched in brain tissues and neural cell types — a finding consistent with a central-nervous-system role and parallel to the brain-enriched ME/CFS findings of DecodeME and the Maccallini meta-GWAS (Kerrebijn et al. 2026). The shared risk regions (OLFM4, RABGAP1L; DCC in the earlier overlap) and the pervasive genetic correlation of fibromyalgia with IBS (rg > 0.7) and depression are analyzed in detail in Section Three-Line Genetic Convergence on Neuronal Biology of Chapter Genetic and Epigenetic Factors.

Hypothesis: PEM Status as the Cross-Disease Determinant of Exercise Response

Certainty: 0.50. The exercise response spectrum across ME/CFS, fibromyalgia, Gulf War illness, and Long COVID may be predicted by a single variable: PEM status. Depression represents one pole—exercise has established biological mechanisms (BDNF upregulation, neuroplasticity) and genuine benefit survives blinding correction . Fibromyalgia occupies an intermediate position: graded aerobic exercise can improve pain in PEM-negative FM patients, while PEM-positive FM patients deteriorate. ME/CFS and PEM-positive Long COVID represent the other pole—exercise causes harm (Section [NO SECTION] of Chapter [NO CHAPTER]). Gulf War illness shares the exercise intolerance profile with contested psychosomatic framing and identical blinding vulnerabilities. Not yet replicated—the PEM-stratified analysis across conditions has not been performed.

The clinical implication is that PEM status, not the disease label, should determine exercise recommendations. A fibromyalgia patient with PEM should receive the same exercise cautions as an ME/CFS patient.

Testable predictions. (1) FM exercise meta-analyses stratified by PEM status should show: PEM-negative FM retains robust exercise benefit; PEM-positive FM shows null benefit identical to the ME/CFS pattern. (2) Gulf War illness exercise meta-analyses should show the same subjective-only improvement pattern, predicted by the BRANDO framework (Section [NO SECTION] of Chapter [NO CHAPTER]). (3) Cross-condition meta-regression with PEM status as moderator should show PEM as the strongest predictor of exercise harm, stronger than the condition label.

13 The Autoimmune Disease Spectrum

ME/CFS may sit on a continuum with recognized autoimmune diseases:

Sjögren’s Syndrome (Gougerot-Sjögren).

The overlap between primary Sjögren’s syndrome (pSS) and ME/CFS is both clinically significant and mechanistically informative. Fatigue affects 65–70% of pSS patients and is consistently rated as the most debilitating extraglandular symptom, predicting work disability and reduced quality of life independently of glandular involvement .

The “seronegative Sjögren’s” hypothesis has historical roots. Already in 1996, Nishikai et al. found that one-third of CFS patients presenting with sicca symptoms (dry eyes, dry mouth) fulfilled Sjögren’s diagnostic criteria while remaining seronegative—lacking the classic anti-SSA/SSB antibodies that define primary SS . This raised the possibility that a subset of ME/CFS patients harbor subclinical or atypical Sjögren’s disease that escapes conventional serological detection.

A 2023 Berlin study brought contemporary rigor to this question. Kim et al. assessed 19 pSS patients against the Canadian Consensus Criteria for ME/CFS and found that 22% (4/18) fulfilled full CCC criteria, while 32% (6/19) had severe ME/CFS-like symptoms . Crucially, PEM in pSS was primarily triggered by mental and emotional exertion rather than physical exertion, and hand grip strength recovered normally within 60 minutes after exercise—in contrast to the abnormal recovery pattern characteristic of ME/CFS. This suggests that while the fatigue phenotypes overlap, the underlying exercise physiology diverges.

The immunological connections are striking. Both pSS and ME/CFS show elevated $\beta_1$/$\beta_2$-adrenergic and M3/M4-muscarinic receptor autoantibodies. However, in pSS these autoantibodies correlate with systemic disease activity (ESSDAI, $p < 0.05$) but not with fatigue severity —whereas in ME/CFS, GPCR autoantibodies are hypothesized to mediate fatigue and autonomic dysfunction directly. The same molecular target thus appears to have different functional consequences in the two conditions.

Type I interferon dysregulation provides another convergence point. In pSS, elevated plasmacytoid dendritic cells drive a type I IFN signature that correlates with extraglandular manifestations and autoantibody production. Similar plasmacytoid dendritic cell changes have been reported in ME/CFS, though the blood↓/tissue↑ pattern documented across autoimmune diseases (SLE, RA, psoriasis) and viral infections (19/23 COVID-19 studies) suggests that “elevated” pDCs in these reports may reflect tissue-predominant expansion with concurrent blood depletion — a chemokine-driven migration pattern (CXCR3/CXCL9-11) rather than true numerical increase. Functional pDC assays (DNase1L3 production, IFN-α secretion capacity) have not been directly compared between ME/CFS and pSS. The apparent convergence on IFN-α-driven autoimmunity may be real but the cellular basis (tissue vs. blood pDCs) remains unresolved both conditions.

CSF neuroinflammation further links the two. In pSS, IL-1Ra in cerebrospinal fluid is significantly associated with fatigue severity, implicating IL-1$\beta$-mediated neuroinflammation . Parallel findings of elevated IL-1$\beta$ and neuroinflammatory markers in ME/CFS CSF suggest a shared central inflammatory driver of fatigue.

Small fiber neuropathy (SFN) may represent a unifying structural substrate across pSS, fibromyalgia, and ME/CFS. SFN prevalence reaches 49% in fibromyalgia by meta-analysis, is well-documented in pSS (presenting predominantly as a non-length-dependent pattern), and has been reported in ME/CFS cohorts at up to 34% . This shared neuropathic substrate could underlie the overlapping pain, autonomic, and sensory symptoms across all three conditions. The diagnostic implications are bidirectional. The average diagnostic delay for Sjögren’s disease is 3–6 years, and pSS patients commonly receive initial labels of fibromyalgia, CFS, or functional disorder before correct diagnosis . Conversely, 7–8% of pSS patients have no sicca symptoms at diagnosis, creating a risk that Sjögren’s is missed in ME/CFS patients who lack the classic dry eyes/mouth presentation. In any ME/CFS patient presenting with fatigue, dry eyes or mouth, joint pain, and sensory symptoms, Sjögren’s evaluation (Schirmer’s test, salivary flow, lip biopsy, SSA/SSB antibodies) is warranted—including when seronegative, given the Nishikai data .

Beyond the shared serology and small-fibre substrate, the ocular surface itself is a candidate non-invasive measurement site that Sjögren’s research has developed to a level of maturity ME/CFS work has not yet exploited. Two independent readouts are relevant: tear-film proteomics and corneal confocal microscopy (CCM). In Sjögren’s dry eye, targeted tear proteomic panels (LC-MRM) discriminate autoimmune from non-autoimmune sicca (Lépine, Robert, and Sleno 2024) (George, Kurien, and Scofield 2023), and tear inflammatory mediators—IL-6, IL-17, MMP-9, BAFF—track systemic disease activity (K. Y. Wu et al. 2024). In parallel, CCM of the corneal subbasal nerve plexus (unmyelinated C-fibres) detects small-fibre loss in Sjögren’s (Luzu et al. 2022), predicts serological activity (Y. Wang et al. 2025), and reveals corneal nerve abnormalities even in fibromyalgia-associated dry eye (Vergés et al. 2025). That CCM can image corneal small fibres in both Sjögren’s and ME/CFS is independently documented (Néstor Azcue et al. 2025) Cañadas et al. (2023), making the corneal nerve plexus, on technical grounds, a candidate common small-fibre observable spanning Sjögren’s, fibromyalgia, long COVID, and ME/CFS — though formal cross-disease normative comparisons in ME/CFS are pending and the single ME/CFS CCM dataset (Azcue group) awaits independent replication. Whether the ocular surface additionally integrates immune and small-fibre signals into a single systemic window in ME/CFS is a separate and untested claim, developed (with its caveats) below.

Speculation: The Ocular Surface as a Shared Neuro-Immune Window Across Sjögren’s and ME/CFS

Certainty: 0.25. The ocular surface may integrate two signals that are dysregulated in both primary Sjögren’s syndrome and ME/CFS: tear-film inflammatory mediators (IL-6, IL-17, MMP-9, BAFF), which in Sjögren’s track systemic immune activity (K. Y. Wu et al. 2024) (George, Kurien, and Scofield 2023), and corneal subbasal nerve density, an unmyelinated C-fiber readout of small-fiber integrity measurable by corneal confocal microscopy (Luzu et al. 2022) (Y. Wang et al. 2025). The same inflammatory cytokine family is elevated in ME/CFS serum, and CCM-detected small-fiber loss is documented in ME/CFS and post-COVID cohorts (Néstor Azcue et al. 2025) (Néstor Azcue et al. 2023) Cañadas et al. (2023). The lacrimal functional unit is parasympathetically innervated (cranial nerve VII), so ME/CFS dysautonomia could plausibly disrupt tear production downstream of the same autonomic pathology implicated elsewhere in the illness. If these strands converge, the ocular surface would provide a mostly non-invasive site reflecting both the neuro-immune and small-fiber arms of ME/CFS pathophysiology. (Tear collection is genuinely low-burden; corneal confocal microscopy, by contrast, requires a specialist fixed device and operator, so “accessible” applies unevenly across the two readouts.)

Prediction (falsifiable). In an ME/CFS cohort versus matched controls, CCM will show reduced corneal nerve fiber density (effect direction matching the Sjögren’s/fibromyalgia literature: lower CNFD in cases), AND tear IL-6/MMP-9 will be elevated in the subset with objective dry eye. Falsified if CCM shows no group difference in corneal nerve density, OR tear inflammatory markers do not differ between ME/CFS cases with sicca and controls (failure of either conjunct refutes the joint window claim).

Limitations. No study has yet measured the tear proteome or performed CCM as a primary endpoint in a dedicated ME/CFS cohort; every ME/CFS-specific claim here is analogical, extrapolated from Sjögren’s, fibromyalgia, and post-COVID data. Sicca in ME/CFS is multifactorial (anticholinergic medications, dehydration, autonomic dysfunction) and need not reflect autoimmune glandular pathology. Corneal nerve changes are non-specific across many systemic conditions.

Consequence: If confirmed, tear sampling (genuinely low-burden) together with corneal confocal microscopy (a specialist procedure, not a bedside test) could offer a largely non-invasive way to detect the small-fiber and inflammatory features of ME/CFS, and to distinguish patients who share biology with Sjögren’s from those who do not — of particular value where invasive skin-biopsy nerve testing is impractical. Until a direct ME/CFS study is done, this remains a promising but untested extrapolation. (Origin: brainstorm-adjacent literature synthesis.)

Open Question: Does the ME/CFS tear proteome resemble the Sjögren’s signature?

The tear proteome of ME/CFS patients is entirely uncharacterized. Sjögren’s tear proteomics has matured to the point of validated discriminant panels (Lépine, Robert, and Sleno 2024), and fibromyalgia—another fatigue-spectrum condition—shows ocular-surface and corneal-nerve abnormalities (Vergés et al. 2025), yet no study has applied the same LC-MRM tear panel to an ME/CFS cohort. Whether ME/CFS tears carry a Sjögren’s-like inflammatory signature (supporting a shared subclinical-autoimmune subgroup), a distinct signature, or no consistent signature is unknown. Resolving this would either strengthen or weaken the long-standing “seronegative Sjögren’s” hypothesis for a subset of ME/CFS patients.

Consequence: Answering this determines whether a cheap, non-invasive tear test could ever help identify the subset of ME/CFS patients whose illness overlaps biologically with an autoimmune disease that has approved treatments—or whether that hoped-for overlap is absent. (Origin: brainstorm-adjacent literature synthesis.)

Speculation: Functional Lacrimal Denervation — a Candidate Non-Autoimmune Route to Sicca in ME/CFS

Certainty: 0.28. (Origin: brainstorm.) The sicca of ME/CFS need not be autoimmune. Whereas Sjögren’s produces dry eye through lymphocytic destruction of the lacrimal gland, and the M3-autoantibody hypothesis (Shared M3 Muscarinic Autoantibodies Link Sjögren’s Syndrome Sicca and ME/CFS Autonomic Dysfunction) posits antibody-mediated impairment of muscarinic signaling across multiple end-organs (exocrine glands, GI smooth muscle, autonomic ganglia), a third route is functional denervation: withdrawal of parasympathetic drive to the lacrimal functional unit (pterygopalatine ganglion, cranial nerve VII) as one facet of the broad dysautonomia documented in ME/CFS (Néstor Azcue et al. 2023). This predicts preserved gland architecture (normal salivary-gland ultrasound), sicca severity that tracks other autonomic measures (heart-rate variability, pupillometry) rather than inflammatory markers, and — critically — a preserved or supersensitive secretory response to cholinergic stimulation, in contrast to the blunted response of glandular destruction.

Prediction (falsifiable). A standardised pilocarpine challenge (oral M3 agonist; Schirmer’s measured at 0/30/60/90 min) will yield a $\geq 5$ mm tear increase in the majority of ME/CFS sicca patients but in a minority of primary Sjögren’s sicca controls, and pilocarpine responsiveness will correlate with resting HRV. Falsified if ME/CFS sicca patients show salivary-gland ultrasound abnormalities matching Sjögren’s, or if the pilocarpine response is as blunted as in Sjögren’s. Two caveats bound what this test can establish. First, it distinguishes functional impairment from structural gland destruction, but does not by itself separate functional denervation from M3-antibody-mediated blockade — both can coexist and both predict a preserved response; discriminating them requires stratifying the pilocarpine response by M3-autoantibody titre (the non-autoimmune route predicts preserved response in the antibody-negative subset), which is the test of the title’s specific claim (see Shared M3 Muscarinic Autoantibodies Link Sjögren’s Syndrome Sicca and ME/CFS Autonomic Dysfunction). Second, chronic anticholinergic-medication use can itself cause glandular atrophy, so a blunted response or abnormal ultrasound must be interpreted against anticholinergic burden before it is read as evidence against denervation.

Limitations. No direct ME/CFS data exist; the mechanism is inferred from documented ME/CFS parasympathetic dysfunction plus lacrimal neuroanatomy. It overlaps with the M3-autoantibody account (Shared M3 Muscarinic Autoantibodies Link Sjögren’s Syndrome Sicca and ME/CFS Autonomic Dysfunction) — functional denervation and antibody-mediated M3 impairment are not mutually exclusive, and the pilocarpine challenge cannot distinguish them. This is a research hypothesis, not a treatment recommendation: cholinergic agonists (pilocarpine, cevimeline) are not approved for ME/CFS, carry side effects (sweating, nausea, diarrhoea, bradycardia) that may limit tolerability, and are contraindicated in uncontrolled asthma (bronchoconstriction risk), narrow-angle glaucoma, acute iritis, gastrointestinal or biliary obstruction, sick sinus syndrome, and concurrent beta-blocker use. In a population with prevalent orthostatic intolerance and autonomic instability, any provocation test with a cholinergic agonist requires cardiac monitoring. Any clinical use would require on-site clinician supervision.

Consequence: If dry eyes and dry mouth in ME/CFS come from the nervous system failing to “switch on” tear and saliva glands—rather than immune attack—then a medication that stimulates those glands (used under medical supervision with cardiac monitoring, and only where not contraindicated) might relieve the symptom, and a supervised in-clinic challenge test could tell which patients that would help. Until tested directly in ME/CFS this is a plausible but unproven mechanism, not a treatment recommendation.

Limitation: Why the Ocular-Surface Signal in ME/CFS May Be an Artefact, Not a Window

(Origin: brainstorm — self-critique of the ocular-surface hypotheses above.) The case for the ocular surface as a neuro-immune window (The Ocular Surface as a Shared Neuro-Immune Window Across Sjögren’s and ME/CFS, Does the ME/CFS tear proteome resemble the Sjögren’s signature?, Functional Lacrimal Denervation — a Candidate Non-Autoimmune Route to Sicca in ME/CFS) rests entirely on analogy: no study has measured tear cytokines, tear proteomics, or corneal confocal microscopy as a primary endpoint in a criteria-defined ME/CFS cohort (certainty of this gap: ~0.95). Several alternative explanations could account for any ocular findings without implicating ME/CFS systemic biology. (1) Medication confound (the simplest, ~0.50): ME/CFS patients are frequently prescribed anticholinergic drugs (tricyclics, antihistamines for mast-cell activation, antispasmodics, trazodone), all of which reduce tear and saliva secretion — sicca and any secondary ocular-surface inflammation may be largely iatrogenic. (2) Dry-eye confound (~0.45): dry eye disease itself elevates tear IL-6/IL-17/MMP-9 and remodels corneal nerves (reduced density, increased tortuosity) through purely local inflammation, so ocular findings may reflect comorbid dry eye rather than a systemic signal. (3) Spurious overlap (~0.40): Sjögren’s and ME/CFS may both converge on a common final pathway of ocular-surface inflammation for entirely different upstream reasons, making any tear-proteomic “similarity” a shared endotype of dry eye rather than shared disease mechanism. Additional weaknesses: the one available fatigue-spectrum ocular study (fibromyalgia; (Vergés et al. 2025)) did not control for anticholinergic medication; the Sjögren’s tear panels were trained to separate autoimmune from non-autoimmune dry eye and may misclassify autonomic dry eye; tear cytokine measurements vary with time of day, blink, collection method, and assay; corneal nerve density and tear cytokines are both age-dependent (risking spurious tear–nerve correlation in an older cohort absent shared pathophysiology); housebound patients likely have elevated screen time, which independently destabilises the tear film and alters tear cytokines; applying a multi-analyte Sjögren’s panel to an uncharacterized population without correction for the number of analytes is a fishing-expedition risk; and referral-clinic selection bias likely inflates apparent Sjögren’s–ME/CFS overlap relative to community prevalence. Finally, no single study has combined corneal confocal microscopy, tear cytokines, skin-biopsy nerve density, and autonomic testing in the same participants, so the “integrative window” claim is untested at the level it is stated. On present evidence the parsimonious position is explicit: each confound above is individually assigned a higher probability than the systemic-window interpretation it competes with (0.50/0.45/0.40 versus the window/denervation certainties of 0.30/0.28), so the ocular findings — if they exist at all in ME/CFS — are more likely explained by medication and comorbid dry eye than by a shared systemic mechanism. The speculations are retained as testable, low-certainty research hypotheses precisely because they make predictions that would discriminate them from these confounds, not because they are the leading explanation.

Consequence: Before any tear or eye test could be trusted as a marker of ME/CFS, researchers must first rule out that the findings are caused by patients’ medications or ordinary dry eye — otherwise a hopeful-looking result could mislead both patients and clinicians. The honest current status is: promising analogy, zero direct evidence.

Systemic Lupus Erythematosus.

Caveat on construct comparison: SLE fatigue is measured by FACIT-F, FSS, and VAS-fatigue; post-exertional malaise, the defining ME/CFS symptom, has never been systematically assessed in SLE (Schwarting et al. 2019) (Parodis et al. 2025). Direct ME/CFS-SLE comparisons use different fatigue instruments and may conflate distinct phenomena.
Fatigue is the most disabling symptom in two-thirds of SLE patients; one-third report severe fatigue (Mertz et al. 2020)
Neuropsychiatric lupus (NPSLE) affects >50%, producing fatigue and cognitive dysfunction through brain-reactive autoantibodies crossing a compromised blood-brain barrier (Weissman-Tsukamoto, Carroll, and Diamond 2025)
58 shared genes between SLE and ME/CFS enriched in monocytes — IL1$\beta$, CCL2, TLR2, STAT1, IFIH1 — from a bioinformatics study only; whether these genes are co-dysregulated in the same direction in both diseases remains untested (Zheng et al. 2024)
NK cell mitochondrial dysfunction in SLE (enlarged mitochondria, impaired mitophagy, defective cytotoxicity (Fluder et al. 2026)) raises the question of whether ME/CFS NK impairment similarly involves a mitochondrial component — the structural abnormality in SLE NK cells provides a testable hypothesis for ME/CFS, not a demonstrated shared mechanism
Anti-NR2 (NMDAR) antibodies in SLE correlate with fatigue independent of neuropsychiatric lupus and downregulate neuronal energy metabolism without cytotoxicity (Schwarting et al. 2019) — a cross-sectional association (n=426) with an observational treatment subgroup (belimumab n=86, non-randomized for this endpoint) that has not been tested in ME/CFS
Fatigue in SLE dissociates from inflammation: 13.6–26.2% of patients in DORIS remission still report clinically significant fatigue (majority resolve) (Parodis et al. 2025); type 2 (fatigue/pain) symptoms show no correlation with interferon signatures (Arcani et al. 2023)
Divergent finding: Exercise modulates the IFN signature and improves fatigue in SLE (Spinelli et al. 2023) — notably opposite to ME/CFS where exercise causes post-exertional malaise and symptom worsening
Speculative link: ME/CFS might involve lupus-like autoimmunity below diagnostic thresholds — UCTD (undifferentiated connective tissue disease) patients with autoimmune fatigue not fulfilling SLE/SSc criteria may overlap with ME/CFS populations; whether UCTD fatigue shares ME/CFS pathophysiology or reflects a distinct connective tissue disease spectrum is unknown (Rubio and Kyttaris 2023) (Andreoli and Tincani 2017)
Speculative link: Anti-NR2-mediated neuronal metabolic suppression in SLE provides a potential analogy for how autoantibodies targeting neural receptors might produce fatigue without systemic inflammation — direct testing of anti-NR2 antibodies in ME/CFS is required before extrapolation

Multiple Sclerosis.

Fatigue is major symptom
Cognitive dysfunction similar
Both may involve HERV reactivation
Speculative link: ME/CFS might be “diffuse MS” without discrete lesions, or MS-related autoimmunity affecting different neural targets

Autoimmune Encephalitis.

Can present with fatigue, cognitive dysfunction, psychiatric symptoms
Antibodies against neural proteins
Often triggered by infection
Speculative link: ME/CFS might be low-grade autoimmune encephalitis affecting widespread but subtle neural dysfunction

PANS/PANDAS.

Acute-onset OCD, tics, behavioral regression temporally linked to infection (most commonly streptococcal) (Swedo et al. 1998) (K. Chang et al. 2015)
Anti-neuronal antibodies cross-reacting with basal ganglia targets via proposed molecular mimicry: GAS N-acetyl-beta-D-glucosamine antibodies may recognize neuronal lysoganglioside, activating CaMKII in dopaminergic neurons — this pathway is characterized in the Cunningham laboratory but independent replication has been inconsistent; the mechanism represents the leading model, not a settled fact (Kirvan et al. 2003) (Kirvan et al. 2006) (Cesaroni et al. 2026) (La Bella et al. 2023)
Treatment response is variable and interpretation contested: open-label series report 83–89% response rates but the single double-blind RCT (n=35) found IVIG not significantly superior to placebo; the post-hoc GAS-positive subgroup analysis (smaller sample within the already-underpowered trial) is hypothesis-generating, not confirmatory (Latimer et al. 2015) (Kovacevic, Grant, and Swedo 2015) (K. A. Williams et al. 2016) (Sigra, Hesselmark, and Bejerot 2018). Immunotherapy for PANDAS/PANS is not FDA-approved for any neuropsychiatric indication; costs range $ 30,000–$ 100,000 per course; serious adverse events include aseptic meningitis, thromboembolism, anaphylaxis (IVIG), and hypotension/catheter-related infection (TPE). Treatment should only be pursued through specialist PANDAS/PANS centers following established protocols (antibiotics→NSAIDs/steroids→specialist referral→IVIG→TPE as last resort)
PANDAS/PANS sits in a nosological middle ground: more severe than the proposed ME/CFS low-grade autoimmune encephalitis (basal ganglia circuit dysfunction producing florid OCD/tics vs diffuse neuroinflammation), but less destructive than anti-NMDAR encephalitis (functional CaMKII activation without overt neuronal loss) (La Bella et al. 2023) (Vreeland et al. 2023)
The PANDAS controversy — whether it “really exists” as a distinct entity — is structurally identical to the historical debate over ME/CFS as biological vs psychosomatic (Chiarello et al. 2017). The resolution, emerging from 25 years of research, is that the question is mis-framed: some acute-onset paediatric neuropsychiatric presentations are immune-mediated, some are not; the diagnostic label “PANDAS” is a syndromic bin that collapses both populations (Masterson et al. 2025)
Speculative link for ME/CFS: If PANDAS represents basal ganglia-directed autoimmune neuropsychiatric disease intermediate between full autoimmune encephalitis and the proposed ME/CFS “low-grade” autoimmune state, then a broader neuroimmune encephalopathy spectrum connects these conditions: different circuits targeted (basal ganglia → OCD/tics in PANDAS; brainstem/thalamus → autonomic dysfunction and fatigue in ME/CFS; widespread cortical → cognitive dysfunction in Long COVID) but a shared disease class — post-infectious, autoantibody-mediated, circuit-specific, potentially immunomodulable. The specific presentation depends on which circuits are hit. The existence of one well-characterized member of this class (PANDAS) raises the prior probability that others (ME/CFS-acquired ADHD/ASD-like features, some cases of regressive autism) belong to the same biological category (Dominique Endres et al. 2022) (Whiteley et al. 2021)
The maternal immune activation literature provides a prenatal variant of this spectrum: ~23% of mothers of children with ASD carry anti-fetal brain autoantibodies targeting specific neuronal proteins — these antibodies are associated with increased ASD risk, but the causal pathway from maternal antibodies to the child’s diagnosis has not been proven (Braunschweig et al. 2013) (Croen et al. 2008) (Meltzer and Van de Water 2017). While the “attributable” fraction depends on assumptions about antibody pathogenicity that remain untested in prospective cohorts, these findings nonetheless define a distinct immune-associated ASD subgroup with potential therapeutic implications. These research findings have no current clinical application — no prenatal screening for anti-fetal brain antibodies is recommended, and immunomodulation during pregnancy for this indication is untested and potentially harmful. The postnatal equivalent — infection-triggered anti-neuronal antibodies producing autism-like symptoms in previously neurotypical children — is structurally analogous to PANDAS and to the proposed ME/CFS Architecture B, differing in which developmental window the immune insult occurs and which circuits it targets
A critical evidence gap: no prospective study has screened children presenting with “regressive autism” for PANS/PANDAS diagnostic criteria. The question of what proportion of “autism” diagnoses represent undiagnosed immune-mediated neuropsychiatric syndromes — the question the user’s clinical observation isolates — remains empirically unanswered. The null finding that anti-neuronal antibodies do not differ between regressive and classic ASD (Aslan et al. 2021) suggests immune mechanisms do not neatly align with regression phenotype, but ~20% of ASD patients carry elevated anti-neuronal antibodies regardless of regression status (Aslan et al. 2021) — consistent with an immune-mediated subset that does not respect syndromic boundaries
Speculative link: ME/CFS might sit elsewhere on this same neuroimmune encephalopathy spectrum — not as basal ganglia OCD/tics (PANDAS) or florid encephalitis (anti-NMDAR) but as a low-grade, diffuse, multi-circuit neuroimmune dysfunction producing fatigue, cognitive impairment, and autonomic instability rather than discrete neuropsychiatric episodes. The ME/CFS predilection for basal ganglia involvement (diminished globus pallidus activation during reward processing, metabolite abnormalities) and dopamine receptor autoantibodies suggests partial overlap with PANDAS targets, but the full clinical picture diverges — consistent with a broader, less anatomically focused autoimmune process (Dominique Endres et al. 2022)

Type 1 Narcolepsy.

Type 1 narcolepsy is now confirmed as autoimmune: CD4+ T-cells selectively destroy hypocretin (orexin)-producing neurons in the lateral hypothalamus, with CD4+ T-cell density 11-fold higher than other T-cell types in the affected region . The immune signatures persist decades after initial neuronal destruction. The condition is frequently post-infectious in origin (notably H1N1, Streptococcus) and strongly associated with HLA-DQB1*06:02.

The ME/CFS parallels are notable:

Orexin system involvement: Reduced orexin-A levels have been reported in ME/CFS across multiple studies , and cytokine-driven PGE₂ suppression of orexin neurons is proposed as a mechanism for inflammation-induced fatigue and sleep pressure (see Hypothesis Maladaptive Sickness Behavior Program, Chapter [NO CHAPTER])
Post-infectious autoimmune trigger: Both conditions arise after infection via autoimmune attack on CNS targets—orexin neurons in narcolepsy, potentially broader neural and autonomic targets in ME/CFS
Comorbid presentation: A case of ME/CFS + POTS + narcolepsy with compound heterozygous MTHFR mutations suggests shared vulnerability in some patients
Speculative link: If ME/CFS involves autoimmune suppression of orexin neuron function (via receptor autoantibodies or cytokine-mediated inhibition) rather than outright neuronal destruction, this would explain the overlapping sleep-wake dysregulation without the complete hypocretin deficiency seen in narcolepsy. The narcolepsy finding also raises the question of whether early immunotherapy could prevent orexin system damage in ME/CFS patients showing narcolepsy-pattern polysomnography

Open Question: Subclinical Autoimmunity?

What if ME/CFS represents autoimmune disease below conventional detection thresholds? The autoantibodies might:

Target functional receptors/channels rather than structural proteins
Be present at low titers that affect function without triggering standard assays
Target intracellular or unusual epitopes not covered by standard panels

This “subclinical autoimmunity” hypothesis would explain why immunomodulation helps some patients while standard autoimmune panels are negative. The PANS/PANDAS precedent is instructive: anti-neuronal antibodies in PANDAS were initially undetectable by standard clinical assays and required functional characterization (CaMKII activation) to establish pathogenicity — see PANS/PANDAS as a Plausibility Precedent for Acquired Neuropsychiatric Phenotypes. If the same pattern holds in ME/CFS, detection requires mechanism-informed assays, not standard panels.

Speculation: The Neuroimmune Encephalopathy Spectrum (NES): A Testable Unified Disease Class

Certainty: 0.30 (bumped from 0.25: PANS/PANDAS — integrated in Phase 6 — provides a molecularly characterized anchor member for the spectrum, reducing the purely speculative nature of the disease-class claim). The comparative analysis above — PANDAS, autoimmune encephalitis, Multiple Sclerosis, maternal immune activation in ASD — suggests that a single disease class, the Neuroimmune Encephalopathy Spectrum (NES), may connect what syndromic classification separates. NES is defined by three shared properties: (1) an immune trigger (infection, autoantibody generation, or maternal antibody transfer), (2) anti-neuronal antibodies targeting specific CNS circuits, and (3) a clinical phenotype determined by which circuits are hit, not by which pathogen triggered the response. Under NES:

Basal ganglia-directed autoantibodies (anti-D1/D2, anti-lysoganglioside, anti-tubulin) produce OCD, tics, and behavioral regression — the PANDAS/PANS end of the spectrum (Kirvan et al. 2003) (Kirvan et al. 2006)
Brainstem/thalamic-directed autoantibodies (anti-adrenergic, anti-muscarinic) produce fatigue, autonomic instability, and orthostatic intolerance — the ME/CFS-POTS end (La Bella et al. 2023)
Diffuse cortical/limbic-directed autoantibodies (anti-NMDAR, anti-GABA) produce psychosis, catatonia, and cognitive collapse — autoimmune encephalitis
Prenatal maternal antibody transfer targeting fetal brain proteins produces an ASD-like developmental phenotype — the immune-mediated autism subset (Braunschweig et al. 2013) (Meltzer and Van de Water 2017)

The NES framework makes two falsifiable predictions that syndromic classification does not: (a) antibody profiles should cluster by circuit-target pattern, not by diagnostic label — some “autism” patients will have basal-ganglia antibody profiles indistinguishable from PANDAS, and some “ME/CFS” patients will have them too; (b) immunotherapy response should be predicted by antibody circuit-target (basal ganglia → IVIG/TPE response; autonomic → variable; diffuse cortical → steroid-responsive) rather than by DSM/ICD diagnosis. The Connery 2018 vs Plioplys 1998 IVIG-in-ASD contrast (Connery et al. 2018) (Plioplys 1998) — dramatic response in autoantibody-stratified ASD, null in unselected ASD — is a proof-of-concept for prediction (b). No study has directly tested prediction (a) because no study has run the same autoantibody panel simultaneously across PANDAS, ME/CFS, ASD, and healthy cohorts. The failure to do so — and the institutional barriers that make multi-diagnosis antibody profiling rare — may be the single largest obstacle to resolving the nosological structure of post-infectious neuropsychiatric illness.

Falsification condition: A multi-disease study ($n ≥ 50$ per group: PANDAS/PANS, post-infectious ME/CFS with neuropsychiatric features, regressive ASD, and healthy controls) running the same anti-neuronal antibody panel (≥ 10 specificities: anti-basal ganglia, anti-D1/D2, anti-lysoganglioside, anti-CaMKII, anti-NMDAR, anti-CASPR2, anti-GABA_A, anti-adrenergic, anti-muscarinic, anti-tubulin) across all groups will test two conjoint predictions. Prediction (A): Unsupervised clustering of antibody profiles will produce clusters that are cross-labeled by diagnostic category — i.e., the adjusted Rand index (ARI) between antibody-profile clusters and diagnostic labels will be < 0.30, indicating that antibody-target pattern carries information orthogonal to clinical diagnosis. Prediction (B): In a platform clinical trial assigning immunotherapy (IVIG or TPE) to participants from all diagnostic groups, the interaction term (antibody-profile-cluster × treatment) will be significant (p < 0.05) while the main effect of diagnostic label on treatment response will not (p > 0.20). Falsified if ARI between antibody-profile clusters and diagnostic labels is ≥ 0.50 (diagnostic labels predict antibody profiles better than chance) AND the diagnostic-label × treatment interaction is significant (p < 0.05) while the antibody-profile × treatment interaction is not (p > 0.20). Failure of either prediction alone (e.g. A true but B false) partially refutes NES; failure of both refutes it entirely. The NES framework is not structurally unfalsifiable — it makes specific claims about clustering structure and treatment-response prediction — but no study of this design has ever been conducted.

Consequence: If NES is correct, diagnostic labels (PANDAS, ME/CFS, autism) are circuit-identity surrogates of limited precision — the same biology produces all three depending on which circuits the autoantibodies target.

Severity applicability: Unknown — the NES framework is a disease-class proposal, not a clinical staging model. Individual conditions within NES have their own severity spectra (e.g. mild PANS vs progressive PANS (Masterson et al. 2025); mild ME/CFS vs very severe).

Limitations: Entirely speculative. No multi-disease anti-neuronal antibody profiling study exists. The specific circuit-target → phenotype mapping is inferred from existing single-disease antibody studies and has not been tested prospectively. The existence of ME/CFS cases without detectable autoantibodies, and PANS cases without immunotherapy response, confirms that NES covers a subset within each diagnostic category, not the entire condition. The PANS/PANDAS evidence base consists predominantly of narrative reviews and case series ($n = 12$–$ 35$ for treatment studies) from specialist referral centers; the single RCT was underpowered and equivocal (K. A. Williams et al. 2016). The PANDAS literature has a population-weight discount (paediatric, tertiary referral, rare condition) that should temper generalization to adult ME/CFS.

Synthesis: Post-Infectious Immune-Mediated Neuropsychiatric Syndromes as a Trans-Disease Biological Category

Certainty: 0.35. Five independent evidence lines, drawn from different disease domains and experimental systems, converge on a single proposition: post-infectious immune-mediated neuropsychiatric syndromes are a recognized biological category that transcends syndromic diagnostic boundaries.

The convergent lines are:

PANS/PANDAS: The best-characterized post-infectious autoimmune neuropsychiatric pipeline — GAS → anti-neuronal IgG → CaMKII activation → basal ganglia dysfunction → OCD/tics — with documented immunotherapy responsiveness in a subset PANS/PANDAS as a Plausibility Precedent for Acquired Neuropsychiatric Phenotypes
Long COVID neuropsychiatric manifestations: ADHD-like executive dysfunction, depressive symptoms, and cognitive impairment arising post-SARS-CoV-2 in a phenotypically consistent pattern (N. Azcue et al. 2022)
Anti-NMDAR encephalitis: The most severe end of the spectrum — florid psychosis, catatonia, cognitive collapse — established as treatable autoimmune disease (Dalmau et al. 2008)
Maternal anti-fetal brain autoantibodies in ASD: ~23% of ASD cases are attributable to maternal antibodies targeting specific fetal neuronal proteins, defining an immune-mechanism ASD subgroup that is structurally identical to the postnatal post-infectious pattern, differing only in the developmental window of the immune insult (Braunschweig et al. 2013) (Croen et al. 2008)
ME/CFS neuropsychiatric overlap: Elevated ADHD, ASD, and depression in ME/CFS, with a plausible immune-mediated subset identified by circuit-target autoantibody profiles The Neuroimmune Encephalopathy Spectrum (NES): A Testable Unified Disease Class

These five lines originate from separate literatures (paediatric neuropsychiatry, post-viral neurology, autoimmune encephalitis, developmental immunology, chronic fatigue neuroimmunology) that rarely cite one another. Their convergence on the same disease logic — infection or antibody transfer → anti-neuronal immune response → circuit-specific neuropsychiatric phenotype → potentially immunomodulable — is unlikely to be coincidental.

The category is clinically consequential: if post-infectious immune-mediated neuropsychiatric syndromes form a trans-disease biological category, then (a) etiological investigations should profile anti-neuronal antibodies by circuit target, not by diagnostic label; (b) immunotherapy trials should enroll across diagnostic categories and stratify by antibody circuit-target pattern; and (c) patients currently assigned to DSM/ICD bins that share an underlying immune mechanism should be identifiable and treatable regardless of which syndromic label they carry. The Connery 2018 vs. Plioplys 1998 IVIG-in-ASD contrast — dramatic response in autoantibody-stratified ASD, null in unselected ASD — is the proof-of-concept that this approach works when the right subset is selected, and that it fails when syndromic criteria alone determine enrollment (Connery et al. 2018) (Plioplys 1998).

Limitations: The convergence is inferential, not experimental. No study has run the same anti-neuronal antibody panel simultaneously across PANS, Long COVID, ME/CFS, ASD, and healthy cohorts. The five lines are individually plausible but collectively underdetermined — convergence does not raise collective certainty above the weakest link. Until a multi-disease antibody profiling study exists, this synthesis is a research framework, not a finding.

Prediction: In a multi-disease study profiling ≥10 anti-neuronal antibody specificities across ≥200 participants from ≥4 diagnostic categories, unsupervised clustering of antibody profiles will produce clusters that are cross-labeled by diagnostic category (ARI < 0.30), confirming that the biological category is circuit-target-defined, not syndrome-defined. Falsified if antibody profiles cluster cleanly by diagnostic label (ARI ≥ 0.50).

Severity applicability: Unknown — each constituent line has its own severity spectrum; the synthesis is a disease-class proposition, not a clinical staging model.

14 The Mitochondrial Disease Connection

Primary mitochondrial diseases share features with ME/CFS:

Overlapping Features.

Exercise intolerance (defining in both)
Post-exertional symptoms (delayed recovery in both)
Cognitive dysfunction (both)
Multi-system involvement (both)

Differences.

Primary mitochondrial disease: genetic mutations, progressive
ME/CFS: acquired, stable or fluctuating

Open Question: Acquired Mitochondriopathy?

What if ME/CFS represents an “acquired mitochondrial disease” where the genetic code is intact but epigenetic changes or post-translational modifications create mitochondria that function as if mutated? The mitochondria might be:

Epigenetically silencing key respiratory chain components
Maintaining a “fission” state inappropriate for energy demands
Preferentially undergoing mitophagy, reducing functional mitochondrial mass

This would explain the mitochondrial dysfunction without genetic mutations.

Speculation: MtDNA Haplogroup Symptom Modulation Without Onset Predisposition

(Certainty: 0.45 — directly evidenced by two-cohort mtDNA sequencing study; symptom associations replicated in direction but not in individual markers.)

Three independent whole-genome mitochondrial sequencing studies consistently find no association between inherited mtDNA variants and ME/CFS onset . The null finding for onset predisposition is robust. What the evidence does support is a distinct claim: haplogroup background modulates symptom cluster expression after ME/CFS is established. Billing-Ross et al. found that haplogroups J, U, and H each associated with distinct symptom profiles in 193 ME/CFS cases — haplogroup J correlating with protection against joint pain metrics, haplogroup U with less severe bloating, and haplogroup H with increased “feeling dead after exercise.” This effect operates on symptom expression rather than susceptibility, and has not been independently replicated.

This distinction matters clinically: mtDNA haplogroup testing could theoretically help predict symptom severity profiles in established ME/CFS patients — which symptom clusters are most likely to be prominent — without offering any information about onset risk. If replicated, haplogroup-based symptom profiling would provide a genetic framework for personalizing symptom management (e.g., predicting which patients are at higher risk for severe post-exertional crashes based on haplogroup H status).

Falsifiable prediction: In an independent ME/CFS cohort of ≥ 300 patients, haplogroup H will correlate with higher post-exertional malaise scores (DSQ-PEM subscale) relative to haplogroup J and U patients matched for disease duration and baseline severity. If no haplogroup-symptom association is found, the symptom-modulation claim does not extend beyond the single Billing-Ross cohort.

Limitations: The symptom associations in are unreplicated; the study was predominantly European-descent, limiting generalizability; haplogroup effects on phenotype could be confounded by nuclear genetic background linked to haplogroup; no causal mechanism linking haplogroup to specific symptom pathways has been proposed. Not yet replicated independently.

Speculation: Haplogroup U as Shared ADHD–ME/CFS Mitochondrial Modifier

(Certainty: 0.20 — rests on a single cross-study haplogroup U overlap; no direct combined-cohort study exists.)

Haplogroup U appears as a modifier in both ME/CFS and ADHD independently. In a meta-analysis of 2,076 ADHD cases across three European-American cohorts, haplogroup U (and K) was protective against ADHD diagnosis . In the Billing-Ross ME/CFS cohort, haplogroup U was associated with attenuated bloating symptoms . These findings emerge from different populations, different study designs, and different endpoints — one measuring disease risk, the other symptom severity — yet they converge on the same haplogroup as a protective modifier across both conditions.

This single overlap point raises a speculative but testable hypothesis: haplogroup U confers a mitochondrial bioenergetic configuration that is protective or moderating across neurodevelopmental and post-infectious fatigue conditions, with the specific phenotypic expression depending on which system is most stressed. Under this shared-substrate model, low mitochondrial reserve (non-U haplogroup background) might manifest as dopaminergic prefrontal dysfunction (ADHD phenotype) when the energetically expensive prefrontal cortex is the primary bottleneck, and as post-exertional metabolic collapse (ME/CFS phenotype) when whole-body energy metabolism is overwhelmed by a trigger.

ADHD itself shows direct evidence of mitochondrial bioenergetic impairment: cybrid cell lines derived from ADHD patient platelets demonstrate lower cellular respiration, reduced Complex V (ATPase) activity, diminished mitochondrial membrane potential, and elevated oxidative stress compared to controls — and these defects are transferable via the patient’s own mitochondria, not just their nuclear environment . Multiple additional studies document elevated mtDNA copy number (≈ 1.3$\\times$ controls), SNP associations, and haplogroup effects across ADHD populations, though sample sizes and methodology are heterogeneous .

Falsifiable prediction: In a cohort of ME/CFS patients with ADHD comorbidity (estimated 20–40% based on general prevalence), haplogroup distribution will differ significantly from ME/CFS patients without ADHD comorbidity, with haplogroup U depleted in the non-ADHD ME/CFS group relative to the ADHD-comorbid group. This is falsified if haplogroup distribution is identical across the two ME/CFS subgroups.

Limitations: The cross-study haplogroup U overlap is a single data point and could reflect chance, population stratification, or study-specific confounders. No study has measured mtDNA haplogroup, ADHD comorbidity, and ME/CFS diagnosis in a single cohort. The cybrid evidence for ADHD mitochondrial dysfunction uses n=3 ADHD patients — mechanistically informative but statistically fragile . The causal direction between ADHD dopamine dysregulation and mitochondrial dysfunction is unresolved: mitochondrial dysfunction could drive dopamine deficiency, or chronic dopaminergic dysregulation could secondarily stress mitochondria. Not yet replicated.

Open Question: Maternal Inheritance Clustering in ME/CFS: Untested

The largest family study of ME/CFS heritability demonstrated first-degree relative risk of 2.70 and a signal persisting to third-degree relatives — arguing against shared-environment explanations and supporting a heritable component. However, the study does not separate maternal from paternal inheritance. This distinction is diagnostically critical for the mitochondrial predisposition hypothesis: mitochondrial DNA is maternally inherited almost exclusively, so purely maternal clustering (mother→child) would constitute evidence for mtDNA-mediated predisposition, while paternal clustering would rule it out.

No published ME/CFS family study has reported the maternal-to-paternal inheritance ratio. The practical barriers are modest: pedigree analysis of ME/CFS family registries (e.g., the Utah Population Database used in , or the UK Biobank data with ME/CFS self-report) could determine whether affected relatives cluster disproportionately on the maternal lineage. This analysis is feasible with existing data.

If maternal clustering exceeds chance (expected ratio ~50% maternal in simple autosomal predisposition), this would motivate whole-mitochondrial-genome sequencing studies powered for onset rather than severity. If maternal and paternal clustering are equivalent, the mitochondrial predisposition hypothesis loses its primary mechanistic rationale, and nuclear genetic or epigenetic transmission models become comparatively more plausible.

Open Question: Why Do ME/CFS Patients Carry Fewer Deleterious mtDNA Variants?

Venter et al. found an unexpected pattern in two independent ME/CFS cohorts (UK and South Africa): patients carry fewer mildly deleterious mtDNA variants than healthy controls (UK: 55% vs 27% without such variants, p = 0.0008; SA: 56% vs 41%, p = 0.03). This is the reverse of what a “damaged mitochondria = predisposition” model would predict. The replicated finding has no accepted explanation.

Three non-exclusive interpretations warrant investigation: (a) Selection pressure — individuals with more deleterious mtDNA variants may respond to the same trigger by developing a different (perhaps more acute and resolved) illness rather than ME/CFS; (b) Threshold inversion — it is not the presence of deleterious variants but the absence of protective rare variants that matters; the “fewer deleterious” group may lack specific haplogroup-associated protective SNPs; (c) Nuclear-mitochondrial mismatch — the relevant predisposition variable may be nuclear-encoded mitochondrial gene variants (NDUFAF2, UCP2, FBXL4, CCPG1) that interact with mtDNA background, making the mtDNA-only sequencing approach incomplete.

This finding challenges simple “damaged mtDNA” framing and may point toward a more complex haplogroup-by-nuclear-gene interaction as the true predisposition substrate.

Speculation: Heteroplasmy Sex Asymmetry as Contributor to Female Predominance

(Certainty: 0.30 — mechanistically plausible; female sex predominance in ME/CFS is well-established; direct heteroplasmy measurement by sex has not been done in ME/CFS.)

ME/CFS affects women at 3–4 times the rate of men. Mitochondrial bottleneck during oogenesis fixes heteroplasmy ratios in offspring. Female-biased X-linked nuclear-encoded mitochondrial proteins (e.g., NDUFA1, NDUFB11) interact with mtDNA-encoded OXPHOS subunits. A heteroplasmy burden tolerated in males — where hemizygosity at X-linked mitochondrial assembly factors is the baseline — may exceed the symptom-modulation threshold in females, where X-inactivation mosaicism creates cell-by-cell variability in mitochondrial assembly efficiency. This could amplify the impact of even modest mtDNA heteroplasmy on whole-tissue energy output.

Falsifiable prediction: In a mixed-sex ME/CFS cohort with deep-coverage mtDNA sequencing (>1000× depth, enabling detection of heteroplasmy at 0.5% allele frequency), female patients will show higher heteroplasmy variance at OXPHOS-relevant sites (Complex I/IV subunit genes) than male patients matched for disease severity. Limitations: Existing ME/CFS mtDNA studies (Billing-Ross 2016, Venter 2019) found overall heteroplasmy levels low and comparable between cases and controls; sex-stratified analysis has not been reported. The X-linked nuclear-mitochondrial interaction mechanism is established in principle but not measured in ME/CFS tissue. Not yet replicated.

Speculation: Nuclear-Mitochondrial Mismatch as Predisposition Substrate

(Certainty: 0.30 — mechanistically plausible; onset-null mtDNA results from all published studies may obscure a real mismatch signal; direct haplogroup × nuclear-genome interaction in ME/CFS untested.)

The consistent null finding for mtDNA onset-predisposition ( ) analyzes mtDNA alone. Yet OXPHOS function depends on the coordinated expression of ~90 nuclear-encoded genes and 13 mtDNA-encoded proteins; evolutionary co-adaptation between haplogroup and nuclear background optimizes this coordination. Recent maternal-lineage admixture — an individual carrying a non-European haplogroup alongside a European nuclear-encoded mitochondrial proteome, or vice versa — may produce reduced OXPHOS efficiency without any individual variant being deleterious. The predisposition signal could be an interaction term not captured by mtDNA-only sequencing.

Falsifiable prediction: ME/CFS prevalence will be elevated in populations with measurably high recent maternal-paternal continental admixture (quantifiable by comparing mitochondrial haplogroup continental origin with autosomal ancestry estimates), compared to populations with stable haplogroup-nuclear pairing, after adjusting for socioeconomic and healthcare access confounders. Limitations: This prediction requires whole-genome data linked to ME/CFS diagnosis in admixed populations — a dataset not currently available. The population genetics methodology is established; its application to ME/CFS is novel and untested. Not yet replicated.

Open Question: Trio Sequencing in ME/CFS Families: The Definitive Maternal-Paternal Test

The strongest direct test of mitochondrial predisposition in ME/CFS is trio sequencing: sequencing the affected proband plus both parents in 100 families, allowing formal partition of heritability between mtDNA (mother-only transmission) and nuclear (biparental) contributions. The Albright 2011 pedigree study establishes the overall family relative risk (2.70 first-degree) but does not report the maternal vs. paternal breakdown. Trio sequencing in 100 families would generate both haplogroup determination (from proband + mother) and maternal-vs-paternal clustering data simultaneously.

This is feasible with existing research infrastructure: GWAS study protocols and saliva-based DNA collection kits require no clinic visit. Family-based recruitment through established ME/CFS patient registries (ME/CFS Biobank UK, Solve ME) could yield sufficient family trios within 12–24 months.

If maternal-side relatives of probands show significantly higher ME/CFS prevalence than paternal-side relatives at p < 0.05, this would constitute the first positive evidence for maternal transmission as a meaningful component of ME/CFS predisposition. If maternal and paternal clustering are equivalent, the mtDNA predisposition hypothesis requires reassessment in favor of nuclear genetic or epigenetic transmission.

Open Question: Long-Read mtDNA Sequencing to Retest the Onset-Null Finding

All three ME/CFS mtDNA studies finding null onset-predisposition (Billing-Ross 2016 , Venter 2019 , and the review by Dibble 2020 ) relied on short-read next-generation sequencing. Short-read approaches underresolve low-frequency heteroplasmy (typically insensitive below 5% allele frequency) and may miss structural variants (large deletions, rearrangements) that accumulate in aged tissues. Nanopore and PacBio duplex sequencing now achieve heteroplasmy detection down to ~0.1% allele frequency with strand-discrimination error correction.

Replication of the Billing-Ross or Venter study design with long-read mtDNA sequencing in n ≥ 200 ME/CFS patients would definitively test whether the null finding reflects genuine absence of mtDNA predisposition or a technical limitation of short-read sensitivity. If long-read sequencing identifies at least one structural variant or low-heteroplasmy site (≥1% allele frequency) enriched in cases vs. controls at p < 0.05 (corrected for multiple testing), the null finding would require revision.

Speculation: Pathogenic mtDNA Carrier Relatives as a Natural Experiment

(Certainty: 0.35 — mechanistically coherent; natural experiment avoids many confounds; requires accessing mitochondrial disease registries and ME/CFS diagnosis data jointly.)

Asymptomatic carriers of pathogenic mtDNA mutations — relatives of individuals with MELAS, MERRF, or Leigh syndrome who carry the mutation at low heteroplasmy and remain clinically unaffected — live with measurably reduced mitochondrial reserve throughout their lives. Under the reserve-threshold model (Section [NO SPECULATION] in Chapter [NO CHAPTER]), these carriers should show elevated ME/CFS rates after viral triggers compared to their non-carrier siblings, because their reserve floor is already lowered by the pathogenic variant before any stressor is applied.

Falsifiable prediction: Surveying maternal relatives in mitochondrial disease registries (MitoCohort, MITOMAP-linked registries, UMDF data) will reveal post-viral ME/CFS-like illness incidence ≥3× the age/sex-adjusted general population rate in individuals confirmed as low-heteroplasmy pathogenic variant carriers. If carrier relatives show ME/CFS rates indistinguishable from the general population, the reserve-threshold model cannot explain predisposition via germline mitochondrial vulnerability. Limitations: Mitochondrial disease registries are not designed for ME/CFS outcome ascertainment; retrospective survey methodology has ascertainment bias; “ME/CFS-like illness” without formal diagnostic validation is a weak endpoint. Not yet replicated.

Open Question: WURS Score as a Zero-Cost ME/CFS Risk Stratifier

The Dopaminergic Demand Crossover hypothesis (Section [NO SPECULATION] in Chapter [NO CHAPTER]) predicts that childhood ADHD history reflects the same low mitochondrial reserve substrate that confers ME/CFS susceptibility. If correct, adult ADHD or subclinical childhood attention symptoms — retrospectively quantifiable by the Wender Utah Rating Scale (WURS-25, a validated retrospective childhood ADHD symptom scale) — could serve as a clinical marker enriching at-risk cohorts for baseline studies or prevention trials without requiring any genetic testing.

The WURS-25 is a self-administered, 25-item questionnaire with established psychometric properties and a validated threshold of score >36 for probable childhood ADHD . It is zero-cost, zero-blood-draw, and can be administered remotely. In a prospective post-viral cohort, WURS score at enrollment could be tested as a predictor of ME/CFS conversion at 6 and 12 months.

Testable prediction: WURS-25 score >36 in pre-illness adults will predict ME/CFS development after acute viral illness with OR ≥ 2.0, in a prospective post-viral cohort of ≥ 300 participants with 12-month follow-up. If the OR is not distinguishable from 1.0, childhood ADHD history does not serve as a risk enrichment tool, and the crossover hypothesis loses its clinical utility even if mechanistically valid.

15 The Psychiatric Overlap—Reframed

ME/CFS has historically been conflated with depression and anxiety. A mechanistic reframing:

Shared Biology, Not Shared Psychology.

Both ME/CFS and depression involve inflammatory cytokines
Both involve kynurenine pathway abnormalities
Both involve HPA axis dysregulation
Both involve neurotransmitter changes

The Cytokine Theory of Depression.

Depression may be, in part, an inflammatory brain state
Cytokines cause “sickness behavior” that resembles depression
Speculative link: ME/CFS and inflammatory depression might be the same phenomenon with different tissue distributions or lock combinations

Speculation: Depression: Shared Glutamatergic Genes, Divergent Circuit Expression

(Certainty: 0.50.) The genetic correlation between ME/CFS and depression ($r_g = 0.60$, $p < 0.00001$) reflects shared glutamatergic synaptic genes, not shared psychiatric etiology (DecodeME Consortium, Ponting, et al. 2025) (Maccallini 2026). DecodeME found NO shared causal variants between ME/CFS and depression — glutamatergic genes are shared risk factors, but circuit expression diverges: prefrontal-limbic circuits in depression versus cortico-cerebellar and brainstem circuits in ME/CFS (Wirth and Scheibenbogen 2026). This circuit-level divergence explains how the same genetic variants can produce clinically distinct conditions, resolved by Maccallini 2026 cell-type resolution showing cerebellar and subcortical enrichment in ME/CFS that is absent in depression GWAS (Maccallini 2026).

The depression correlation is a signal of shared synaptic biology, not shared disease mechanism. This has direct clinical implications: antidepressants targeting prefrontal serotonin/norepinephrine pathways may relieve depressive symptoms in comorbid ME/CFS patients without addressing the underlying glutamatergic-cerebellar dysfunction driving ME/CFS fatigue and PEM (Hirsch et al. 2025).

Evidence base. DecodeME (rg=0.60, no shared causal variants); Maccallini 2026 (cell-type resolution, cerebellar vs. prefrontal divergence); Hirsch 2025 (comparative GWAS confirming circuit-specific enrichment patterns); Peng 2026 (single-cell hippocampal atlas showing molecular disruption of the hippocampal trisynaptic circuit in depression, consistent with the limbic-circuit locus of the divergence) (Peng et al. 2026).

Open Question: Neuroimmune Spectrum Disorders?

What if ME/CFS, inflammatory depression, “brain fog” conditions, and some anxiety disorders all represent points on a “neuroimmune spectrum”? The common feature would be immune activation affecting brain function through:

Direct cytokine effects on neurons
Microglial activation
Kynurenine pathway shifts
Blood-brain barrier dysfunction

Different presentations might reflect which brain regions are most affected, not fundamentally different diseases.

Speculation: The Inflammatory-Mediated Neurodegeneration Template: Bidirectional Mental-Illness–Inflammation Link

(Certainty: 0.40; raw 0.64–0.60, dementia/mental-illness populations → cross-disease discount; reduced from 0.45 on adversarial review because the two supporting meta-analyses share a single research group testing the same hypothesis, and the inference is a double-hop from dementia to ME/CFS. Origin: literature.) A meta-analysis of 41 studies testing the “inflammatory-mediated neurodegeneration” hypothesis in dementia (Kuring et al. 2026) found that Alzheimer’s disease is associated with elevated IL-17A, IL-1$\alpha$, IL-10, G-CSF, GM-CSF, and IL-3 independent of comorbid psychiatric and physical inflammatory conditions. A companion meta-analysis found depression, anxiety, and PTSD are each associated with inflammation independent of such comorbidities (Kuring et al. 2023). IL-17A is elevated in both Alzheimer’s disease and depression, and G-CSF in Alzheimer’s, depression, and PTSD — a shared inflammatory pathway candidate linking mental illness to CNS pathology.

This template is the reverse direction of the “neuroinflammation as common upstream driver” model above Shared Neuroinflammatory Cascade as Generator of All Three Phenotypes. That model treats depression as a secondary manifestation of ME/CFS neuroinflammation; the inflammatory-mediated-neurodegeneration template posits that pre-existing mental-illness-associated inflammation can itself drive CNS degeneration. The two are not contradictory — they imply a bidirectional inflammation–mental-state loop, with the direction of causality unresolved in cross-sectional data. Critically for ME/CFS, the dementia experience shows that excluding comorbid inflammatory conditions removes the classically reported IL-6 and TNF-$\alpha$ elevations (Kuring et al. 2026) — a methodological caution that comorbidity masking may inflate or distort inflammatory-marker claims in any chronic condition, including ME/CFS (Chapter Immune System Dysfunction).

Falsifiable prediction: Within ME/CFS, a depression/ anxiety–comorbid subgroup—selected to exclude other inflammatory conditions—will show a distinct inflammatory-marker profile (higher IL-17A, G-CSF) that correlates with the severity of depressive symptoms, and prospective longitudinal sampling will show the emotional/mental symptom onset preceding, not following, that marker rise. Conversely, if mental symptoms in ME/CFS are uniformly secondary to neuroinflammation, the marker rise will follow mental-symptom onset in all patients.

Critique: structurally falsifiable; the critical uncertainty is causality direction, which the existing cross-sectional evidence cannot resolve. Two further caveats temper the cross-disease parallel. First, the dementia and mental-illness meta-analyses come from the same research group testing the same hypothesis, so the template rests on a single group’s meta-analytic approach and its cross-sectional primary studies — it is a hypothesis-generating parallel, not independently replicated causal evidence. Second, the ME/CFS inflammatory signal itself is contested (only TGF-$\beta$ is consistently elevated across CFS studies (Blundell et al. 2015); the primary TSPO-PET neuroinflammation finding awaits replication), so inflammation may be less central in ME/CFS than in dementia — the template adds plausibility to, but does not prove, an inflammatory role in ME/CFS.

Consequence: none currently — this is a cross-disease mechanistic parallel whose specific dementia markers do not match the ME/CFS profile; its practical value for ME/CFS is limited to a caution about how comorbidities can distort inflammatory-marker interpretation and a recurring framing that mental and inflammatory states may drive each other rather than one always causing the other. ] {#spec-bidirectional-mood-inflammation-template}

Speculation: Encephalitis–Dementia as the Acute-Exposure Precedent for ME/CFS Neuroinflammation→Neurodegeneration

(Certainty: 0.42; raw 0.70, general-population electronic-health-record cohort → cross-disease population weight 0.75 × an additional 0.80 analogical-distance factor, because the acute, high-intensity florid encephalitides driving the autoimmune finding are farther from chronic low-grade ME/CFS neuroinflammation than the sibling dementia-metabolic parallels; within the crude ±0.10 certainty granularity this is calibration to the analogical gap, not merely the source’s internal strength. Origin: literature.) A large US electronic-health-record cohort (TriNetX, ~129 M patients, 72 organizations) found that surviving encephalitis — an acute, direct inflammation of the brain parenchyma — predicts long-term dementia, with a composite risk ratio of 2.11 in adults over 60 and 5.16 in adults 40–60 over a ten-year follow-up; the association remained robust when restricted to specific dementia codes (Alzheimer’s, vascular, specified) in adults 40–60 and >60, whereas the much higher relative-risk figures in younger cohorts (20–40 and under 20) were no longer significant once restricted to those codes, indicating the younger-age signal largely reflects post-encephalitic neurological sequelae (G31) rather than delayed Alzheimer-type degeneration (Aditi et al. 2026). An independent UK primary-care cohort replicated elevated post-encephalitis cognitive and dementia sequelae (with risk highest early in follow-up, consistent with a component of direct post-infectious cognitive sequelae) (Granerod et al. 2017). The strongest dementia risk was conferred by non-infectious/post-infectious inflammatory (autoimmune) etiologies (risk ratio 3.93) rather than bacterial infection (1.35, 0.97–1.87, not significant) (Aditi et al. 2026). The mechanism by which CNS inflammation drives neurodegeneration is established in Alzheimer’s disease — microglial/astroglial activation (Heneka et al. 2025) with a causal NLRP3-inflammasome role in both tau (Ising et al. 2019) and amyloid (Heneka et al. 2013) pathology — and ME/CFS is grouped with chronic neuroimmune-dysfunction disorders in this framing (Cohen et al. 2024). The cross-disease claim is that this same acute-CNS-inflammation → long-term-neurodegeneration axis, demonstrated in encephalitis and mechanistically grounded in Alzheimer’s disease, is the template by which sustained ME/CFS neuroinflammation (TSPO-PET microglial activation (Nakatomi et al. 2014); serum exosomes activating microglia to release IL-1$\beta$, a finding reported after exercise rather than at rest (Tsilioni, Natelson, and Theoharides 2022)) could, if it persists, contribute to progressive cognitive decline — tempered throughout by the G31-sequelae-vs-AD endpoint confound above and by the absence of any ME/CFS dementia cohort. All of this is indirect for ME/CFS: there is no ME/CFS dementia cohort, and no demonstration that ME/CFS neuroinflammation is degenerative rather than reversible.

Falsifiable prediction: If sustained ME/CFS neuroinflammation drives neurodegeneration through the encephalitis/Alzheimer axis, then longitudinal ME/CFS cohorts (n$ >= $ 200, ≥ 5-yr follow-up) will show (i) elevated baseline plasma NfL/GFAP or CSF neuroinflammation markers that predict worsening objective cognitive performance over time, and (ii) a dose–response between TSPO-PET signal (or an equivalent glial marker) and the rate of cognitive decline. Falsified if ME/CFS patients show stable cognition over ≥ 5 years despite elevated neuroinflammation markers, or if no neuroinflammation marker predicts any cognitive trajectory — which would argue ME/CFS neuroinflammation is non-degenerative (reversible/regional), unlike the encephalitic or Alzheimer patterns.

Critique: structurally falsifiable. The transfer rests on analogy, not ME/CFS data: the demonstrable link is encephalitis (or Alzheimer neuroinflammation) to dementia, not ME/CFS to dementia. Two caveats temper the parallel (compare bidirectional mood inflammation template). First, the ME/CFS neuroinflammation signal itself is contested — the single published TSPO-PET study in ME/CFS found no significant difference vs controls (Raijmakers et al. 2021) (Neuroinflammation Measurement Challenges) — so the premise that ME/CFS has sustained neuroinflammation is not yet established. Second, etiology specificity in the encephalitis data warns against over-generalizing: the effect is strongest for non-infectious/post-infectious inflammatory (autoimmune) causes and absent for bacterial, and animal work ((Javonillo et al. 2026), preprint) cautions that acute viral neuroinflammation does not uniformly accelerate amyloid — so inflammation quality, not mere presence, would matter.

Consequence: If confirmed, ME/CFS neuroinflammation would shift from a symptom correlate toward a progressive-pathology driver, justifying earlier anti-neuroinflammatory intervention and long-term cognitive monitoring — but at present this is an analogy across diseases; it remains unknown whether ME/CFS neuroinflammation is degenerative, and (Severity applicability: unknown — general-population dementia/encephalitis cohort, not a ME/CFS population and not severity-stratified.) ] {#spec-encephalitis-dementia-precedent}

Speculation: Shared Th17/Myeloid-Axis Signal Between ME/CFS and Inflammatory CNS Conditions

(Certainty: 0.30. Origin: brainstorm.) Against the overall divergence between the ME/CFS inflammatory profile (TNF-$\alpha$, IL-2, IL-4, TGF-$\beta$, CRP elevated (Strawbridge et al. 2019)) and the Alzheimer profile (IL-17A, IL-1$\alpha$, IL-10, G-CSF, GM-CSF, IL-3 elevated (Kuring et al. 2026)), one specific axis converges: myeloid/Th17 hematopoiesis. G-CSF and GM-CSF are severity-correlated inflammatory markers in ME/CFS (Montoya cohort (Montoya et al. 2017)) and are among the markers elevated in Alzheimer’s disease independent of comorbidity (Kuring et al. 2026); IL-17F (a Th17 cytokine) is part of early-disease ME/CFS signatures (Hornig 2015) while IL-17A is elevated in Alzheimer’s and depression (Kuring et al. 2023). This raises the possibility of a shared set of G-CSF/GM-CSF/IL-17 markers across chronic inflammatory-CNS conditions — grouped here by their observed co-elevation, not by a demonstrated shared regulatory node (G-CSF and GM-CSF are myeloid/hematopoietic growth factors, while IL-17A/IL-17F are T-cell-derived, so any single underlying axis would require direct evidence that does not yet exist).

Falsifiable prediction: A comorbidity-screened ME/CFS cohort (excluding other inflammatory conditions) will show elevated G-CSF and GM-CSF relative to controls with a Th17-correlated profile (raised IL-17F or IL-17A), and these will track fatigue/cognitive severity. Falsified if comorbidity-screened ME/CFS cohorts show neither elevated G-CSF/GM-CSF nor a Th17-correlated signal.

Critique: structurally falsifiable; the convergence is suggestive but the marker panels differ across studies, and none have been run on comorbidity-screened ME/CFS cohorts.

Consequence: if confirmed, this would give ME/CFS researchers a specific, measurable immune axis (rather than vague “inflammation”) to test as a biomarker and as a stratification tool — but it is currently an untested inference from cross-disease panels. ] {#spec-th17-myeloid-shared-signal}

Synthesis: Inflammation as Cross-Disease Driver: Template, Direction, and a Methodological Caution

Taken together, the dementia and mental-illness meta-analyses (Kuring et al. 2026) (Kuring et al. 2023), the CFS inflammatory-protein meta-analysis (Strawbridge et al. 2019), and the encephalitis–dementia cohort finding (Aditi et al. 2026) yield three distinct take-aways relevant to ME/CFS on separate evidentiary grounds. First, low-grade inflammation can be dissociated from psychiatric comorbidity — and disease-inherent markers survive that exclusion — supporting the inflammatory component of ME/CFS as real rather than purely conferred by coexisting mood or inflammatory disorders (bidirectional mood inflammation template; Comorbidity Masking of Cytokine Signals — A Cross-Disease Caution); the encephalitis–dementia cohort strengthens the causal side of this from cross-sectional correlation to an exposure-defined longitudinal prediction that acute CNS inflammation precedes and predicts later neurodegeneration (encephalitis dementia precedent). Second, the causal direction between mental state and inflammation is not resolved and may run in both directions; the dementia template (mental-illness inflammation → CNS pathology) and the neuroinflammation-cascade model (Shared Neuroinflammatory Cascade as Generator of All Three Phenotypes) are two poles of one unresolved bidirectional loop rather than competing explanations. Third, a specific measurable set of G-CSF/GM-CSF/IL-17 markers appears co-elevated across chronic inflammatory-CNS conditions, including an overlap with severity-correlated ME/CFS markers, and is worth testing as a distinct biomarker set rather than as nonspecific “inflammation” (th17 myeloid shared signal). None of these claims is yet confirmed in comorbidity-screened ME/CFS cohorts, and the specific dementia markers do not transfer directly to ME/CFS; the value is a template, a methodological warning, and a specific testable marker set, not a proven ME/CFS mechanism.

Consequence: together these cross-disease parallels suggest that a shared, measurable marker set — not vague inflammation — may be a productive ME/CFS biomarker and stratification target, and that mental and inflammatory states in ME/CFS should be considered coupled rather than one always causing the other. ]

16 Architecture C Cross-Disease Bridges

The metabolic reserve framework generates predictions about ME/CFS connections to diseases not traditionally associated with it. These speculative bridges suggest testable hypotheses and potential shared mechanisms. These bridges are illustrative analogies exploring where the metabolic reserve metaphor generates novel predictions — they are not claims that Architecture C explains these diseases. Energy metabolism is involved in most disease processes; the question is whether the specific reserve-threshold framing adds predictive value beyond “mitochondria are important.”

Speculation: ADHD and ME/CFS as Same-Root Etiology — Inflammation-Driven Energy Failure

Certainty: 0.20. ADHD and ME/CFS (along with Alzheimer’s, Parkinson’s, and other neuroinflammatory conditions) may share the same root cause: chronic inflammation driving mitochondrial dysfunction and energy failure, with phenotypic differences determined by tissue compartment affected and temporal profile. The causal chain is: inflammation → cytokine-mediated metabolic suppression (IDO pathway, BH4 oxidation, Complex I/II inhibition) → mitochondrial ATP deficit → CNS energy failure (ADHD phenotype when compartmentalised) or systemic energy failure (ME/CFS phenotype when generalised). Inflammation is the primary driver; mitochondrial dysfunction is the mechanistic intermediate; symptom expression is determined by which tissues cross their energy threshold first.

Shared mechanisms supporting a unified inflammatory etiology include: (a) elevated pro-inflammatory cytokines (IL-6, TNF-alpha) documented independently in both ADHD and ME/CFS (Quadt et al. 2024) (Dunn et al. 2019); (b) co-localised dopaminergic deficit and microglial activation in ADHD (Yokokura et al. 2021), paralleling neuroinflammation patterns in ME/CFS; (c) inflammatory diversion of BH4 from catecholamine synthesis toward neopterin production — the same BH4 bottleneck in both conditions (Fanet et al. 2021) (G. E. Williams et al. 2025); (d) shared prefrontal cerebral hypoperfusion (Berthier et al. 2025); (e) convergent stimulant pharmacology — both conditions respond to dopamine/norepinephrine reuptake inhibition (Blockmans et al. 2006) (Eckey et al. 2025); (f) overlapping mtDNA haplogroup effects: haplogroup U is protective in both ADHD (X. Chang et al. 2020) and symptom-modifying in ME/CFS; (g) mitochondrial OXPHOS impairment documented independently (Verma et al. 2016) (Almutairi et al. 2024). The ALSPAC finding that ADHD traits predict 2x chronic fatigue risk with IL-6 mediation (Quadt et al. 2024) directly supports inflammation as the shared driver preceding both phenotypes. The same inflammation-mitochondria-energy cascade is implicated in Alzheimer’s and Parkinson’s, suggesting a broad spectrum of inflammation-driven brain energy failure where the specific diagnosis depends on which neural circuits are most affected. Key prediction: PBMC spare respiratory capacity will form a gradient across controls > ADHD-only > ME/CFS-only > ADHD+ME/CFS, correlating with serum IL-6 and TNF-alpha levels. Falsified if ADHD-only patients show normal inflammatory markers AND normal mitochondrial function. Not yet replicated.

Open Question: Does SARS-CoV-2 Infection Drive New-Onset AD(H)D, or Reveal Preexisting Neurodivergence?

Certainty: 0.40. A translational framework proposes that Long Covid and AD(H)D are linked through convergent neuroimmune dysfunction — frontal-striatal-hippocampal circuit impairment, neuroimmune dysregulation of catecholamine (dopamine/noradrenaline) systems, tryptophan-kynurenine overactivation, and mitochondrial bioenergetic defects (Spanoghe et al. 2026). Within this framework, a subset of Long Covid patients may develop a distinct, infection-triggered, neuroimmune-mediated dopaminergic/noradrenergic dysfunction — an “acquired” executive/cognitive syndrome phenotypically resembling AD(H)D (Spanoghe et al. 2026). The causal claim that SARS-CoV-2 infection drives new AD(H)D is genuinely unresolved and currently points in both directions:

Supporting: Post-COVID ADHD symptom onset/worsening is reported across peer communities (Spanoghe et al. 2026); stimulant prescribing rose after COVID (Koonce and Martin 2024) (Gimbach et al. 2024); AD(H)D is a risk factor for Long Covid (Merzon et al. 2022); and COVID may correlate with ADHD diagnosis (Ferrara et al. 2023).
Against: A 20-year national cohort found no independent effect of COVID-19 on AD(H)D diagnosis or treatment rates (Shkalim Zemer et al. 2024). The post-COVID rise in ADHD medication is also consistent with catch-up diagnosis or symptom overlap rather than infection-driven new-onset disease (Gimbach et al. 2024).

Resolution path: The two evidence streams differ in outcome measured (population diagnosis/treatment rates versus stimulant prescribing and symptom report) and in population, so they are not directly contradictory in method; but they reach opposite conclusions about infection-driven causality. Because the certainty gap between the strongest null ((Shkalim Zemer et al. 2024)) and the positive epidemiological associations is within measurement uncertainty (≤0.15), this is presented as an unresolved open question rather than weighted toward either side. A decisive test would be a prospective post-infectious cohort measuring validated AD(H)D symptom scales, neuropsychological executive function, and an AD(H)D biomarker profile (e.g., dopaminergic or immune-metabolic measures) at enrollment and longitudinally, rather than relying on retrospective diagnosis or prescription records.

Limitations: The anchor is a viewpoint (hypothesis-generating synthesis, no primary data); the population-relevance discount applies because much of the support is in Long Covid rather than ME/CFS directly. The framework’s own counterpoint (Zemer2024) is high-certainty and null. Severity applicability: unknown — studies were not stratified by ME/CFS severity.

Consequence: This determines whether a patient’s new attention and concentration difficulties after a viral illness are treated as newly acquired brain dysfunction requiring investigation, or as the unmasking of preexisting neurodivergence that was already present — which changes clinical interpretation, expectations, and management even though it does not change the need to address the symptoms themselves.

Speculation: Shared Neuroimmune Substrate Linking Long Covid, AD(H)D, and ME/CFS Executive Dysfunction

Certainty: 0.35. Spanoghe et al. propose a convergence of four neuroimmune mechanisms across Long Covid and AD(H)D: (1) frontal-striatal-hippocampal circuit dysfunction, (2) neuroimmune dysregulation of dopamine/noradrenaline, (3) tryptophan-kynurenine overactivation, and (4) mitochondrial bioenergetic defects (Spanoghe et al. 2026). This framework is directly relevant to ME/CFS because ME/CFS shares the same post-infectious executive/cognitive phenotype and much of the same underlying physiology already documented in this chapter — mitochondrial dysfunction (Verma et al. 2016), kynurenine-pathway activation, and inflammation-driven energy failure ADHD and ME/CFS as Same-Root Etiology — Inflammation-Driven Energy Failure.

The most defensible arm of the framework for ME/CFS is the kynurenine-to-monoamine link: kynurenine-pathway overactivation correlates with post-acute COVID cognitive impairment in humans (Cysique et al. 2023), and kynurenic acid reduces striatal dopamine at nanomolar concentrations in animal models (Rassoulpour et al. 2005) — a biochemical bridge connecting neuroimmune activation to the dopamine deficit implicated in both ADHD and, speculatively, ME/CFS executive dysfunction. The kynurenine pathway also intersects mitochondrial function (Tanaka et al. 2022), and mitochondrial complex gene suppression accompanies COVID cognitive decline (Xu et al. 2025). Dopamine itself has immunomodulatory roles, providing a bidirectional route between monoaminergic and immune systems (Sarkar et al. 2010) (Feng and Lu 2021).

The framework’s therapeutic proposal is a precision-immunopsychiatry approach: stratifying patients by immune and neurocognitive endophenotype rather than broad syndrome, and trialling targeted immunomodulation or tailored stimulant therapy in biomarker-selected subgroups (Spanoghe et al. 2026). Clinical reports describe partial symptomatic benefit in Long Covid patients from AD(H)D-targeted medications — methylphenidate, guanfacine, low-dose lithium, and (lis)dexamfetamine — with preexisting autonomic dysregulation and post-exertional malaise as limiting factors (Krishnan et al. 2022) (Fesharaki-Zadeh, Lowe, and Arnsten 2023) (Spanoghe et al. 2026). These are clinical observations only; they are not dosing guidance and not a treatment recommendation for ME/CFS.

Falsifiable prediction: If kynurenine-driven dopamine suppression is a shared substrate, then in a cohort of ME/CFS patients with executive dysfunction, plasma kynurenine/tryptophan (KYN/TRP) ratio will correlate positively with an objective executive-function deficit and negatively with a dopaminergic-function proxy (e.g., in vivo or neuropsychological measures), and this correlation will persist after controlling for fatigue severity. Falsified if KYN/TRP does not track executive dysfunction independent of overall fatigue.

Limitations: The anchor is a viewpoint without primary data; the framework’s convergence claim is a synthesis rather than a single tested model. Kynurenic-acid dopamine suppression is animal data (translation gap). Off-label pharmacotherapy evidence is observational and partially contradictory in its safety profile. Not yet replicated as a unified framework.

Severity applicability: unknown — the framework discusses Long Covid and AD(H)D without stratifying by ME/CFS severity; the executive-dysfunction and off-label-treatment content most plausibly concerns mild-to-moderate patients who can tolerate medication, but this is not stated in the source.

Consequence: This gives researchers a testable, mechanism-specific target (kynurenine-to-dopamine signaling) rather than only the broad “inflammation causes brain fog” claim, and it gives clinicians a caution: any trial of AD(H)D-type medication for cognitive symptoms must weigh preexisting autonomic instability and PEM against partial symptomatic benefit, and is not currently a validated ME/CFS treatment.

Clinical Finding

Guanfacine + N-acetylcysteine for post-infectious cognitive deficits: the combination as a distinct clinical observation. Separate from the single-agent AD(H)D medications listed above, the fixed combination of low-dose guanfacine (α2A-adrenoceptor agonist) and N-acetylcysteine (Nrf2 antioxidant) is reported as an open-label intervention for cognitive deficits in Long COVID (Fesharaki-Zadeh, Lowe, and Arnsten 2023) and in traumatic brain injury (Khasnavis et al. 2024) (Fesharaki-Zadeh et al. 2025). The combination rationale is that guanfacine restores prefrontal α2A noradrenergic signaling while NAC supports glutathione/Nrf2 redox balance against neuroinflammatory oxidative stress (Cherneva et al. 2025), addressing two non-overlapping bottlenecks in post-infectious cognitive deficits. This is distinct from the guanfacine+NAC single-agent pharmacology treated elsewhere; here the two are co-administered as a deliberate combination. Evidence type: open-label case series (no control group, small n). There are no ME/CFS-specific data, and no direct α2A-adrenoceptor → Nrf2 molecular interaction has been demonstrated (0 PubMed hits). The combination’s cognitive benefit, where reported, is partial and unblinded. Not a treatment recommendation for ME/CFS.

Falsifiable prediction: In post-infectious cohorts with cognitive deficits and elevated oxidative-stress markers, guanfacine+NAC improves objective executive function more than either agent alone, with the added benefit tracking baseline oxidative burden. Falsified if combination benefit equals monotherapy or is oxidative-burden-independent. No ME/CFS trial exists to date.

Consequence: This documents a real-world, tolerability-favorable combination already used off-label for post-infectious cognitive deficits that maps onto two mechanisms implicated in ME/CFS brain fog (prefrontal noradrenergic deficit and oxidative stress). It is hypothesis-generating: a mechanism to test in ME/CFS, not a current treatment. (Severity applicability: unknown; reported populations were Long COVID and TBI, not stratified by ME/CFS severity.)

Safety context (not a dosing recommendation): Guanfacine is a centrally acting α2A agonist with dose-related hypotension, sedation, and bradyarrhythmia risk (disease-interaction warnings: hypotension, bradyarrhythmia) and carries a substantial drug-interaction load; the ME/CFS-relevant cautions — additive hypotension with other sympatholytics and the midodrine-prerequisite pattern in sympathetically dependent patients — are documented in ch34 Diurnal Response Window as Circadian Pharmacodiagnostic Probe. No human pregnancy data exist for guanfacine (animal fetal-survival reduction at maternal-toxic doses; some authorities do not recommend use); breastfeeding effects unknown. N-acetylcysteine is generally well tolerated. No validated monitoring protocol or stopping criteria exist for this combination in ME/CFS.

Speculation: Mitochondrial-Kynurenine NAD⁺ Sink: A Unifying Executive-Dysfunction Mechanism

Certainty: 0.40. The kynurenine and mitochondrial arms of the LC-AD(H)D framework may converge on a single bioenergetic bottleneck. Kynurenine-pathway overactivation diverts tryptophan toward the kynurenine-NAD⁺ catabolic arm (Cysique et al. 2023), while mitochondrial complex suppression in COVID cognitive decline already limits energy production (Xu et al. 2025). The kynurenine pathway intersects mitochondrial function directly (Tanaka et al. 2022). If chronic kynurenine activation draws down the NAD⁺ pool that mitochondria require for ATP production, then the prefrontal cortex — the most energetically expensive brain region — would be the first to show an ATP deficit, producing the executive/inattentive phenotype common to Long Covid, AD(H)D, and ME/CFS. This unifies mechanisms (3) and (4) of the Spanoghe framework into a single molecular mechanism rather than two parallel ones (Spanoghe et al. 2026).

Falsifiable prediction: In ME/CFS patients with executive dysfunction, prefrontal ATP/phosphocreatine (measured by MR spectroscopy) will correlate inversely with kynurenine-pathway load (e.g., plasma KYN/TRP ratio) and with executive-deficit severity, independent of general fatigue. Falsified if prefrontal bioenergetic markers do not track kynurenine-pathway load after controlling for fatigue.

Limitations: This is a synthesis of three separate literatures (kynurenine, mitochondrial, prefrontal-energetics); no study has measured prefrontal NAD⁺/ATP and kynurenine load in the same ME/CFS or Long Covid cohort. The mitochondrial-kynurenine connection is documented but not at the point of prefrontal NAD⁺ depletion. Not yet replicated.

Severity applicability: unknown — the executive-dysfunction phenotype most plausibly concerns mild-to-moderate patients, but the source framework does not stratify by ME/CFS severity.

Consequence: If confirmed, this gives a single, druggable target (kynurenine-to-NAD⁺ balance) behind the otherwise separate kynurenine and mitochondrial hypotheses, and suggests that shifting the kynurenine pathway away from the NAD⁺-consuming arm might preserve prefrontal energetics — a mechanism to test, not a current treatment.

Speculation: Parkinson’s as Terminal Reserve Depletion in the Dopaminergic System

Certainty: 0.15. Parkinson’s manifests when >60% of dopaminergic neurons in the substantia nigra are lost — a reserve concept. Pre-existing conditions reducing dopaminergic reserve (including ADHD lifelong dopaminergic inefficiency) might show either earlier PD onset (cumulative stress) or protective compensation (lifelong upregulation). Caffeine’s protective effect on PD risk aligns with the reserve model: it reduces adenosine-mediated dopaminergic inhibition, effectively increasing functional $R_\text{headroom}$. Testable with existing PD registries cross-referenced with neurodevelopmental history. Entirely speculative. Not yet replicated.

Speculation: Compounded Glymphatic Failure in Neurodivergent ME/CFS

Certainty: 0.15. Neurodivergent individuals have chronic sleep disruption (ADHD delayed phase, ASD fragmented sleep) leading to impaired glymphatic clearance. If ME/CFS further degrades sleep quality, neurodivergent ME/CFS patients face compounded glymphatic failure — potentially accelerating amyloid and tau accumulation. Predicts faster cognitive decline on longitudinal testing and lower CSF amyloid-beta 42 in neurodivergent ME/CFS vs non-neurodivergent ME/CFS patients. Multiple speculative steps; each link has evidence but the chain is untested. Not yet replicated.

Speculation: Bipolar State-Switching as Oscillation Around Metabolic Reserve Threshold

Certainty: 0.20. Bipolar disorder involves state-dependent mitochondrial function — mania shows hypermetabolism, depression shows hypometabolism. In the reserve model, this maps to oscillation around $R_\text{crit}$ rather than sustained depletion below it. Prediction: bipolar patients should develop ME/CFS more frequently during or after depressive episodes (when $R_\text{headroom}$ is minimized) than during euthymia. Infection during a depressive phase should carry higher ME/CFS risk than infection during mania. Novel framing; no data. Not yet replicated.

Speculation: Fibromyalgia Central Sensitization as Self-Amplifying Metabolic Cost

Certainty: 0.35. Central sensitization in fibromyalgia (amplified pain signaling) is metabolically expensive — maintaining heightened neural excitability requires more ATP via Na⁺/K⁺-ATPase demand. In Architecture C, fibromyalgia is both a consequence of reduced metabolic reserve AND a cause of further reduction: the sensitization itself consumes energy. This creates a self-amplifying loop distinct from but synergistic with the PEM damage cycle. The memantine rationale ([NO SPECULATION]) targets this demand-side amplification. Individual links documented; self-amplifying loop framing novel. Not yet replicated.

Speculation: Depression as BH4-Mediated Reserve Reduction and Consequence

Certainty: 0.25. If BH4 is the convergent bottleneck ([NO SPECULATION]), depression (serotonin depletion) could be both cause and consequence of reserve reduction. BH4 depletion reduces serotonin, causing depression; depression disrupts sleep, impairing mitochondrial repair; further reserve reduction worsens BH4 depletion. This bidirectional loop explains why depressed ME/CFS patients respond poorly to SSRIs alone — SSRIs address serotonin reuptake without fixing the upstream BH4 deficit. Prediction: depressed ME/CFS patients with low BH4 should respond better to BH4 supplementation + SSRI than SSRI alone. BH4 levels should predict SSRI response magnitude. Mechanistically logical; no clinical data. Not yet replicated.

17 The Cancer Cachexia Connection

Cancer-associated cachexia shares surprising features with ME/CFS:

Shared Features.

Profound fatigue out of proportion to activity
Muscle wasting/weakness
Metabolic abnormalities
Inflammatory cytokine elevation
Anorexia and weight issues

Mechanistic Overlap.

Both involve TNF-$\alpha$ (“cachexin”) elevation
Both show muscle protein catabolism
Both have mitochondrial dysfunction
Both may involve the same metabolic “shutdown” program

Open Question: Cachexia Without Cancer?

What if ME/CFS is essentially “cachexia without cancer”—the same metabolic shutdown program activated by inflammation, but without a tumor driving it? The “safe mode” hypothesis becomes even more plausible: the body is running a program designed for survival during severe illness (cancer, infection, trauma) but triggered inappropriately or locked on.

18 The Hibernation/Torpor Analogy

Some researchers have noted similarities between ME/CFS and hibernation:

Hibernation Features.

Profound metabolic suppression
Reduced body temperature
Altered fuel utilization (lipid preference)
Immune quiescence
Rapid reversibility (in hibernators)

ME/CFS Parallels.

Metabolic suppression (documented)
Some patients report feeling cold
Altered fuel utilization (documented)
Immune changes (documented)
NOT rapidly reversible (the “lock”)

Open Question: Stuck in Torpor?

What if ME/CFS involves activation of ancient metabolic programs related to torpor or hibernation—programs that are suppressed in humans but not deleted from our genome? A severe enough trigger might activate these dormant programs. In hibernating animals, specific signals trigger arousal. In ME/CFS patients, those arousal signals might be missing or ineffective.

If true, studying the molecular biology of hibernation arousal might reveal therapeutic targets for ME/CFS.

19 Symptom-Specific Speculations

Some specific ME/CFS symptoms suggest particular connections:

Coat Hanger Pain (Neck/Shoulder Pain in Distribution of Trapezius).

Classic dysautonomia symptom from muscle ischemia during orthostatic stress
Speculative link: May indicate small vessel disease or microvascular dysfunction; could also reflect craniocervical issues

Post-Exertional Malaise Delay (24-72 Hours).

Not immediate like normal fatigue
Speculative link: Time course matches delayed-type hypersensitivity immune responses; may indicate immune-mediated component to PEM

“Wired but Tired” (Exhausted but Unable to Sleep).

Paradoxical hyper-arousal with fatigue
Speculative link: Classic presentation of ion channel dysfunction affecting both excitation (hyperactive) and energy (depleted); or circadian desynchronization with misaligned sleep drive and circadian alerting

Alcohol Intolerance.

Many ME/CFS patients cannot tolerate even small amounts
Speculative link: Could indicate ALDH dysfunction, already-compromised NAD+ pools (alcohol metabolism consumes NAD+), or mast cell activation (alcohol triggers mast cell degranulation)

Orthostatic Cognitive Impairment (Worse When Standing).

Cognitive function declines in upright position
Speculative link: Cerebral hypoperfusion from autonomic dysfunction, but could also indicate position-sensitive CSF dynamics affecting brain function (supporting glymphatic hypothesis)

Symptom Fluctuation with Menstrual Cycle.

Many female patients report cycle-dependent symptoms
Speculative link: Estrogen and progesterone affect immune function, mast cells, mitochondria, and virtually every proposed mechanism; hormonal influence on HERV expression might explain cyclical viral-like symptoms

20 ADHD, Autism Spectrum, and Depression as Secondary Manifestations of ME/CFS

A striking feature of ME/CFS is the high co-occurrence of ADHD, autism spectrum disorder (ASD), and depression. These relationships are conventionally treated as comorbidities—parallel conditions sharing genetic susceptibility or diagnostic overlap. However, the ADHD–ASD relationship itself is one of the strongest genetic overlaps in psychiatry. Twin studies — which compare identical and fraternal twins to estimate how much of a trait is inherited — find that 54–60% of the genetic influences on ADHD and ASD are shared (a measure called the genetic correlation, denoted $r_\text{g}$) (Polderman et al. 2014). Large genome-wide studies (GWAS), which scan the entire genome for common DNA variants associated with each condition, find a similar overlap of 32–41% (Demontis et al. 2023) (Grove et al. 2019), and when all five major psychiatric disorders are analyzed together, ADHD and ASD genetically cluster as a single neurodevelopmental group, distinct from the psychotic/mood disorder group (Cross-Disorder Group of the Psychiatric Genomics Consortium 2013). The practical implication for ME/CFS is that co-occurrence of ADHD and ASD is the genetically expected outcome, not an incidental finding: when an ME/CFS patient presents with one neurodevelopmental condition, the other should be actively screened for because dual occurrence is the norm, not the exception. This shared genetic foundation also means the metabolic pathways discussed throughout this chapter — BH4 depletion, dopamine inefficiency, mitochondrial dysfunction — are likely present in amplified form when both conditions coexist, compounding the risk of ME/CFS onset after an immune trigger. An alternative framing warrants rigorous examination: can all three emerge as secondary manifestations of ME/CFS, arising from the same neurobiological cascades that produce the primary illness? And can body-wandering—the brain’s spontaneous, internally-oriented bodily attention described by Banellis et al. —provide a unifying mechanistic lens?

This section develops that hypothesis with strict attention to the difference between what the evidence establishes and what remains speculative. The discussion is a research-level mechanistic analysis, not clinical guidance; patients should not alter treatment plans based on the hypotheses presented here without consulting their physician.

20.1 Epidemiological Signal: Prevalence and Temporal Structure

The epidemiological co-occurrence is well-documented. Children with ADHD are twice as likely to develop chronic disabling fatigue by age 18 compared to the general population . Autistic children show a 78% elevated risk of chronic disabling fatigue by age 18, independent of depressive symptoms . Rates of ADHD in ME/CFS cohorts reach 40–47% with childhood-onset ADHD and ~21% with persistent adult ADHD . Among autistic adults, 60% score at or above clinical cutoff for central sensitivity syndrome symptoms and 21% carry a formal ME/CFS or fibromyalgia diagnosis . Critically, ADHD and ASD themselves co-occur far beyond chance: a meta-analysis of 96 studies found pooled ADHD prevalence of 28% (95% CI 25–32) in ASD — the highest of all co-occurring mental health conditions (Lai et al. 2019). A Swedish register study of 1.9 million births found that a diagnosed ASD elevated the odds of also having ADHD 22-fold (OR = 22.33). The strength of this association tracked with genetic relatedness: identical twins (who share all their genes) had 18-fold elevated odds; fraternal twins (who share half) had 4-fold; and full siblings had 5-fold — a gradient that confirms shared genes, not shared environment, is the driving force (Ghirardi et al. 2018). The dimensional picture is supported by data from 17,770 adult twins: the observable correlation between autistic and ADHD traits was $r = 0.51$ (on a scale where 0 = no relationship and 1 = perfect correspondence), and the genetic contribution to this overlap was 54–60% (Polderman et al. 2014). The consequence for ME/CFS is that many patients will carry traits or diagnoses of both neurodevelopmental conditions simultaneously. Since each condition independently reduces metabolic reserve through its own mechanism (dopaminergic inefficiency in ADHD; systemic mitochondrial dysfunction in ASD), the combination means these patients start from a doubly lowered baseline before any infection. In clinical terms: a patient presenting with both ADHD and ASD traits should be considered at substantially higher risk for developing ME/CFS after a triggering infection than someone with either condition alone, and treating their modifiable metabolic deficits — such as iron deficiency or BH4 depletion — may offer greater protective benefit because the same intervention can partially correct deficits arising from two converging pathways.

These figures establish comorbidity but not causality. Critically, the temporal directionality in the available data does not support Architecture B straightforwardly: the Norris and Rimes studies both show ADHD preceding fatigue onset (childhood ADHD predicting later chronic fatigue), not the reverse. This is consistent with Architecture A (shared vulnerability) or bidirectional amplification, and constitutes evidence against a simple “ME/CFS causes ADHD” narrative. Architecture B may still apply to a subset of patients who develop new-onset inattention after ME/CFS, but the epidemiological data suggest this subset is smaller than the shared-vulnerability population.

Two competing architectures are logically possible:

Architecture A (shared vulnerability): A common upstream factor—genetic, immune, or developmental—predisposes individuals to both ME/CFS and neurodevelopmental/psychiatric conditions independently. The co-occurrence is then a selection artefact: the same underlying vulnerability produces all four phenotypes. The genetic architecture of ADHD and ASD themselves is partially shared — large genome-wide studies consistently find that 32–41% of the common genetic variants contributing to one condition also contribute to the other (Demontis et al. 2023) (Grove et al. 2019), and when analyzed alongside other psychiatric disorders, ADHD and ASD form a single neurodevelopmental cluster, genetically distinct from the psychotic and mood disorder cluster (Cross-Disorder Group of the Psychiatric Genomics Consortium 2013). This shared genetic architecture is the most parsimonious starting point for Architecture A: variants that jointly increase risk for ADHD and ASD may also influence the immune and metabolic pathways that determine ME/CFS susceptibility. For clinicians, this means the co-occurrence of neurodevelopmental conditions with ME/CFS should not be dismissed as coincidental or psychosomatic — it is a genetically grounded pattern with predictable biological mediators (IL-6, BH4, mitochondrial function) that are measurable and, in some cases, modifiable.

Architecture B (secondary cascade): ME/CFS pathophysiology—neuroinflammation, dopaminergic depletion, HPA axis dysfunction, interoceptive disruption—produces downstream neuropsychiatric phenotypes in individuals who would not otherwise develop them. The ADHD/ASD/depression arises because of ME/CFS, not alongside it.

No definitive longitudinal study has yet distinguished these architectures in ME/CFS. However, two independent lines of evidence demonstrate that infection-triggered immune cascades can produce acquired neuropsychiatric phenotypes, establishing the minimum requirement for Architecture B’s plausibility. The two lines differ in evidentiary weight: one provides phenomenological consistency in an ME/CFS-proximal population (Long COVID); the other provides molecular-level mechanistic characterization in a structurally parallel disease domain (PANS/PANDAS). Together they are complementary — each compensates for the other’s limitation.

Long COVID. Post-COVID patients develop ADHD-like executive dysfunction and neuroinflammation-mediated depressive symptoms through mechanisms paralleling those proposed for ME/CFS. This consistency does not constitute proof—long COVID data were not predicted by ME/CFS Architecture B models but were retrospectively fitted to the framework — but provides one line of evidence from the most ME/CFS-proximal post-infectious population.

PANS/PANDAS. A second line comes from a different disease domain with partial mechanistic characterization (contested — see (Chiarello et al. 2017) (Sigra, Hesselmark, and Bejerot 2018)). Pediatric Acute-onset Neuropsychiatric Syndrome (PANS) and its streptococcal-triggered subtype (PANDAS) are defined by abrupt-onset OCD, tics, and behavioral regression temporally linked to infection — most commonly group A streptococcal (GAS) pharyngitis (Swedo et al. 1998). The diagnostic criteria require acute, dramatic symptom onset with a documented infectious or inflammatory trigger, producing a clinical picture that can closely resemble severe autism exacerbation or late-presenting neurodevelopmental regression (K. Chang et al. 2015) (Masterson et al. 2025). The leading molecular model — from the Cunningham laboratory — holds that GAS-directed antibodies cross-react with neuronal lysoganglioside in the basal ganglia via molecular mimicry, activating CaMKII in dopaminergic neurons and producing circuit-specific dysfunction in the same basal ganglia-thalamocortical loops that subserve behavioral control (Kirvan et al. 2003) (Kirvan et al. 2006) (Cesaroni et al. 2026). Independent replication has been inconsistent, and the Cunningham Panel (based on this pathway) is research-use only and not FDA-approved (Chiarello et al. 2017) (Sigra, Hesselmark, and Bejerot 2018).

PANS/PANDAS provides a plausibility precedent for Architecture B that ME/CFS alone cannot yet offer — though the analogy is imperfect given the biological divergences: (1) a partially characterized infection-to-autoantibody pipeline (GAS → anti-neuronal IgG → CaMKII activation — proposed, contested), (2) an acquired neuropsychiatric phenotype (OCD, tics, regression) that is clinically indistinguishable from developmental neuropsychiatric conditions, and (3) treatment response — a subset of PANS/PANDAS patients improve dramatically with IVIG or therapeutic plasma exchange in unblinded series (Latimer et al. 2015) (Kovacevic, Grant, and Swedo 2015), suggesting the symptoms are immune-mediated in at least some cases, though the single double-blind RCT (n=35) failed to demonstrate a significant advantage over placebo overall (K. A. Williams et al. 2016). A systematic review rated overall evidence quality as low, citing study heterogeneity (Sigra, Hesselmark, and Bejerot 2018). The pattern — dramatic responders alongside non-responders, positive unblinded series but equivocal RCT — is structurally similar to the immunotherapy data in ME/CFS and reflects the same underlying problem: syndromic inclusion criteria capture a biologically heterogeneous population, and immunotherapy only works in the immune-mediated subset (La Bella et al. 2023). However, this pattern is equally consistent with substantial open-label expectation effects in a condition where symptoms are subjective and treatment cannot be truly blinded.

The PANDAS controversy itself is instructive. Whether PANDAS “really exists” as a distinct entity has been debated for two decades (Chiarello et al. 2017) (Cesaroni et al. 2026). The debate has legitimate methodological concerns on both sides: proponents argue the syndrome is immune-mediated (the Swedo/Frankovich line), while skeptics note that GAS infection and OCD are common in the pediatric population, diagnostic criteria lack specificity, and the one RCT was null (the Gilbert/Singer line) (Chiarello et al. 2017). Unlike the ME/CFS biological-psychosomatic polarization, the PANDAS debate involves methodologically substantive disagreements and is not settled. The emerging view from the Stanford PANS clinic’s 10-year data (Masterson et al. 2025) — which itself originates from the pro-PANDAS side of the debate — is that the question may be mis-framed: rather than asking whether PANDAS “really exists,” the clinically tractable question is what proportion of acute-onset pediatric neuropsychiatric presentations are immune-mediated. This is precisely the question the present paper asks about ME/CFS and acquired ADHD/ASD-like features, though with the critical difference that in ME/CFS this hypothesis has no direct evidence at all.

For the paper’s purposes, PANS/PANDAS provides Architecture B’s strongest analogy: a disease domain in which an infection-to-autoantibody-to-neuropsychiatric-phenotype pipeline has been proposed, and in which treatment response — partial, stratified, contested, but real — has been documented. The circuit is different (basal ganglia in PANDAS vs diffuse neuroinflammatory/thalamocortical/metabolic in ME/CFS), but the disease logic is analogous: infection triggers an immune response that produces neuropsychiatric symptoms that syndromic classification attributes to a “different” disorder. The analogy is imperfect — PANDAS is pediatric, streptococcal, and basal-ganglia-specific, while ME/CFS is predominantly adult, post-viral, and diffuse — and the evidential support for the PANDAS mechanism is substantially weaker than a “well-characterized” label would imply. PANDAS does not prove Architecture B so much as it makes it a less remarkable claim than if no post-infectious autoimmune neuropsychiatric syndrome existed at all.

Consequence: PANDAS does not prove Architecture B, but it demonstrates that infection-triggered immune responses can produce neuropsychiatric syndromes in a subset of patients — making Architecture B in ME/CFS more plausible than if no such parallel existed, though the specific magnitude of the plausibility increase is small given the biological divergences between the conditions.

Speculation: PANS/PANDAS as a Plausibility Precedent for Acquired Neuropsychiatric Phenotypes

Certainty: 0.35. The PANS/PANDAS literature — a proposed molecular mimicry pathway (contested, see below) (Kirvan et al. 2003) (Kirvan et al. 2006), immunotherapy response in some patients (unblinded series positive, single RCT null) (Latimer et al. 2015) (Kovacevic, Grant, and Swedo 2015) (K. A. Williams et al. 2016), and clinical recognition of immune-mediated acquired behavioral regression (Masterson et al. 2025) (La Bella et al. 2023) — constitutes a plausibility precedent for Architecture B in a closely related disease domain. The proposed mechanism differs from the ME/CFS cascade (streptococcal→basal ganglia in PANDAS vs viral/immune→diffuse neuroinflammatory→thalamocortical/metabolic in ME/CFS), but the disease logic — infection→immune response→neuropsychiatric phenotype→potentially reversible in a subset with immunomodulation — is analogous. The PANDAS controversy’s resolution through stratification (some patients are immune-mediated, some are not) prefigures the same resolution the present paper advocates for ME/CFS-associated ADHD/ASD-like features. Falsification condition: In a cohort of n ≥ 80 ME/CFS patients with documented post-infectious onset (≤ 3 months from infection to ME/CFS symptom onset) and new-onset inattention or executive dysfunction post-dating ME/CFS onset, systematic anti-neuronal antibody screening (anti-basal ganglia, anti-D1/D2 receptor, anti-CaMKII, anti-lysoganglioside IgG) versus matched healthy controls (n ≥ 80) will show enrichment of ≥ 1 specificity at ≥ 2× control prevalence (p < 0.05 after FDR correction for the panel size). Falsified if the panel shows no significant enrichment (all q > 0.30) and the antibody burden does not correlate with ADHD-symptom-severity change-from-premorbid-baseline (Spearman ρ < 0.15). A null result refutes the claim that the PANDAS mechanism — infection-triggered anti-neuronal antibodies producing acquired neuropsychiatric symptoms — generalizes to ME/CFS-associated inattention. (The PANDAS literature remains a worked example that immune-mediated acquired neuropsychiatric phenotypes are possible in humans regardless of ME/CFS outcome; the speculation tested here is the claim that this same mechanism operates in the ME/CFS-ADHD overlap population.)

Limitations: PANDAS is paediatric, streptococcal-triggered, and basal-ganglia-specific — all three diverge from the proposed ME/CFS Architecture B mechanism. The analogy strengthens Architecture B’s plausibility without directly testing it. The PANDAS immunotherapy RCT was underpowered and its GAS+ subgroup analysis is hypothesis-generating, not confirmatory (K. A. Williams et al. 2016). No study has examined whether PANDAS/PANS resolves into an ME/CFS-like state.

Consequence: The PANDAS precedent raises the research question whether some ME/CFS-associated inattention and sensory rigidity is immune-mediated and potentially immunotherapy-responsive, rather than representing comorbid primary ADHD or ASD. This hypothesis is entirely untested — no ME/CFS patient has been screened for the PANDAS antibody panel, and the analogy rests on disease-logic similarity across different circuits, not on direct evidence. The PANDAS lesson that selecting the immune-mediated subset is the hard part should temper any expectation of rapid clinical translation. Research-stage only — no clinical implications can be derived without direct testing of anti-neuronal antibody prevalence in ME/CFS.

Limitation: Epidemiology Establishes Comorbidity, Not Causality

The prevalence figures cited above reflect cross-sectional or retrospective designs in specialist referral cohorts. No prospective study has tracked ADHD, ASD, or depression emergence after confirmed ME/CFS onset in a pre-illness-characterized cohort. The elevated comorbidity rates are consistent with shared vulnerability (Architecture A), secondary cascade (Architecture B), or bidirectional amplification. Without temporal precedence data, causal direction cannot be established from epidemiology alone.

Synthesis: Syndromic Labels Obscure Mechanism-Based Subgroups: The PANDAS-Architecture B Convergence

Certainty: 0.30. Three parallel lines of reasoning, each drawn from a different disease domain, converge on the same structural claim: diagnostic labels codified in DSM and ICD bin biologically heterogeneous populations into syndromic categories, and this binning systematically obscures the immune-mediated subset within each category.

The convergent lines are:

The PANDAS controversy resolution: Two decades of debate over whether PANDAS “really exists” resolved not with a verdict but with a reframing — the question is not whether the syndrome is real but what proportion of acute-onset paediatric neuropsychiatric presentations are immune-mediated (Chiarello et al. 2017) (Masterson et al. 2025). The answer is “some, not all.” The residual is that a diagnostic label functions as a heterogeneous bin, and the clinically consequential distinction — immune-mediated vs not — is orthogonal to the label itself.
Architecture B in ME/CFS: The same structural problem recurs in the ME/CFS-ADHD/ASD overlap. Whether acquired inattention and sensory rigidity are immune-mediated (Architecture B), shared-vulnerability (Architecture A), or bidirectional amplification cannot be determined from diagnostic labels alone — because the labels collapse these biologically distinct trajectories into a single “comorbidity” category PANS/PANDAS as a Plausibility Precedent for Acquired Neuropsychiatric Phenotypes
IVIG response in ASD: The Connery 2018 vs Plioplys 1998 contrast — dramatic immunotherapy response in autoantibody-stratified ASD, null in unselected ASD — is the clearest proof that the therapeutic consequence of this binning problem is real, not theoretical. When recruitment uses syndromic criteria, the trial is underpowered by the inclusion of non-immune-mediated cases; when recruitment uses antibody profiling, the same intervention shows efficacy in the subset it targets (Connery et al. 2018) (Plioplys 1998)

The convergence is that each of these lines independently demonstrates the same structural pattern in a different diagnostic bin: PANDAS (immune-mediated subset within paediatric acute-onset neuropsychiatric syndrome), ME/CFS-ADHD (immune-mediated subset within post-infectious fatigue + inattention), and ASD (immune-mediated subset within neurodevelopmental autism). The pattern is not disease-specific — it is a property of syndromic classification itself. Syndromic labels are diagnostically convenient but therapeutically misleading when the actionable mechanism does not respect the label’s boundary.

Practical consequence: This convergence supports the paper’s central argument — mechanism-based stratification over syndromic labeling for treatment selection. It provides a worked example drawn from a molecularly characterized disease domain (PANDAS) where the mechanism-to-treatment logic has been partially validated (IVIG responders). Whether this logic extends to ME/CFS is the research question this paper raises; the convergence establishes that the question is well-posed, not that the answer is known.

Limitations: The PANDAS RCT was underpowered and equivocal (K. A. Williams et al. 2016). The Connery 2018 IVIG-in-ASD study is a single specialist-center series, not a multi-site trial. No prospective study has simultaneously applied the PANDAS antibody panel, ASD maternal-antibody panel, and ME/CFS functional autoantibody panel to the same cohort. The claim that syndromic labels systematically obscure immune-mediated subsets is supported by multiple independent examples but has not been proven as a general principle.

Falsification condition: A multi-disease immunotherapy platform trial (IVIG or TPE) recruiting across four diagnostic categories (PANDAS/PANS, ME/CFS with neuropsychiatric features, regressive ASD, and post-COVID neuropsychiatric syndrome) and stratifying by anti-neuronal antibody profile will test the central claim. If antibody-profile-stratified subgroups show equivalent immunotherapy response regardless of diagnostic label, the convergence is validated. If diagnostic label significantly predicts response independent of antibody profile, the convergence is refuted — mechanism-based stratification offers no advantage over syndromic criteria, and the PANDAS precedent does not generalise.

Severity applicability: Unknown — the convergence is a structural argument about classification, not a clinical staging model. Each constituent condition has its own severity spectrum.

20.2 Mechanism I: Neuroinflammation as a Common Upstream Driver

The most frequently invoked neurobiological bridge between ME/CFS and all three psychiatric phenotypes is neuroinflammation—though “bridge” may overstate the evidence, since neuroinflammation is documented in each condition independently without proof that ME/CFS-specific neuroinflammation causes the others. One PET study ($n = 9$ ME/CFS patients) reported 45–199% elevation in microglial activation markers across six brain regions , but a subsequent study using the same TSPO radioligand in women with CFS and Q fever fatigue syndrome ($n = 31$) found no signs of neuroinflammation . The neuroinflammation foundation for Architecture B is therefore contested, not established; all downstream claims inherit this uncertainty. Separately, blood-brain barrier dysfunction in ME/CFS may allow peripheral cytokines (TNF-$\alpha$, IL-1$\beta$, IL-8) and immune cells to enter the CNS , providing an alternative route to central immune activation that does not require resident microglial involvement.

These same neuroinflammatory mechanisms link independently to ADHD, ASD, and depression:

ADHD and neuroinflammation: Substantial evidence supports neuroinflammation in ADHD pathophysiology , including elevated pro-inflammatory cytokines in children with ADHD and microglial activation in post-mortem and imaging studies.
ASD and neuroinflammation: Altered neuroinflammation has been documented across four decades of ASD research , with microglial and astroglial activation characteristic of ASD post-mortem brain tissue.
Depression and neuroinflammation: Cytokine-induced activation of the indoleamine 2,3-dioxygenase (IDO) pathway depletes serotonin and produces the neurotoxic metabolite quinolinic acid , providing a mechanistic route from inflammation to depressive phenotype distinct from classical monoamine depletion.

The critical question is whether ME/CFS neuroinflammation—if confirmed by future replication—is severe enough, and sufficiently targeted to the relevant circuits, to produce these downstream effects. The Nakatomi study reported involvement of cingulate cortex, hippocampus, amygdala, thalamus, midbrain, and pons , which overlaps with circuits implicated in each of the three conditions; but this finding awaits independent replication (see above).

Speculation: Shared Neuroinflammatory Cascade as Generator of All Three Phenotypes

Certainty: 0.35. Chronic microglial activation in ME/CFS—particularly in mesolimbic, thalamocortical, and prefrontal circuits—may be sufficient to produce ADHD-like, ASD-like, and depressive phenotypes in individuals with intact pre-illness neurodevelopment. If this cascade is the proximate cause, interventions targeting neuroinflammation should partially reverse these secondary phenotypes—a testable and not yet tested prediction. Candidate agents include low-dose naltrexone (off-label, physician-supervised; see Ch. 18 for dosing and contraindications); anti-cytokine biologics (tocilizumab, anakinra) are research-context only, require specialist prescribing, and have no ME/CFS trial evidence. This hypothesis does not require identical mechanisms to primary ADHD, primary ASD, or primary depression; it only requires that neuroinflammation degrades the same circuits that those conditions affect through developmental means.

20.3 Mechanism II: Dopaminergic Dysfunction and the Effort-Computation Failure

In animal models and computational accounts, dopamine encodes not only reward prediction errors but tonic behavioral vigor—the speed and energy with which organisms execute goal-directed actions . Tonic dopamine depletion in the striatum produces effort aversion: organisms maintain intact reward discrimination (they still want the reward) but shift preference sharply toward low-effort/low-reward options, producing a motivational profile that may be clinically difficult to distinguish from ADHD-type executive dysfunction.

In ME/CFS, preliminary evidence for dopaminergic deficiency exists but remains limited. The NIH deep phenotyping study found significantly reduced homovanillic acid (HVA, the primary dopamine metabolite) in cerebrospinal fluid, alongside reduced putamen activity correlating with poor reward sensitivity (Walitt et al. 2024). ADHD stimulants improve brain fog in 77.1% and fatigue perception in 71.7% of ME/CFS patients surveyed —a pharmacological response pattern consistent with dopaminergic insufficiency rather than primary ADHD.

The effort-cost computation depends on a circuit involving the dorsal anterior cingulate cortex (dACC), ventromedial prefrontal cortex (VMPFC), and striatum . In the neuroeconomic framework, action value is modeled as:

\[ A = \mathbb{E}[R] - k dot C_{\text{effort}} \]

where $\mathbb{E}[R]$ is expected reward, $C_{\text{effort}}$ is the effort cost, and $k$ is a scaling coefficient regulated by dopaminergic tone. This equation is qualitative in the ME/CFS context—no parameter values have been estimated from ME/CFS data—but it makes a directional prediction: If this model applies to ME/CFS, the documented dopamine depletion would effectively raise $k$, making effort costs feel prohibitively high even for low-cost actions. The predicted behavioral signature—prolonged decision latency, preference for minimal-demand tasks, and collapse of sustained attention when reward salience drops—matches the cognitive profile reported by ME/CFS patients, though this match is circumstantial rather than mechanistically proven.

According to one influential account of primary ADHD, the deficit is characteristically one of incentive salience instability: dopaminergic tone fluctuates rather than being chronically depleted, producing intermittent hyperfocus alongside inattention . If this distinction holds, ME/CFS-secondary “ADHD-like” presentation should be more uniform—a tonic suppression of effort invigoration—and may respond differently to stimulants. This prediction has not been directly tested.

Hypothesis: Dopaminergic Tonic Depletion as Generator of ADHD-Like Phenotype in ME/CFS

Certainty: 0.45. Chronic dopaminergic depletion in ME/CFS—mediated by neuroinflammation (IDO pathway diverting tryptophan away from catecholamine synthesis), metabolic constraint (reduced CNS energy supply for neurotransmitter turnover), and HPA axis hypocortisolism (cortisol normally upregulates tyrosine hydroxylase)—produces a tonic effort-invigoration deficit. Clinically, this manifests as inattention, executive dysfunction, and motivational collapse that closely resembles ADHD but arises from a different mechanism. Falsification condition: If stimulant response in ME/CFS-associated inattention shows the same variability and concentration-dependence as primary ADHD, primary ADHD mechanisms are implicated; if stimulant response is more uniform and dose-linear, tonic depletion is more likely. CSF HVA level should predict response.

20.4 Mechanism III: Interoceptive Hierarchy Disruption and the Body-Wandering Paradox

A third proposed mechanism—more elaborate than neuroinflammation or dopaminergic depletion alone—operates at the level of interoceptive predictive processing. A parsimony caveat is warranted: the simpler cytokine-induced sickness behavior pathway (Mechanism I) may already account for depressive and cognitive symptoms without requiring the theoretical apparatus of predictive processing. The framework below is therefore offered not as the only explanation but as one that, if validated, would provide a more differentiated account of why ME/CFS produces ADHD-like features in some patients, ASD-like features in others, and depression in most—a specificity that simple neuroinflammation models do not explain. The theoretical components are imported from other fields (computational psychiatry, interoceptive neuroscience) and none has been directly validated in ME/CFS.

The interoceptive hierarchy. The brain does not passively receive bodily signals; it generates predictions about internal states and computes prediction errors when actual afferent signals diverge from expectation . This interoceptive inference operates through a hierarchical network: brainstem nuclei supply raw visceral afference; posterior insula maps first-order bodily states; mid-insula contextualises them; anterior insula and dorsal ACC assign allostatic significance; prefrontal cortex encodes higher-order beliefs about self-regulatory efficacy. Each level sends downward predictions and receives upward prediction errors.

Precision-weighting and its failure. Healthy interoceptive inference requires adaptive precision-weighting: the relative confidence assigned to prior predictions versus incoming sensory evidence must adjust dynamically to context. Powers et al. demonstrated, across anxiety, depression, eating disorders, and substance use disorders, a transdiagnostic failure to adaptively update precision estimates during an interoceptive perturbation (breath-holding to amplify cardiac afference). All clinical groups failed to increase precision-weighting on ascending cardiac signals as healthy controls did. The result is a locked system: the brain cannot accurately register whether its regulatory commands are working, and therefore cannot calibrate them.

Body-wandering as endogenous interoceptive monitoring. Banellis et al. characterized body-wandering—spontaneous, task-unrelated thought directed toward somatomotor and visceral bodily sensations—as a distinct dimension of mind-wandering with its own neural signature: increased connectivity among somatomotor, interoceptive, and thalamocortical networks, with elevated negative affect and physiological arousal during episodes. The paradox is that individuals with higher habitual propensity for body-wandering show lower ADHD severity and lower depression—the opposite of what simple negative-affect coupling would predict. Their cross-validated canonical correlation analysis ($N = 536$, CCA canonical loadings: stomach 0.576, arousal 0.442, breathing 0.345, movement 0.300, negative affect 0.286) established this as a reproducible individual-differences dimension.

The proposed explanation: habitual body-wandering reflects the brain’s capacity for endogenous interoceptive monitoring—actively sampling bodily states even when aversive. This monitoring maintains the precision-weighting update cycles that keep allostatic regulation functional. Those who cannot body-wander (whether due to alexithymia, ADHD-type external attention orientation, or dissociation from interoceptive signals) may lose the endogenous feedback channel that normally prevents allostatic drift—though this causal interpretation goes beyond what the Banellis et al. correlational data establish.

ME/CFS and pathological body-wandering. In ME/CFS, however, the relationship inverts. Patients do not lack body-attention; they are often characterized by interoceptive hypervigilance—sustained, distress-driven monitoring of symptoms . This is not equivalent to adaptive body-wandering. Adaptive body-wandering draws on accurate predictive models of bodily states; hypervigilance reflects amplified attention to prediction errors without the ability to resolve them. The body-wandering literature’s protective paradox—more body-attention predicts better outcomes—may apply only when the underlying interoceptive generative models are intact. If the models are corrupted, as Architecture B predicts in ME/CFS, increased attention to bodily signals would amplify awareness of unresolvable discrepancies, producing symptom amplification rather than allostatic correction.

This distinction maps onto the distinction between primary ASD interoceptive profiles and ME/CFS-acquired interoceptive disruption. ASD involves altered precision of priors—overprecise top-down predictions that suppress sensory updating —whereas ME/CFS, according to the proposed model, involves corrupted lower-level signals (brainstem, thalamocortical) that prevent accurate afference from reaching the cortical models. The phenomenological result may resemble each other (sensory overwhelm, difficulty interpreting internal states, alexithymia), but the computational locus differs.

Speculation: Interoceptive Hierarchy Failure as Generator of Autism-Like Features in ME/CFS

Certainty: 0.30. Thalamocortical disruption and brainstem RAS dysfunction in ME/CFS corrupt the lower levels of the interoceptive hierarchy, generating persistent prediction errors that reach anterior insula and ACC as unresolvable “error signals.” The brain’s compensatory response—increasing prior precision to suppress the noise—produces the same phenomenology as ASD interoceptive hyperprecision: rigid perceptual filtering, difficulty in novel or unpredictable sensory environments, withdrawal from socially demanding contexts, and apparent alexithymia (not from absent emotional response but from inability to link it reliably to a specific bodily state). Crucially, this is acquired rigidity—it should be more context-dependent and potentially reversible with treatment of the underlying neuroinflammation—distinguishing it from developmental ASD where the precision profile is stable and trait-like. Falsification condition: If ME/CFS-associated sensory hypersensitivity shows the same trait stability and context-independence as developmental ASD—persisting unchanged across PEM cycles and not correlating with neuroinflammatory markers—the acquired-rigidity model is not supported.

Speculation: Failed Body-Wandering as the Pathway to ME/CFS-Secondary Depression

Certainty: 0.40. Adaptive body-wandering maintains allostatic regulation by providing the endogenous feedback that allows the brain’s generative model to update beliefs about bodily state. In ME/CFS, the corrupted interoceptive hierarchy degrades this feedback: the brain attends to bodily signals (hypervigilance) but cannot resolve the prediction errors they generate, because the generative model itself is miscalibrated. The accumulation of unresolvable prediction errors—persistently high free energy —produces a learned belief of allostatic self-inefficacy: the brain predicts that it cannot successfully regulate its own physiological state. This belief is not irrational given the functional evidence the brain has accumulated; it is a coherent inference from a corrupted information channel. The depressive phenotype that follows—anhedonia, motivational collapse, negative future prediction—is structurally identical to what Stephan et al. term allostatic interoceptive overload and what Seth terms the affective consequence of persistent interoceptive prediction error. Depression in ME/CFS is therefore not a mood disorder with a separate etiology that happens to co-occur; it is the fourth-order cognitive consequence of the same hierarchy failure that generates fatigue, pain, and post-exertional malaise at lower levels. Falsification condition: If depressive affect in ME/CFS does not correlate with interoceptive accuracy measures (cardiac interoception tasks) more strongly than with rumination scores, or if SSRI-resistance is not linked to interoceptive disruption severity, this pathway is not supported.

20.5 A Formal Causal Structure

The proposed Architecture B can be represented as a directed acyclic graph (DAG) with explicit uncertainty annotations. The following represents the core causal skeleton; all edges are speculative except those annotated as established (E): Figure Failed Body-Wandering as the Pathway to ME/CFS-Secondary Depression presents the proposed causal structure.

Root nodes: Viral/immune trigger → Systemic inflammation
BBB disruption [E: long COVID] → CNS microglial activation
Microglial activation [E: neuroinflammation] → {IDO pathway activation; HPA hypocortisolism; Thalamocortical disruption; RAS dysfunction}
IDO pathway → {Serotonin depletion; Dopamine depletion; Quinolinic acid excitotoxicity}
HPA hypocortisolism → {Unconstrained cytokine cascades; Reduced tyrosine hydroxylase activity}
Thalamocortical disruption → {Brainstem interoceptive signal corruption; Precision-weighting rigidity}
Dopamine depletion → {Effort aversion [ADHD-like phenotype]; Anhedonia [depressive phenotype]}
Precision-weighting rigidity → {Acquired interoceptive inflexibility [ASD-like phenotype]; Failed body-wandering feedback}
Failed body-wandering feedback → {Unresolvable prediction errors → Allostatic self-inefficacy [depressive phenotype]}

Confounders (C): Shared genetic susceptibility (immune dysregulation loci, HLA variants; ADHD and ASD share 32–41% of their common genetic risk variants (Demontis et al. 2023) (Grove et al. 2019)) acts as a common cause of both ME/CFS trigger-susceptibility and elevated baseline risk for ADHD/ASD/depression, creating non-causal association even if Architecture B is false.

The DAG identifies the key confounders that must be controlled and the colliders that must not be conditioned on. If shared genetic susceptibility (C) is the dominant pathway, adjusting for polygenic risk scores for immune dysregulation should substantially attenuate the ME/CFS → psychiatric phenotype associations. If Architecture B is dominant, it should not.

20.6 What Distinguishes Secondary from Primary Conditions

If Architecture B is correct, the clinical and mechanistic distinction between primary conditions and ME/CFS-secondary phenotypes would be important for treatment selection. The predicted distinguishing features listed below are derived from the proposed mechanisms and have not been systematically confirmed in prospective ME/CFS cohort studies. They require professional evaluation and cannot be reliably self-applied.

Predicted secondary ADHD-like presentation:

Onset after ME/CFS onset (not childhood history)
Predominantly inattentive type (rather than combined or hyperactive-impulsive)
Deficits correlate with fatigue severity and neuroinflammatory markers
Stimulant response uniform and dose-linear (tonic depletion model) rather than variable (incentive salience model)
May coexist with primary ADHD (25–40% of ME/CFS patients report childhood ADHD, suggesting Architecture A operates in a significant subset)

Predicted secondary ASD-like features:

Acquired sensory hypersensitivity, not a lifelong trait
Context-dependent: worse during post-exertional malaise, better in low-demand periods
Alexithymia correlates with illness duration and severity, not developmental history
Social withdrawal driven by energy limitation and symptom management, rather than social motivation difference
May be partially reversible with treatment of neuroinflammation

Predicted secondary depression:

Neurobiologically distinct from primary major depressive disorder: IDO/kynurenine mechanism versus monoamine depletion
Correlates with inflammatory markers (CRP, IL-6, TNF-$\alpha$) rather than early adverse experience or rumination scores alone
May respond less completely to SSRIs alone than to combined anti-inflammatory and antidepressant strategies; however, SSRIs remain a standard treatment for symptom management and must not be discontinued without medical supervision
In the allostatic self-inefficacy framework, depressive affect reflects accumulated evidence that self-regulatory actions have failed—a mechanistic framing, not a claim that hopelessness is rational or that treatment is futile. Depression in ME/CFS carries serious morbidity and suicide risk regardless of its mechanistic origin and warrants the same clinical urgency as primary depression. Ecological data from multiple meta-analyses (OR 0.76, 95% CI 0.66–0.87, combined n > 113 million) show a consistent inverse association between trace lithium concentrations in drinking water and suicide rates (Barjasteh-Askari et al. 2020) (Memon et al. 2020). These are ecological designs and do not establish causation; unmeasured confounders (socioeconomic status, healthcare access, other trace minerals) may account for the association. The data provide contextual epidemiological evidence that suicide risk is sensitive to trace-level neurobiological variables, but cannot speak to the mechanism of depression in ME/CFS specifically.

Open Question: Can Neuroinflammation-Targeted Treatment Reverse Secondary Neuropsychiatric Phenotypes?

If ADHD-like features, ASD-like sensory disruption, and depression in ME/CFS arise through the proposed cascade, anti-neuroinflammatory interventions should selectively improve these phenotypes beyond their effect on core fatigue. Low-dose naltrexone (LDN), which modulates microglial activation via toll-like receptor 4, already shows benefit for brain fog and fatigue . LDN is off-label, contraindicated with concurrent opioid use, and requires physician supervision. Whether it also reduces ADHD symptom scores, interoceptive rigidity, or depressive affect in ME/CFS patients—controlling for fatigue improvement—is entirely untested. This is a tractable, clinically relevant question that current trial designs do not address.

Open Question: Does Habitual Body-Wandering Predict ME/CFS Onset, Severity, or Trajectory?

The Banellis et al. finding—that individuals with higher propensity for body-wandering have lower ADHD and depression—raises a specific prediction for ME/CFS: pre-illness body-wandering capacity (a proxy for intact interoceptive feedback circuits) may predict whether post-infectious fatigue resolves or becomes chronic. Individuals with robust endogenous interoceptive monitoring may have stronger allostatic error-correction capacity following the initial immune insult. This is testable in post-infection cohort studies with pre-illness cognitive phenotyping. Body-wandering propensity also predicts the profile of psychiatric sequelae: low baseline body-wandering capacity → greater vulnerability to ME/CFS-secondary depression; disrupted (hypervigilant) body-wandering → greater vulnerability to symptom amplification and acquired ASD-like features.

20.7 Implications for Clinical Management

Warning: Research Hypotheses, Not Treatment Protocols

The implications below derive from Architecture B, which remains unproven. They are intended for clinicians and researchers, not as patient-directed guidance. Patients should not discontinue prescribed psychiatric medications or decline psychiatric referrals on the basis of this speculative framework. ADHD, depression, and sensory processing difficulties warrant treatment in their own right, regardless of whether they are primary or secondary to ME/CFS. All medication changes require physician supervision.

The Architecture B framing, if validated, would have clinical implications that differ from the standard comorbidity framing:

Consider the inflammatory substrate alongside the psychiatric phenotype. If ADHD-like and depressive features arise partly from neuroinflammation, targeting inflammation may complement—not replace—standard psychiatric treatment. This is an argument for adding anti-inflammatory strategies, not for withdrawing effective psychiatric medications.
Stimulants for cognitive symptoms provide real benefit. In a patient-reported outcomes survey ($N = 3{,}925$, not an RCT), ADHD stimulants improved subjective brain fog in 77.1% and fatigue perception in 71.7% of ME/CFS respondents . Self-selection and placebo effects limit this evidence, but the signal is consistent with dopaminergic insufficiency. Addressing cognitive dysfunction is a legitimate and valuable treatment goal even if it does not alter the primary illness. However, stimulants’ negligible or negative effect on PEM means they do not address exertional limits. Clinicians must also consider the sympathomimetic effects of stimulants in the 30–40% of ME/CFS patients with comorbid POTS: methylphenidate and amphetamine salts can worsen tachycardia and interact with midodrine, fludrocortisone, or beta-blocker regimens. Stimulant prescribing in ME/CFS requires cardiac-aware monitoring.
Interoceptive rehabilitation is experimental and unvalidated. If the pathway to depression in ME/CFS runs through failed body-wandering feedback and allostatic self-inefficacy, interventions that restore accurate interoceptive awareness might complement standard depression treatment. However, no validated protocol exists for PEM-safe interoceptive rehabilitation in ME/CFS. Body-scan practices and biofeedback are well-established in other populations but have not been tested in ME/CFS. Any future protocol must be clinician-supervised, respect strict pacing constraints, and exclude severely ill patients who cannot safely engage in structured cognitive activities. This is a research direction, not a current recommendation; it should not substitute for established psychological support or antidepressant treatment.
Acquired sensory features warrant support, not dismissal. Acquired sensory hypersensitivity and social withdrawal in ME/CFS may serve an adaptive function (reducing stimulation load during energy depletion). Recognising this may reduce iatrogenic harm from demands for social normalization. However, patients experiencing these features may still benefit from occupational therapy, environmental accommodations, and autism-informed support services; the point is to offer appropriate support rather than pathologize illness adaptations.

Limitation: Architectural Uncertainty: Architecture A Cannot Be Ruled Out

The entire section above develops Architecture B (ME/CFS → secondary phenotypes) in mechanistic detail, because this architecture has clinical implications that the standard comorbidity framing misses. However, the evidence does not yet rule out Architecture A. The shared vulnerability model predicts identical epidemiological patterns (elevated comorbidity) and identical pharmacological responses (stimulants help both primary ADHD and ME/CFS executive dysfunction) through entirely different causal pathways. Distinguishing these architectures requires longitudinal studies with pre-illness psychiatric phenotyping, Mendelian randomisation studies using ME/CFS genetic instruments, and treatment trials designed to test whether anti-inflammatory interventions reduce psychiatric symptom burden independently of fatigue improvement. Until such studies exist, both architectures should inform clinical reasoning.

20.8 Architecture C: The Metabolic Reserve Hypothesis — Pre-Existing Energy Deficits as ME/CFS Predisposition

The preceding sections develop Architectures A (shared vulnerability) and B (ME/CFS → secondary phenotypes). A third architecture warrants separate treatment because it makes distinct predictions and identifies a different set of at-risk populations: certain pre-existing conditions chronically reduce the brain’s metabolic reserve — the buffer between baseline energy demand and maximum energy production capacity — so that an immune trigger (EBV, SARS-CoV-2, other pathogens) that would produce transient post-infectious fatigue in a metabolically robust individual instead precipitates sustained decompensation into ME/CFS.

Note on terminology: “Metabolic reserve” is used throughout this section at three levels of description — (1) the theoretical whole-organism quantity $R_\text{headroom} = (J_\text{production,max} - J_\text{demand,peak}) / J_\text{production,max}$, (2) the measurable cellular proxy of PBMC spare respiratory capacity, and (3) the clinical concept of functional buffer before symptom threshold. These are related but not identical; cellular measurements may not directly reflect whole-organism reserve.

Architecture C differs from Architecture A in a critical respect: Architecture A posits shared genetic vulnerability without specifying the mechanism by which it predisposes. Architecture C identifies the mechanism as reduced metabolic reserve — a measurable, potentially modifiable state. It differs from Architecture B by reversing the causal arrow: the neurodevelopmental condition precedes and predisposes to ME/CFS rather than arising from it.

20.8.1 ADHD as Brain Energy Failure: The Primary Case

ADHD is the condition for which the energy-deficit framing is most directly supported. Zametkin et al. demonstrated in the foundational PET study that ADHD adults show 8.1% lower global cerebral glucose metabolism than controls, with the largest reductions in premotor and superior prefrontal cortex — the regions most metabolically demanding and most critical for sustained attention . A systematic review of 20 cerebral blood flow studies (total $n = 1{,}652$ ADHD participants) confirmed resting-state hypoperfusion in prefrontal, temporal, and basal ganglia regions, with methylphenidate normalising striatal and thalamic blood flow . Both glucose metabolism and cerebral perfusion data converge: the ADHD brain operates with less energy delivery to the regions it needs most.

The energy framing extends beyond perfusion. Mitochondrial dysfunction in ADHD — decreased membrane potential, impaired oxidative phosphorylation, and variants in Complex~I assembly genes (NDUFAF2, UCP2) — has been documented primarily in preclinical models but is convergent with the ME/CFS mitochondrial literature . Dual-tracer PET imaging in drug-naive ADHD adults ($n = 24$ per group) simultaneously demonstrated reduced D1 receptor availability in the anterior cingulate cortex and increased microglial activation in the dorsolateral prefrontal cortex, with the two measures correlated only in ADHD subjects . This co-imaging finding is mechanistically important: the same brain shows both catecholamine deficit and neuroinflammation — precisely the combination documented in ME/CFS by the NIH deep phenotyping study (Walitt et al. 2024).

Iron deficiency compounds the energy deficit through a dual mechanism: iron is a required cofactor for both dopamine/norepinephrine synthesis (via tyrosine hydroxylase) and mitochondrial complex~I/II function . Iron-deficient neurodivergent individuals thus suffer impaired neurotransmitter production and reduced ATP yield — a modifiable metabolic reserve reducer. Per neurodevelopmental guidelines, ferritin thresholds below which iron supplementation is recommended are $< 30$ ng/mL for ADHD and $< 50$ ng/mL for ASD sleep phenotypes ; whether these thresholds are relevant to ME/CFS predisposition risk specifically has not been tested.

Whether ADHD is fundamentally “developmental” (structural wiring) or reflects chronic catecholamine insufficiency from birth remains an open question. The Dunedin birth cohort study ($n = 1{,}037$; 95% retention to age 38) found that 90% of adult ADHD cases had no childhood ADHD diagnosis, and childhood and adult ADHD populations were “virtually non-overlapping sets” with distinct neuropsychological and genetic profiles (Moffitt et al. 2015). That reading has since been qualified by later evidence, though the case rests on a small research cluster rather than settled consensus: repeated comprehensive assessments of a comparison group (MTA local normative comparison cohort, $n = 239$) followed from ages 10 to 25 found that roughly 95% of those who screened positive were excluded on careful assessment, with the remainder being adolescent-onset or explained by substance use or comorbidity rather than de novo adult-onset (Sibley et al. 2018); a dedicated review concluded that most late-onset cases develop between ages 12–16, rarely in the complete absence of childhood precursors, while cautioning that the data cannot yet clarify whether early and late onset share underlying neural mechanisms (Asherson and Agnew-Blais 2019); and retrospective recall of childhood ADHD symptoms is itself unreliable (accuracy about 55%), so some reported adult onset reflects misrecalled childhood symptoms (Breda et al. 2020). Taken together, evidence to date favors a mix — much apparent adult onset is adolescent-onset or missed childhood ADHD, not a clearly distinct adult-onset syndrome — but this rests largely on one US comparison cohort and a self-referential research cluster, so the question remains open. The interpretation most consistent with this evidence is that ADHD functions mainly as a childhood- and adolescent-onset predisposing trait; the “acquired catecholamine insufficiency in adulthood” reading is a residual that the current evidence does not clearly support, not an established alternative.

The epidemiological data are consistent with the metabolic reserve hypothesis. In the ALSPAC birth cohort ($n = 4{,}563$), ADHD traits at age 9 conferred OR$= 2.18$ (95% CI 1.33–3.56) for chronic disabling fatigue at age 18 . Critically, this relationship was partially mediated by IL-6 at age 9, suggesting that pre-existing low-grade neuroinflammation — itself an energy drain — serves as the mechanistic bridge. Among adult CFS patients, 29.7% had childhood ADHD and 20.9% had persistent adult ADHD; those with comorbid ADHD had earlier CFS onset and worse prognosis .

The stimulant response pattern in ME/CFS is particularly informative. A patient-reported outcomes survey ($N = 3{,}925$) found that ADHD stimulants improved brain fog in 77.1% and fatigue perception in 71.7% of ME/CFS respondents, but had a net $-1.5%$ effect on PEM itself . This dissociation is precisely what the metabolic reserve hypothesis predicts: stimulants boost catecholamine availability at the prefrontal synapse (patching the energy-starved executive system) without addressing the systemic metabolic crisis that produces post-exertional malaise.

20.8.2 Autism Spectrum Disorder: Mitochondrial Baseline Deficit

The metabolic reserve case for ASD rests on stronger biological evidence than for ADHD. A meta-analysis of 204 studies found systemic mitochondrial dysfunction in ASD: elevated lactate (17%), pyruvate (41%), alanine (15%), and creatine kinase (9%), with significant ATP deficit and elevated lactate:pyruvate ratio at moderate-to-large effect sizes ($d >= 0.6$) . These are the same electron transport chain impairments documented in ME/CFS energy metabolism studies (see Ch.~6). In ASD, they represent a baseline feature — present before any infectious trigger — constituting a pre-existing lower metabolic floor.

The BH4 cofactor bottleneck provides a molecular link between ASD and ME/CFS. A systematic review found consistently lower BH4 levels in ASD biological samples compared to controls . Since BH4 is the essential cofactor for tyrosine hydroxylase (dopamine/NE synthesis), tryptophan hydroxylase (serotonin synthesis), and all three nitric oxide synthase isoforms, low BH4 simultaneously impairs neurotransmitter production and vascular autoregulation — the same pathways affected in ME/CFS orthostatic intolerance. GCH1 rs841, the variant encoding the rate-limiting enzyme for BH4 synthesis, is homozygous in approximately 4% of the population and has been associated with both ADHD and ASD traits . This represents a high-frequency shared genetic vulnerability linking both neurodevelopmental conditions to ME/CFS susceptibility through a single enzymatic deficit.

Autistic traits at age 7 independently predicted chronic disabling fatigue at 18 (OR$= 1.78$, 95% CI 1.17–2.72), with IL-6 mediating the pathway . Among autistic adults, 60% score at or above clinical cutoff for central sensitivity syndrome symptoms and 21% carry a formal ME/CFS or fibromyalgia diagnosis .

Speculation: Neurodivergent Mitochondria: Constitutionally Lower Reserve Capacity

Certainty: 0.20. The preceding sections frame neurodivergent energy deficit primarily as excess demand — hyperperfusion requirements, catecholamine inefficiency, heightened sensory processing costs. A complementary framing inverts this: neurodivergent brains may carry genetically lower-capacity mitochondria, not merely higher demand. The same alleles producing beneficial cognitive traits — rapid pattern recognition, hyperfocused attention, sensory acuity — may be pleiotropically linked to mitochondrial variants that trade coupling efficiency for membrane flexibility or rapid remodeling capacity. On this view, the reduced metabolic reserve is constitutional, encoded in the mitochondrial (and nuclear-encoded mitochondrial) genome, and present from birth.

The evidence base, while not yet definitive, is convergent. The Frye 2024 meta-analysis of 204 studies establishes that ASD involves systemic, not merely CNS, mitochondrial dysfunction: elevated lactate, pyruvate, alanine, and creatine kinase with significant ATP deficit and elevated lactate:pyruvate ratio at moderate-to-large effect sizes . Systemic involvement is the key qualifier — it means the deficit is not an artifact of brain energy demand but a body-wide feature, consistent with a constitutive mitochondrial property. For ADHD, Almutairi 2024 documents variants in NDUFAF2 (Complex I assembly factor), UCP2 (mitochondrial uncoupling protein governing proton leak and thermogenesis), and broader mitochondrial organization genes, identified primarily in preclinical models but with convergent human genetic evidence . These are not secondary consequences of ADHD behavior; they are upstream genetic variants in the core machinery of ATP production.

The formal implication for Architecture C is precise: if $J_\text{production,max}$ — the ceiling of mitochondrial ATP output — is genetically set lower in neurodivergent individuals, then identical environmental insults (infection, immune activation, sustained exertion) produce lower $R_\text{headroom}$ at every demand level. The decompensation threshold is reached sooner, and recovery from each near-threshold episode is slower.

An evolutionary framing — speculative, and flagged as such — may explain why such variants persist at high frequency. Mitochondria with lower coupling efficiency but higher membrane flexibility and faster fission/fusion cycling might support the rapid synaptic remodeling that characterizes neurodivergent cognition: exceptional capacity in specific processing domains, hyperfocused attention, and unusual cross-domain associative thinking. The metabolic trade-off would be reduced systemic buffer for challenges that demand sustained whole-body energy mobilization. This remains entirely inferential; no study has tested mitochondrial membrane dynamics in relation to cognitive phenotype.

Falsifiable prediction: Isolated mitochondria from neurodivergent versus neurotypical skeletal muscle biopsies — controlling for aerobic fitness, age, sex, and body composition — should show lower maximal respiratory capacity (state 3 respiration) and lower spare respiratory capacity (the difference between maximal and basal oxygen consumption). If Complex I-driven respiration is disproportionately affected, consistent with NDUFAF2 variants, this should be detectable as altered Complex I/II ratios in substrate-specific respirometry. If mitochondrial function is identical when confounders are adequately controlled, the genetic constitutional hypothesis fails and the demand-side framing is sufficient.

Not yet replicated. No study has directly compared isolated mitochondrial function in neurodivergent versus neurotypical individuals controlling for fitness, age, sex, and body composition. The Frye 2024 and Almutairi 2024 findings are consistent with but do not prove the constitutional framing; they do not distinguish genetically lower capacity from acquired dysfunction secondary to neurodivergent lifestyle and metabolic patterns.

20.8.3 The Neurodivergent–Hypermobility–Dysautonomia Cluster: Compounded Risk

The highest-risk phenotype for ME/CFS may be the intersection of neurodivergence and connective tissue hypermobility, because this combination creates two independent metabolic reserve reductions simultaneously.

A meta-analysis of 20 studies found autistic individuals are 7.4 times more likely to have Ehlers-Danlos syndrome than comparison groups, with 31% showing clinically assessed joint hypermobility . In children with hEDS or HSD ($n = 201$), ADHD was present in 16% overall but 46% of the 17–18 age group; ASD was present in 6% . Crucially, joint hypermobility statistically mediates the association between neurodivergence and dysautonomia: in 109 neurodivergent adults, 51% had generalized joint hypermobility (vs. 20% general population), and hypermobility mediated both dysautonomia and pain .

The proposed mechanistic chain — connective tissue laxity → impaired venous return → orthostatic cerebral hypoperfusion → chronic energy delivery deficit to the brain — is consistent with the mediation analysis showing hypermobility mediates the neurodivergence-dysautonomia association , though the specific biophysical steps remain inferential. If this chain operates, it would add to the mitochondrial and catecholamine deficits already present from the neurodevelopmental condition itself, producing a compound lower baseline: mitochondrial dysfunction (reduced ATP production) plus cerebral hypoperfusion (reduced ATP delivery). When an immune trigger strikes this already marginal system, the threshold for decompensation into sustained ME/CFS is substantially lower.

20.8.4 Migraine: Cumulative Mitochondrial Depletion

Migraine fits the metabolic reserve model through a different mechanism: cumulative depletion rather than baseline deficit. A Taiwan national cohort ($n = 6{,}902$ migraineurs vs. $n = 27{,}608$ controls) found migraine conferred a 1.5-fold elevated risk of CFS (IRR$\approx 1.5$), with a dose-response relationship (risk scaled with migraine frequency) and age amplification (IRR$= 2.11$ in those $>= 65$) . Migraineurs show elevated blood lactate, decreased activities of NADH dehydrogenase, citrate synthase, and cytochrome c oxidase, and many triggers that provoke migraine attacks (stress, sleep deprivation, fasting, exercise) are also reported as ME/CFS relapse triggers — a convergence consistent with shared energy metabolism vulnerability, though the overlap has not been systematically quantified. Each cortical spreading depression event creates a massive transient energy demand that mitochondrially impaired cortex may not efficiently recover from . If this recovery deficit is cumulative, repeated CSD events could progressively deplete neuronal energy reserves over years — a speculative but mechanistically coherent explanation for the age-dependent amplification of migraine-CFS risk observed by Lau et al.

20.8.5 Neurodivergence as Post-Infectious Risk Factor: Emerging Direct Evidence

The most direct test of the metabolic reserve hypothesis comes from post-COVID studies. In a cross-sectional study of 267 healthcare workers, higher autistic trait scores — specifically the sensory reactivity subscale — predicted COVID-19 symptoms lasting longer than 12 weeks, independent of formal autism diagnosis . This is consistent with central sensitization (a form of neural hyperexcitability that consumes extra metabolic resources) leaving less buffer for post-viral recovery. The dimensional finding (autistic traits, not just diagnosis) is important because it suggests the predisposition operates on a spectrum: any degree of neurodivergent biology that increases baseline neural energy consumption narrows the margin for absorbing an immune insult. This dimensional model is genetically validated: the same DNA variants that produce clinical ADHD and ASD also influence milder, continuous trait measures in the general population — meaning even people without a formal diagnosis carry some degree of the same genetic loading (Taylor et al. 2019). Furthermore, the genetic overlap between ADHD and ASD is not uniform across symptom types: specific dimensions (such as hyperactive-impulsive behaviors and repetitive/restricted behaviors) share the strongest genetic signal ($r_\text{g} = 0.56$), while other combinations share weaker signals ($r_\text{g} = 0.33$) (Ghirardi et al. 2019). This means that composite trait burden — the total load of subthreshold ADHD and ASD features a person carries, not just whether they meet formal criteria for both — determines how much metabolic reserve is genetically reduced. The practical implication is that screening for ME/CFS risk in neurodivergent populations should use dimensional questionnaires that capture subthreshold traits, not just binary diagnostic checklists. A person with moderate ADHD traits and mild ASD features — falling short of formal diagnosis for either — may carry the same metabolic reserve reduction as someone with a single formal diagnosis. In the clinic, ferritin and BH4 status may be more informative risk markers than whether a patient has received a formal ADHD or ASD label.

Hypothesis: Metabolic Reserve Depletion as ME/CFS Predisposition (Architecture C)

Certainty: 0.50. Pre-existing conditions that chronically reduce brain metabolic reserve — through mitochondrial dysfunction (ASD), catecholamine inefficiency (ADHD), cerebral hypoperfusion (hEDS/POTS), cumulative energy depletion (migraine), or shared BH4 cofactor deficiency (GCH1 variants) — create a lower baseline from which immune-triggered energy demands cannot be met, precipitating sustained decompensation into ME/CFS. The 0.50 certainty is justified over the alternative 0.35 specifically by the mechanistic specificity of Architecture C relative to Architecture A: while both predict the same epidemiological pattern (elevated neurodivergent comorbidity), Architecture C additionally predicts a dose-response (more reducers → higher risk), a biomarker signature (pre-illness metabolic markers predict risk), and a modifiable target — three predictions Architecture A does not make and that provide falsifiable traction beyond simple comorbidity. If these mechanistic predictions fail, the certainty should be reduced accordingly. The hypothesis makes several testable predictions:

Epidemiological: ADHD and ASD patients should show higher ME/CFS incidence after documented EBV or COVID-19 infection than non-neurodivergent controls, after controlling for sex, age, socioeconomic status, and pre-existing autoimmune conditions. This is testable now using ADHD diagnosis registries cross-referenced with post-COVID ME/CFS diagnoses.
Dose-response: Among neurodivergent individuals, those with more metabolic reserve reducers (e.g., ASD + hEDS + iron deficiency) should have higher ME/CFS incidence than those with only one. The compound-risk prediction is testable in existing hEDS registries.
Biomarker: Pre-illness metabolic markers (FDG-PET glucose metabolism, MRS lactate, BH4 levels, ferritin, cerebral blood flow by ASL-MRI) should predict post-infectious ME/CFS risk. This requires prospective post-infection cohorts with baseline metabolic phenotyping.
Interventional: Correcting modifiable reserve reducers (iron supplementation, BH4 supplementation for GCH1 carriers, optimizing cerebral perfusion in POTS) should reduce ME/CFS risk after infection or improve ME/CFS outcomes in neurodivergent patients. This is the most clinically actionable prediction.

Falsification: If ADHD/ASD patients show equal or lower ME/CFS incidence after matched infectious triggers, or if pre-illness metabolic markers do not predict post-infectious fatigue trajectory, the metabolic reserve hypothesis is not supported. The existing ALSPAC data (OR$= 2.18$ for ADHD, OR$= 1.78$ for ASD) are consistent with but do not prove this hypothesis because “chronic disabling fatigue” at 18 may not meet ME/CFS criteria.

Not yet replicated. No study has directly tested this hypothesis by prospectively measuring metabolic reserve before infection and tracking ME/CFS outcomes.

The clinical manifestation of reduced metabolic reserve before formal ME/CFS diagnosis is described in [NO SECTION]. Retrospective prodromal signs — the overtrained-but-unfit athlete, the late-game cognitive fader, the always-tired high performer, unexplained fine tremor, cold extremities, and unrefreshing sleep — represent the observable face of the lower $R_\text{headroom}$ predicted by Architecture C. German claims data confirm elevated medical utilisation (fatigue, pain, somatoform, cognitive codes) up to five years before pediatric ME/CFS diagnosis , and prospective birth cohort data show sleep disruption 4–7 years before onset . These findings are consistent with a population operating near its metabolic ceiling for years before an immune trigger pushes production below the minimum needed for daily function.

20.8.6 The Unifying Energy Thread

What connects ADHD, ASD, hEDS/POTS, and migraine as ME/CFS predisposing conditions is not a single gene or pathway but a convergent functional state: insufficient energy production, delivery, or utilisation in the brain. Each condition achieves this state through a different mechanism:

Pre-existing conditions as metabolic reserve reducers. Each condition creates energy deficit through distinct mechanisms, but the convergent functional state — reduced brain metabolic reserve — creates a common predisposition to ME/CFS following immune triggers.
Condition	Mechanism(s) reducing metabolic reserve	Key evidence	ME/CFS risk data
ADHD	↓ prefrontal glucose metabolism; ↓ D1R + ↑ microglia; ↓ dopaminergic efficiency → ↑ neural effort per task; iron deficiency → ↓ complex I/II + ↓ dopamine synthesis	Zametkin 1990 (PET); Yokokura 2021 (dual-PET); Berthier 2025 (CBF review)	OR = 2.18 for CDF at 18 (ALSPAC); 29.7% childhood ADHD in CFS
ASD	Systemic mitochondrial ETC dysfunction (↑ lactate, ↓ ATP); ↓ BH4 → ↓ dopamine/serotonin/NO; chronic IL-6 elevation	Frye 2024 (204-study meta-analysis); ColpaniFilho 2025 (BH4 systematic review)	OR = 1.78 for CDF at 18 (ALSPAC); 21% carry ME/CFS or fibro diagnosis
hEDS / POTS	Connective tissue laxity → venous pooling → ↓ cerebral perfusion; chronic sympathetic activation → ATP drain	Csecs 2022 (mediation); BaezaVelasco 2025 (7.4× EDS in ASD)	~60% POTS prevalence within ME/CFS (cross-sectional; reverse risk untested)
ASD + hEDS	Compound: mitochondrial dysfunction + cerebral hypoperfusion (dual mechanism)	BaezaVelasco 2025; Kindgren 2021 (46% ADHD in hEDS age 17–18)	Highest-risk subgroup (untested directly)
ADHD + ASD	Compound: twice-hit BH4 bottleneck (dopaminergic demand + IL-6 oxidative destruction); double catecholamine inefficiency; ADHD and ASD share roughly 10,000 specific genetic risk variants, far more than either shares with mood disorders (Hindley et al. 2022)	Lai 2019 (28% ADHD in ASD) (Lai et al. 2019); Ghirardi 2018 (ASD elevates ADHD odds 22-fold) (Ghirardi et al. 2018); Polderman 2014 (54–60% shared genetic influence) (Polderman et al. 2014); Hindley 2022 (thousands of shared risk variants) (Hindley et al. 2022)	Untested directly; predicted to be highest-risk neurodevelopmental subgroup — compound BH4 depletion through both consumption (ADHD) and destruction (ASD) routes. Clinically: dual-diagnosis patients merit proactive iron and BH4 status screening even before ME/CFS onset.
Migraine	Cumulative CSD → progressive mitochondrial depletion; ↓ cytochrome c oxidase; shared triggers with ME/CFS	Lau 2015 (national cohort); Wang 2023 (mito review)	IRR ≈ 1.5 for CFS; dose-response
GCH1 rs841 carriers	↓ BH4 → ↓ NO → ↓ vascular autoregulation; ↓ dopamine/serotonin synthesis (~4% homozygous)	Williams 2025 (case series, n = 5); ColpaniFilho 2025	Untested; high-frequency variant
Iron deficiency	↓ Complex I/II → ↓ ATP; ↓ TH activity → ↓ dopamine. Modifiable.	DelRosso 2026 (comparative review)	Untested directly; modifiable

If Architecture~C is correct, ADHD, ASD, migraine, hEDS, and POTS should not be treated merely as comorbidities in ME/CFS but as predisposing conditions whose pre-existing energy deficits contributed to the development of ME/CFS. Architecture~C predicts — but no study has yet tested — that identifying and correcting modifiable components of the metabolic reserve deficit (iron status, BH4 cofactor support, cerebral perfusion optimization) could improve outcomes even after ME/CFS is established. This prediction requires validation through the interventional studies described below.

Limitation: Architecture C Shares Epidemiological Predictions with Architecture A

Architecture C (metabolic reserve deficit predisposes to ME/CFS) makes the same epidemiological prediction as Architecture A (shared genetic vulnerability): elevated comorbidity of ADHD/ASD with ME/CFS. The architectures differ in their mechanistic predictions — Architecture C predicts that pre-illness metabolic markers should predict post-infectious ME/CFS risk, and that correcting metabolic deficits should reduce that risk — but no study has tested these mechanistic predictions directly. The ALSPAC IL-6 mediation finding is more consistent with Architecture C (inflammation as the mechanism reducing reserve) than with a non-mechanistic Architecture A, but does not definitively distinguish them. All three architectures (A, B, and C) likely coexist in the patient population: some patients have shared genetic vulnerability without specific metabolic deficit (A), some develop acquired ADHD/ASD features from ME/CFS (B), and some had pre-existing metabolic fragility that predisposed them to ME/CFS (C).

Limitation: Speculative Elaboration Exceeds Current Evidence

The volume of downstream elaboration from Architecture C — mathematical formalizations, treatment protocols, diagnostic scores, and proposed studies — substantially exceeds what the current evidence base supports. The founding empirical anchors are limited: a single 36-year-old PET study for ADHD glucose hypometabolism (with mixed subsequent replications), a 5-patient case series for BH4 (Williams 2025), and epidemiological associations that cannot distinguish Architecture C from Architecture A (shared genetic vulnerability). The elaboration is presented not as validated clinical guidance but as a structured exploration of what the hypothesis predicts IF correct — generating falsifiable predictions that can be tested efficiently. Readers should interpret the treatment protocols, mathematical models, and diagnostic proposals as research-stage hypothesis development, not as established clinical frameworks. The certainty ratings (0.15–0.50) throughout reflect this preliminary status.

Open Question: ADHD Hyperfocus Crash as Micro-PEM: Same Mechanism at Different Severity?

ADHD communities describe “crashing after hyperfocus” — exhaustion, brain fog, irritability, and hours needed to recover from sustained cognitive effort. If ADHD represents a brain energy deficit, hyperfocus depletes a limited cognitive energy budget just as physical exertion depletes the limited energy budget in ME/CFS. The phenomenological parallel is striking: both involve an activity that feels sustainable in the moment, followed by disproportionate recovery cost. If this analogy holds, ADHD patients are already experiencing micro-PEM — energy envelope depletion at a subclinical scale. This reframes ME/CFS not as a categorically distinct condition but as the severe end of a spectrum that begins with the energy management challenges already familiar to ADHD patients.

Architecture C provides quantitative grounding: ADHD is associated with 8.1% lower global cerebral glucose metabolism and frontal hypoperfusion . If healthy $R_\text{headroom} \approx 0.3$–$ 0.5$, then constitutive ADHD metabolic deficits compress prefrontal $R_\text{headroom}$ to $\approx 0.15$–$ 0.25$ at baseline — already within the range Architecture C identifies as fragile. Hyperfocus represents maximal sustained prefrontal demand: when $J_\text{demand,peak}$ approaches $J_\text{production,max}$ in prefrontal circuits, the same ROS-mediated damage cascade hypothesized for systemic PEM may trigger focally. The post-hyperfocus crash is then the recovery phase from exceeding local $R_\text{crit}$ — a prefrontally distributed, cognitively triggered micro-PEM rather than a systemic, exertion-triggered one.

Testable predictions: (1) Actigraphy and continuous glucose monitoring during ADHD hyperfocus should show disproportionate cognitive fatigue scaling nonlinearly with session duration. (2) Sustained cognitive effort should produce elevated blood lactate in ADHD subjects relative to neurotypical controls matched for effort intensity. (3) Both signatures should be dramatically more pronounced in ADHD patients who subsequently develop ME/CFS. (4) Do ADHD patients’ post-hyperfocus recovery times correlate with metabolic reserve markers? Does the crash pattern worsen after infection (partial decompensation)?

Speculation: Focal PEM: Threshold Phenomenon Independent of Anatomical Distribution

Certainty: 0.25. PEM is currently understood as a systemic response to physical exertion exceeding a whole-body energy threshold. The ADHD hyperfocus analogy suggests a generalization: PEM is a threshold phenomenon that can occur focally, triggered whenever local metabolic demand exceeds local $R_\text{crit}$ in any high-demand tissue. Under this model, systemic ME/CFS PEM and ADHD post-hyperfocus crash are instances of the same mechanism differentiated only by the anatomical distribution (whole-body vs. prefrontal) and the triggering domain (physical vs. cognitive). The threshold in each case is determined by local metabolic reserve — which is why low-reserve individuals (ADHD, prior infection, iron deficiency) hit it first and hardest. This framing predicts that sufficiently severe cognitive exertion in ME/CFS patients should produce PEM even without physical activity, and that prefrontal metabolic markers should predict cognitive PEM threshold independently of physical PEM threshold. Both predictions are testable with existing neuroimaging and actigraphy technology but have not been formally studied.

20.8.7 The Two-Hit Threshold: Why Some Post-Infectious Patients Develop ME/CFS and Others Recover

Architecture C provides a framework for the central unsolved question in post-infectious illness: why do most patients recover from EBV, SARS-CoV-2, or influenza while a minority develop persistent fatigue meeting ME/CFS criteria? If metabolic reserve determines the outcome, then the same infection produces a binary result based on pre-illness headroom. In high-reserve individuals, the metabolic cost of the immune response (cytokine production, lymphocyte proliferation, neuroinflammation) remains within the system’s capacity to sustain repair processes simultaneously — transient fatigue is followed by recovery and eventual return to baseline. In low-reserve individuals, the immune response may push the system below a critical threshold where mitochondrial damage from reactive oxygen species exceeds repair capacity — potentially initiating the kind of self-sustaining feed-forward cycle that has been proposed to characterize ME/CFS (see speculation below).

Speculation: Two-Hit Threshold Model for ME/CFS Onset

Certainty: 0.40. The transition from post-infectious fatigue to ME/CFS occurs when an acute immune insult pushes an individual below a critical metabolic reserve threshold ($R_\text{crit}$), triggering feed-forward mitochondrial damage that does not spontaneously reverse when the infection resolves. Pre-existing metabolic reserve determines who crosses $R_\text{crit}$. The model exhibits hysteresis: reversing the infection does not reverse ME/CFS, because the damage cycle is self-sustaining once initiated. Testable: Pre-COVID metabolic markers (FDG-PET, BH4, ferritin, VO₂max) in biobank samples should predict Long COVID → ME/CFS transition. Neurodivergent Long COVID patients should show higher ME/CFS conversion rates than neurotypical patients matched for infection severity. The Raw 2025 finding (autistic traits predict prolonged COVID symptoms, $n = 267$) is a preliminary directional signal .

Prediction: Pre-Pandemic Biobank Samples Can Test the Two-Hit Model Now

Several large cohorts — UK Biobank, ALSPAC, RECOVER, and DecodeME — hold pre-pandemic metabolic markers (ferritin, BH4 proxies, VO₂max, neurodevelopmental diagnoses), COVID infection records, and Long COVID outcome data in the same participants. The compound-risk prediction is therefore testable without new data collection: individuals carrying multiple low-reserve markers before infection (neurodivergent diagnosis + low ferritin + orthostatic intolerance history) should show substantially higher ME/CFS conversion rates than those with none, even after matching for infection severity. A specific cross-diagnostic prediction follows: neurodivergent Long COVID patients should convert to ME/CFS at higher rates than neurotypical Long COVID patients matched for viral load and acute illness severity — a directional signal already supported by the Raw 2025 autistic-trait finding ($n = 267$) . If biobank reanalysis finds no metabolic-reserve gradient in pre-illness markers across outcome groups, the two-hit threshold model is falsified.

20.8.8 BH4 as Convergent Metabolic Bottleneck

The observation that multiple predisposing conditions converge on a single cofactor is noteworthy. Tetrahydrobiopterin (BH4) is simultaneously required for dopamine synthesis (ADHD relevance), serotonin synthesis (depression, sleep), nitric oxide synthesis (vascular tone, cerebral perfusion), and phenylalanine metabolism (energy substrates). A single cofactor deficit thus impairs neurotransmitter efficiency, vascular delivery, and metabolic substrate availability simultaneously . The convergence is structurally supported by the shared genetic architecture of ADHD and ASD: large genome-wide studies find that 32–41% of the common genetic variants contributing to one condition also contribute to the other (Demontis et al. 2023) (Grove et al. 2019), and across psychiatric disorders, ADHD and ASD genetically cluster as a single neurodevelopmental group (Cross-Disorder Group of the Psychiatric Genomics Consortium 2013). These shared variants do not just elevate risk for both conditions — they plausibly converge on the same metabolic pathways, with BH4 synthesis and recycling among the strongest candidates given its simultaneous role in dopamine, serotonin, and nitric oxide biology. If this convergence is correct, then interventions that increase BH4 availability or support its recycling — such as folinic acid (to fuel the recycling enzyme dihydrofolate reductase) or vitamin C (which protects BH4 from oxidation) — may be disproportionately beneficial in ADHD+ASD patients because they address a single bottleneck being hit from two independent directions. This is a testable clinical prediction: ADHD+ASD individuals should show lower baseline BH4 (measured via urinary neopterin:biopterin ratio) than individuals with either condition alone, and BH4-supportive interventions should produce larger symptomatic improvements in the dual-diagnosis group. Multiple predisposing conditions may deplete BH4 through different mechanisms : high dopaminergic demand (ADHD) increases BH4 consumption through tyrosine hydroxylase turnover; chronic IL-6 elevation (ASD) drives oxidative BH4 destruction ; oxidative stress from any source (including ischemia-reperfusion in orthostatic hypoperfusion) generates peroxynitrite that oxidizes BH4 to dihydrobiopterin (BH2) ; iron deficiency impairs BH4 recycling via dihydropteridine reductase ; GCH1 rs841 variants reduce BH4 production constitutionally ; and post-infectious interferon-gamma surges divert GTP cyclohydrolase~I activity toward neopterin production at the expense of BH4 .

A further complexity: BH4 has different $K_m$ values for its target enzymes — partial depletion may disproportionately affect nitric oxide synthase (higher $K_m$) before tyrosine hydroxylase (lower $K_m$), meaning the BH4 deficit may manifest as vascular dysfunction before catecholamine deficiency. The convergent bottleneck may not affect all downstream pathways equally.

Speculation: BH4 Depletion as the Convergent Bottleneck Linking Predisposing Conditions to ME/CFS

Certainty: 0.35. BH4 depletion may be the single metabolic chokepoint through which ADHD, ASD, hEDS/POTS, iron deficiency, and GCH1 variants all reduce metabolic reserve. If so, urinary neopterin:biopterin ratio — a non-invasive marker of immune activation relative to BH4 status (~$30–50 per sample) — should be elevated across ALL predisposing conditions, not just post-infection. Pre-illness BH4 status should predict post-infectious ME/CFS risk better than any single predisposing diagnosis alone. Falsification: If BH4 levels are normal in ADHD and hEDS patients (only depleted in ASD and post-infection), the convergent bottleneck hypothesis fails and condition-specific mechanisms dominate. A critical evidence gap: BH4 depletion has been directly measured only in ASD populations . The extrapolation to ADHD (via dopaminergic demand), hEDS (via ischemia-reperfusion ROS), migraine, and iron deficiency is mechanistically inferred but not empirically confirmed — each pathway requires direct BH4 measurement in the relevant population.

20.8.9 Why Stimulants Help Symptoms but Not PEM

The dissociation between stimulant benefit for brain fog (77.1%) and negligible effect on PEM ($-1.5%$) is precisely what the metabolic reserve model predicts. Stimulants increase catecholamine availability at the synapse, improving dopaminergic efficiency — the metabolic cost per unit of cognitive output decreases. This is equivalent to increasing cognitive headroom by reducing demand rather than increasing production capacity. However, stimulants do not augment mitochondrial maximum capacity. The improved subjective function enables patients to sustain higher cognitive workloads, potentially increasing total energy expenditure and bringing them closer to the PEM threshold. The net $-1.5%$ PEM effect in the Vernon survey may represent the average of two opposing behavioral responses: patients who use the cognitive improvement to do more (increased PEM risk) and patients who maintain activity levels (reduced PEM risk from lower metabolic cost per task). This predicts that stimulant-treated ME/CFS patients who maintain strict activity pacing should experience lower PEM frequency, while those who increase activity should experience higher PEM frequency — a testable prediction that current trial designs do not capture.

(Note: The Vernon 2025 survey data and the Lau 2015 migraine-CFS association were published before Architecture C was formulated. Their consistency with the model is post-hoc — the model was partly constructed to accommodate these observations. The genuinely prospective predictions of Architecture C are the biobank test (Pre-Pandemic Biobank Samples Can Test the Two-Hit Model Now), the PBMC gradient (PBMC Spare Respiratory Capacity Gradient Across Neurodivergent and ME/CFS Groups), and the stimulant-pacing interaction (Stimulant + Pacing Interaction Determines PEM Outcome), none of which have been tested.)

The stimulant response pattern is consistent with the metabolic reserve model but is not uniquely predicted by it — the same dissociation (cognitive improvement without PEM reduction) would be expected from any model in which stimulants mask subjective fatigue without altering underlying energy pathology.

Prediction: Stimulant + Pacing Interaction Determines PEM Outcome

In neurodivergent ME/CFS patients, stimulants have a dual mechanistic role that the survey cannot disentangle. Their reserve-increasing effect operates through dopaminergic efficiency: reducing the metabolic cost of baseline cognitive function effectively increases $R_\text{headroom}$ for cognitive tasks without augmenting mitochondrial capacity. Their reserve-depleting risk runs in the opposite direction — subjective improvement enables exceeding the physical energy envelope, increasing total expenditure and PEM exposure. Methylphenidate may additionally provide direct mitochondrial benefit by maintaining Parkin-mediated mitophagy and reducing reactive oxygen species , a pathway absent from amphetamine-class stimulants. The specific prediction follows: a trial stratifying stimulant-treated ME/CFS patients by pacing adherence should find that strict pacers show lower PEM frequency than untreated patients, while activity-increasers show higher PEM frequency; the net $-1.5%$ in is the population average of these opposing effects. Current trial designs that measure only mean PEM change will always find near-zero net effects, masking a clinically important interaction that may explain why stimulant benefit appears confined to cognitive symptoms in aggregate survey data.

20.8.10 Progressive Reserve Erosion with Age

Neurodivergent individuals may not just start with lower reserve but experience faster erosion with age. Baseline neural inefficiency produces chronic low-grade oxidative stress that accumulates mitochondrial damage over decades — a process invisible in standard clinical monitoring but potentially detectable through longitudinal metabolic imaging.

Speculation: Progressive Metabolic Reserve Erosion in Neurodivergent Ageing

Certainty: 0.20. Even sub-threshold ROS exposure, if sustained, can slowly degrade mitochondrial function. In neurodivergent brains operating closer to their metabolic ceiling during routine cognition, ROS levels may be chronically slightly elevated — not enough to trigger acute PEM, but enough to accelerate mitochondrial ageing.

This framework explains three converging observations: (1) found that migraine-CFS risk increases with age (IRR 2.11 in the age ≥ 65 cohort versus 1.5 overall), consistent with cumulative erosion reaching a pathological threshold; (2) increasing ADHD functional impairment in middle age is often reported despite stable symptom scores, suggesting a shrinking $R_\text{headroom}$ rather than worsening core symptoms; (3) neurodivergent ME/CFS patients frequently describe gradual decline in baseline function across years before any acute precipitating event, as though $R_\text{crit}$ was being approached incrementally.

Falsifiable predictions: Longitudinal FDG-PET in ADHD adults should show faster decline in prefrontal glucose metabolism than age-matched neurotypical controls. Mitochondrial DNA mutation burden, assessed via long-range PCR or mtDNA sequencing, should be elevated in neurodivergent adults relative to neurotypical controls at matched chronological ages.

Limitations: Entirely speculative. No longitudinal metabolic imaging in ADHD adults exists. The mtDNA prediction is testable with existing biobank samples.

Replication status: Not yet replicated — no study has examined age-related metabolic decline specifically in neurodivergent populations.

20.8.11 Mathematical Formalization: Individual Reserve Capacity

The following formalization is illustrative rather than calibrated — the specific parameter values are order-of-magnitude estimates derived from population studies, not from direct measurement of metabolic reserve. The mathematical structure captures the qualitative prediction (compound conditions produce multiplicative, not additive, reserve reduction) and is intended to generate testable predictions rather than to compute precise individual risk.

The metabolic reserve model can be formalized to incorporate individual variation in maximum production capacity. Rather than treating $J_\text{production,max}$ as a population parameter, Architecture C proposes:

\[ J_\text{production,max} (\text{individual}) = J_\text{production,max} (\text{population}) dot product_{i} (1 - \delta_i) \]

where $\delta_i$ represents the fractional reserve reduction from each predisposing condition:

Condition	Estimated $\delta_i$
ADHD	~0.08 (8.1% reduced glucose metabolism, ; note: mixed replication — Zametkin’s own 1993 adolescent follow-up did not replicate global hypometabolism; original study criticized for sex confounders; treat as approximate magnitude estimate)
ASD	~0.10–0.15 (ETC dysfunction magnitude, )
hEDS/POTS	~0.05–0.10 (cerebral hypoperfusion magnitude)
Migraine	~0.03–0.05 per decade of active migraine (cumulative CSD damage, )
Iron deficiency	0.00–0.15 (variable, depends on ferritin level, )
GCH1 rs841 homozygous	~0.05–0.10 ()

The multiplicative model predicts compound conditions produce MORE than additive reserve reduction. Example: a patient with ASD ($\delta = 0.12$) + hEDS ($\delta = 0.08$) + iron deficiency ($\delta = 0.10$) would have $J_\text{production,max} = J_\text{base} dot 0.88 dot 0.92 dot 0.90 = J_\text{base} dot 0.729$ — a 27% reduction before any infection. The combined ADHD+ASD case is particularly consequential because, at 28% ADHD prevalence within ASD (Lai et al. 2019), roughly 1 in 3–4 autistic individuals may carry this dual burden. ADHD+ASD yields $J_\text{production,max} = J_\text{base} dot 0.92 dot 0.88 = J_\text{base} dot 0.810$ — a 19% reduction before any additional co-occurrences. The BH4 bottleneck is hit from both sides simultaneously: dopaminergic demand driving consumption and IL-6-driven oxidative destruction. Clinically, this means a patient with both ADHD and ASD traits should be treated as having an elevated baseline risk for ME/CFS even before any triggering infection occurs. When such a patient does encounter a significant immune trigger (EBV, COVID-19, other viral illness), the clinician should have a lower threshold for monitoring post-infectious fatigue trajectory and for proactively addressing modifiable metabolic deficits — particularly iron status (target ferritin above 100 ng/mL) and BH4 cofactor support (folinic acid, vitamin C) — during the acute infection window, when interventions may prevent decompensation into sustained ME/CFS rather than merely treating established disease.

Limitation: Delta Values Are Rough Estimates Requiring Calibration

The specific $\delta_i$ values are derived from population-level studies and represent rough estimates. Individual variation is likely enormous. Calibration requires the PBMC respirometry study ([NO SECTION]). The multiplicative assumption (independence of reserve reducers) may not hold if conditions share downstream mechanisms.

20.8.12 Vulnerability Score and Phase Transition

Define a vulnerability score:

\[ V(t) = 1 - R_\text{headroom} (t) = 1 - (J_\text{\prod,max} (\text{individual}, t) - J_\text{demand} (t)) / J_\text{\prod,max} (\text{individual}, t) \]

ME/CFS onset occurs when $V(t)$ crosses a critical threshold $V_\text{crit}$ during infection:

\[ J_\text{demand} (t) = J_\text{demand,baseline} + J_\text{immune} (t) \]

where $J_\text{immune}$ represents the metabolic cost of the immune response.

The key feature is HYSTERESIS: once the system crosses $V_\text{crit}$, reversing the infection does not reverse ME/CFS because the feed-forward damage cycle (ROS → ETC damage → lower $J_\text{\prod,max}$ → more ROS) has been initiated.

Speculation: ME/CFS as Hysteretic Phase Transition

Certainty: 0.30. If the transition to ME/CFS is genuinely hysteretic — exhibiting path-dependence such that the system cannot return to the healthy attractor by simply removing the triggering infection — then treatment requires not just resolving inflammation but actively breaking the feed-forward cycle (ROS scavenging + mitochondrial biogenesis stimulation + metabolic demand reduction simultaneously). Prevention (keeping $V(t) < V_\text{crit}$ during infection) would be far more effective than treatment. Whether the dynamics are truly hysteretic vs simply slow-recovering is an empirical question testable with longitudinal metabolic imaging during and after acute infection. Not yet replicated.

20.8.13 Proposed Research Priorities

The metabolic reserve hypothesis generates several tractable research directions ordered by feasibility:

Reanalysis of existing datasets (fast, low-cost). The ALSPAC, UK Biobank, RECOVER, and DecodeME datasets contain neurodevelopmental history, metabolic markers, infection records, and fatigue outcomes. Cross-referencing these existing data can test the compound-risk prediction without new data collection.
Peripheral blood mitochondrial respirometry (moderate cost). Seahorse XF respirometry from a standard blood draw can measure spare respiratory capacity — the cellular analogue of metabolic reserve — across six groups: ADHD-only, ASD-only, ADHD+ME/CFS, ASD+ME/CFS, ME/CFS-only, and controls. If spare respiratory capacity follows the predicted gradient (controls $>$ neurodivergent-only $>$ ME/CFS-only $>$ neurodivergent+ME/CFS), the reserve model is directly supported.
Prospective post-infection metabolic phenotyping (expensive, definitive). The only design that can definitively prove or disprove Architecture~C: recruit neurodivergent and neurotypical cohorts, measure baseline metabolic parameters, and follow prospectively through viral infections. This requires substantial funding but would resolve a fundamental question in ME/CFS etiology.

Open Question: Can Modifiable Reserve Reducers Be Corrected to Prevent or Treat ME/CFS?

Architecture C identifies several modifiable components of the metabolic reserve deficit. Iron repletion (targeting ferritin $> 100$ ng/mL per clinical opinion, not trial-validated for ME/CFS) targets both mitochondrial complex~I/II function and dopamine synthesis . BH4 cofactor support — either indirect (folinic acid and vitamin C to support BH4 recycling via dihydrofolate reductase ) or direct (BH4 supplementation for confirmed GCH1 carriers, evidence limited to $n = 5$ case reports ) — addresses the proposed convergent bottleneck. Cerebral perfusion optimization through standard POTS management increases energy delivery. These interventions exist and are testable, though none has been validated specifically for ME/CFS prevention or treatment in this context. Whether pre-emptive correction of modifiable reserve reducers in neurodivergent patients reduces ME/CFS incidence after infection — or improves outcomes in established ME/CFS — is entirely untested but represents the most immediate translational prediction of Architecture~C.

Hypothesis: Connective Tissue Matrix Permissivity as the Unifying Mechanism of the hEDS-POTS-MCAS-ME/CFS Cluster

(Certainty: 0.50 – mechanistically plausible; no controlled tissue-level data in ME/CFS.)

The hEDS-POTS-MCAS triad plus ME/CFS may share an underlying structural defect: extracellular matrix (ECM) abnormalities (tenascin-X, collagen VI variants) reduce the physical barriers that normally constrain mast cell migration into perivascular and perineural niches. The same matrix laxity that causes joint hypermobility also permits mast cell tissue infiltration in the vasa nervorum, brainstem, and gut submucosa – enabling the aberrant mast cell-nerve crosstalk documented via CADM1 adhesion structures . Under this “permissive matrix” model, ME/CFS patients with hEDS features do not simply have four comorbid diseases; they have a single ECM vulnerability that expresses differently depending on which tissue is most affected.

Mechanistic pathway: Defective ECM (collagen laxity) → abnormal mast cell tissue localization and density → perivascular and perineural mast cell-mediator release (histamine, VEGF, tryptase) → autonomic nerve sensitization (POTS component) + small fiber neuropathy (ME/CFS neuroinflammation component) + connective tissue instability (hEDS component). Mast cell-derived heparin from sustained low-grade degranulation could also explain the easy bruising and coagulation anomalies reported by some patients in this triad.

Testable prediction: ME/CFS patients with Beighton score $\geq$ 5 will show higher skin biopsy mast cell density ($>$ 20 cells/HPF) than Beighton $\leq$ 2 ME/CFS patients, even when matched for serum tryptase. Skin punch biopsy is the minimal feasible test. A stronger prediction: dermal mast cells in the high-Beighton group will localize closer to nerve fibers ($<$ 20 nm proximity) than in the low-Beighton group.

Limitations: No published ME/CFS study has measured mast cell density versus Beighton score. Tenascin-X and collagen VI variants in hEDS are incompletely characterized; the ECM defect is not definitively established in most hEDS cases. This hypothesis conflates the diverse molecular basis of hEDS (multiple gene variants, undefined pathways) with a single mast cell mechanism – an oversimplification that requires tissue-level validation.

Speculation: Mast Cell Gene Expression Convergence Across ME/CFS, Chronic Lyme, and Bartonellosis

(Certainty: 0.20 — mechanistic inference from mast cell biology with stealth pathogens; no transcriptomic comparison in these three conditions has been performed. Not yet replicated.)

Persistent stealth pathogens — Borrelia burgdorferi (chronic Lyme disease) and Bartonella henselae (bartonellosis) — chronically activate mast cells via TLR2 and TLR9 ligands (lipoproteins, unmethylated CpG DNA). Bartonella has been proposed to exhibit mast cell tropism, with some evidence suggesting intracellular invasion of mast cell cytoplasm and direct phenotypic modulation, though this has not been systematically characterized in human mast cell lines. Post-treatment Lyme disease syndrome (PTLDS) and post-bartonellosis syndrome share with ME/CFS the triad of fatigue, cognitive dysfunction, and orthostatic intolerance — and all three overlap substantially with MCAS in clinical presentation.

We speculate that single-cell RNA sequencing of mast cells from skin biopsies of ME/CFS, PTLDS, and bartonellosis patients would reveal a shared “chronic-pathogen-primed” mast cell transcriptomic state — distinct from the atopic mast cell state seen in allergic disease — characterized by: upregulated TLR signaling, elevated MRGPRX2 expression, suppressed FcεRI-mediated degranulation (atopy-like activation is reduced while IgE-independent activation is enhanced), and elevated CXCL8/CXCL10 production.

This shared mast cell state would provide a common mechanistic substrate for the MCAS overlap observed across post-infectious syndromes and would suggest that mast cell stabilization strategies developed for one condition could be cross-applicable to others.

Testable predictions:

1. scRNA-seq of skin punch biopsy mast cells will identify a transcriptomic cluster present in ME/CFS, PTLDS, and bartonellosis but absent in atopic dermatitis controls, with MRGPRX2 and TLR2/TLR9 upregulation as defining markers
1. ME/CFS patients with positive Bartonella or Borrelia serology will show higher rates of biochemically confirmed MCAS (Valent 2021 criteria) than seronegative ME/CFS patients
1. The MRGPRX2-modulating treatment response pattern (poor H1/H2 response, possible bromelain response) will be more prevalent in PTLDS and bartonellosis-associated ME/CFS than in post-viral-onset ME/CFS

Limitations: scRNA-seq of mast cells from skin biopsies is technically demanding — mast cells are rare in dermis, requiring enzymatic dissociation protocols. Seronegative Lyme and Bartonella are poorly standardized. Chronic Lyme disease itself remains diagnostically controversial. Distinguishing pathogen-induced from coincidental MCAS in these patients is challenging without controlled cohorts.

Synthesis: Post-Viral Persistence: Endothelial Damage, Immune Dysregulation, and Failed Resolution

Five integration cycles address the question of why ME/CFS persists long after the initial trigger has resolved: corticosteroid mechanisms postviral, methylprednisolone failure, NETs/DNase deficiency, plasmacytoid dendritic cell dysfunction, and HIF-2α endothelial postviral activation. These converge on a temporal model in which an acute viral trigger initiates three parallel dysfunctions — (1) endothelial barrier damage via sustained HIF-2α activation (Ribeiro 2026, cert 0.55 Sustained HIF-2$\alpha$ as a Mechanistic Explanation for $\beta_2$AR-Autoantibody-Negative Endothelial Dysfunction), impairing vascular delivery and glymphatic waste clearance; (2) impaired NET clearance via DNase1L3 deficiency, creating microclot-nucleating NET remnants that physically obstruct microvasculature Defective NET Clearance Despite Diminished NET Production; and (3) pDC misdirection from blood to tissue with IFN-α dysregulation pDC Migration Hijack: Blood-to-Tissue Sequestration Model, sustaining a sterile interferonopathy that clinically mimics persistent infection. Corticosteroids worsen this picture: the PoCoVIT trial failure (methylprednisolone in Long COVID) and mechanistic analyses suggest that immune suppression impairs viral clearance, enables EBV/HHV-6 reactivation via NK suppression, and disrupts HPA axis feedback Methylprednisolone Failed in Long COVID. The convergent implication is that post-viral ME/CFS persistence is not a single-pathway failure but a multi-system lock-in state where endothelial, immune, and neuroendocrine dysfunctions mutually reinforce. Treating any one pathway in isolation (e.g., immune suppression without addressing endothelial repair) may be insufficient or harmful. The central challenge is whether the lock-in state can be reversed by sequential pathway interventions, or whether all three must be addressed simultaneously to achieve durable remission.

21 Th1-Fibrotic Disease Comparators: Lichen Sclerosus, Systemic Sclerosis, Morphea, and Vitiligo

ME/CFS and lichen sclerosus share a Th1-dominant cytokine signature (IL-7, IL-15, IFN-γ, TNF-α elevated; IL-10 reduced) and tissue-resident memory T-cell dominance without meeting classical autoimmune criteria. Two related Th1-mediated conditions provide mechanistic comparators that may generate testable ME/CFS predictions: systemic sclerosis (SSc), which adds a vascular and mitochondrial dimension, and vitiligo, which tests the TRM/IL-15 axis without fibrosis.

21.1 Systemic Sclerosis and ME/CFS

Systemic sclerosis (scleroderma; SSc) is an autoimmune fibro-vascular disease combining Th1/IFN-driven inflammation, TGF-β-mediated fibrosis, and progressive microvascular injury. Fatigue is reported by 75–89% of SSc patients and consistently ranks among the most disabling symptoms, independent of organ involvement . In the SPIN prospective registry (n=2,385), SSc fatigue severity exceeded population norms (T-score 54.6 vs. 50) and was most strongly predicted by gastrointestinal involvement and pain rather than skin fibrosis score (mRSS) . This finding shows that SSc fatigue is not simply a by-product of fibrotic burden — though GI involvement in SSc is itself a structural manifestation of the disease, not a purely functional one. The key mechanistic implication is that whatever drives GI-and-pain-linked fatigue in SSc may overlap with ME/CFS pathways more than the stereotypical “fibrosis-equals-disability” model would predict.

A subset of SSc patients satisfies formal ME/CFS criteria. Van Eeden et al. applied the Canadian Consensus Criteria to 24 early SSc patients and identified a subgroup that not only met ME/CFS criteria but also showed a distinct mitochondrial gene-expression profile: reduced cytochrome b (CyB; Complex III of the mitochondrial electron transport chain) independent of skin score, digital ulcers, lung function, or autoantibody subtype . This preliminary finding (n=12 per group, single center) implies that SSc-triggered ME/CFS may arise via Complex III suppression, not fibrotic organ burden — raising the possibility that Complex III is a convergent bottleneck across post-infectious and autoimmune-triggered ME/CFS phenotypes. Critically, PEM has not yet been systematically assessed in SSc cohorts using validated instruments; applying DSQ-PEM or 2-day CPET to SPIN sub-samples is an immediately tractable research gap .

SSc also contributes to the ME/CFS endothelial and autonomic picture. Masini et al. found cardiac autonomic neuropathy (CAN) in 50% of SSc patients (n=26, pilot), correlated with microvascular damage on nailfold capillaroscopy; anti-Scl-70 independently predicted autonomic involvement . ME/CFS literature documents HRV reduction, reduced baroreflex sensitivity, and chronotropic incompetence on CPET (Walitt et al. 2024) — all consistent with autonomic dysfunction — but formal CAN criteria (Ewing battery) have rarely been applied. Whether ME/CFS autonomic findings reflect structural neuropathy analogous to SSc-CAN, or functional dysregulation (autoantibody-mediated, deconditioning-associated), remains unresolved.

Raynaud’s phenomenon affects $>$ 90% of SSc patients and provides a vascular mechanistic bridge to ME/CFS cold sensitivity. Two large GWAS studies have now identified the primary Raynaud’s risk locus as ADRA2A (α2A-adrenoreceptor; OR=1.26, $p < 9.6 \times 10^{-27}$), with NOS3 (endothelial NO synthase) as a second independent locus . The NOS3 finding connects to documented ME/CFS endothelial dysfunction: reduced flow-mediated dilation and impaired microvascular reactivity have been reported in ME/CFS cohorts ; ADMA elevation is mechanistically plausible but not yet confirmed in large ME/CFS cohorts. A subset of ME/CFS patients with Raynaud-positive history and orthostatic intolerance may carry ADRA2A risk variants contributing mechanistically to both peripheral vasospasm and impaired baroreflex feedback.

Speculation: Systemic Sclerosis with High IFN Signature as a Closer ME/CFS Fatigue Model

Certainty: 0.30. Based on shared IFN-γ/TGF-β pathway, documented fatigue in SSc, and the LS-ME/CFS mechanistic overlap. No direct comparative study; hypothesis is cross-disease inference. Not yet replicated.

Morphea (localized scleroderma) and systemic sclerosis (SSc) share LS’s core IFN-γ→TGF-β→fibroblast activation mechanism but frequently co-occur with severe fatigue — a symptom prominent enough in SSc that some patients meet ME/CFS symptomatic criteria . Unlike LS (primarily skin/mucosa) or ME/CFS (primarily systemic/functional), SSc provides a condition where the IFN-γ→TGF-β fibrotic cascade is both measurable and linked to documented fatigue, making it a potential mechanistic bridge. Studying SSc-associated fatigue at the molecular level — IFN signature score, tissue-resident memory T-cell density, IL-15 trans-presentation activity — could yield ME/CFS-relevant findings faster and more definitively than direct ME/CFS tissue studies, given that SSc has established biopsy infrastructure and biomarker assay pipelines.

Falsifiable prediction: SSc patients with high type-I IFN signature scores (using validated IFN-stimulated gene panels) will show higher rates of ME/CFS-criteria-meeting fatigue (assessed by PEM questionnaire) than low-IFN-signature SSc patients, independent of overall SSc severity.

Limitations: SSc fatigue has multiple drivers (pulmonary involvement, pain, sleep disruption) that confound any IFN-fatigue correlation. The prediction requires SSc patients to be phenotyped for PEM specifically — not just generic fatigue — which is not standard in rheumatology settings.

Speculation: Complex III as Convergent ETC Bottleneck Across Post-Infectious and Autoimmune ME/CFS

Certainty: 0.35. Based on van Eeden 2023 CyB deficit in SSc-ME/CFS subgroup and existing ME/CFS mitochondrial dysfunction literature. Cross-disease inference; preliminary n=12; not yet replicated.

Van Eeden et al. found reduced cytochrome b (CyB) expression specifically in SSc patients meeting ME/CFS criteria, independent of organ fibrosis burden . ME/CFS literature reports mitochondrial dysfunction with variable site localization (Complex I, IV, V across studies). Complex III occupies a Q-cycle bottleneck where reverse electron transport generates reactive oxygen species (ROS); CyB is mitochondrially encoded and its translation is uniquely sensitive to mtDNA damage and inflammatory mtROS feedback. This suggests a hypothesis: when any upstream insult — post-viral, autoimmune, or inflammatory — chronically elevates electron pressure on the electron transport chain, Complex III becomes the rate-limiting failure point, regardless of the initiating trigger. SSc-ME/CFS would then represent a “convergent endpoint” model, not a distinct disease, sharing the metabolic collapse mechanism with post-infectious ME/CFS via a common mitochondrial vulnerability.

Falsifiable prediction: In a head-to-head bioenergetic study (Seahorse + RNA-seq) comparing post-viral ME/CFS, SSc-ME/CFS, and Long COVID-ME/CFS patients, a Complex III-specific deficit (reduced CyB expression; reduced antimycin-sensitive respiration fraction) will be present in ≥60% of all three groups despite differing upstream triggers, and will be absent in fatigued SSc patients not meeting ME/CFS criteria.

Limitations: Van Eeden 2023 is limited to n=12 per group with no functional mitochondrial assays; the CyB deficit rests on gene-expression data alone, and CyB mRNA reduction may reflect reduced mitochondrial mass or biogenesis state rather than a Q-cycle bottleneck specifically. The ME/CFS mitochondrial literature cited as support actually implicates Complexes I, IV, and V across different studies — the Complex III specificity is a feature of this single preliminary finding, not of the broader literature. Deconditioning and lymphopenia can independently reduce PBMC mitochondrial transcript abundance. Replication with PBMC respirometry and muscle biopsy in larger, multi-center cohorts is required before this should be treated as more than hypothesis-generating.

Speculation: eNOS/NOS3 Hypofunction as Shared Endothelial Substrate for SSc-Raynaud and ME/CFS Exercise Intolerance

Certainty: 0.40. NOS3 is a validated Raynaud’s GWAS locus ; ME/CFS endothelial dysfunction is documented; BH4-depletion/ADMA accumulation mechanisms are plausible connecting links. Direct NOS3 genotyping in ME/CFS cohorts absent. Partially supported by indirect evidence; not yet replicated as a unified model.

Tervi et al. identified NOS3 (endothelial NO synthase) as a genome-wide significant Raynaud’s risk locus, functionally confirmed by CRISPRi knockdown of the endothelial regulatory element . ME/CFS literature documents impaired flow-mediated dilation and reduced post-occlusive reactive hyperaemia ; ADMA — an endogenous eNOS inhibitor — has not been confirmed as elevated in large ME/CFS cohorts but is a mechanistically plausible mediator of eNOS suppression. NOS3 produces endothelial NO; reduced NO bioavailability causes both vasospasm and impaired microvascular reactivity during exercise — a candidate mechanism for exercise-triggered muscle deoxygenation and the delayed PEM peak. Hypothesis: NOS3 hypofunction (genetic and/or acquired via BH4 depletion, ADMA accumulation, or oxidative eNOS uncoupling) is a shared substrate linking SSc microvascular injury to ME/CFS exercise intolerance. This would explain why Raynaud-positive ME/CFS patients with orthostatic intolerance may respond differently to NO-pathway interventions (L-citrulline, BH4 supplementation) than Raynaud-negative patients.

Falsifiable prediction: ME/CFS patients will show elevated plasma ADMA/SDMA and reduced post-occlusive reactive hyperaemia compared to matched controls, with the deficit correlating with PEM severity score (DSQ-PEM); L-citrulline (6 g/day, 8 weeks) will improve flow-mediated dilation by ≥1.5% absolute in the subgroup with baseline FMD below 6%, in a placebo-controlled crossover (n≥30).

Limitations: The NOS3 GWAS locus was identified in primary Raynaud’s, not SSc or ME/CFS specifically; its relevance to ME/CFS endothelial dysfunction is a cross-disease inference, not a direct finding. ADMA elevation in ME/CFS has not been confirmed in large cohorts. Reduced FMD in ME/CFS may be explained by deconditioning — bedrest studies in healthy controls produce comparable FMD reductions — without requiring a NOS3-specific mechanism. L-citrulline trials in other fatigue conditions (heart failure, COPD) have produced inconsistent FMD improvements, lowering the prior probability for ME/CFS. The BH4/ADMA hypothesis is mechanistically coherent but has not been tested interventionally in ME/CFS.

Speculation: Cardiac Autonomic Neuropathy as Hidden Phenotype in ME/CFS

Certainty: 0.45. Masini 2021 found 50% formal CAN prevalence in SSc correlated with microvascular damage; ME/CFS HRV literature documents autonomic dysfunction but formal CAN criteria (Ewing battery) have not been systematically applied. Evidence is by analogy; not yet replicated in ME/CFS.

Masini et al. found cardiac autonomic neuropathy (CAN) by Ewing battery in 50% of SSc patients, correlated with nailfold capillaroscopy active pattern (microvascular damage) and independently predicted by anti-Scl-70 antibody . ME/CFS literature documents HRV reduction, reduced baroreflex sensitivity, and chronotropic incompetence on CPET — all consistent with autonomic dysfunction — but rarely applies the standardized Ewing 5-test battery used in diabetes and SSc medicine. Hypothesis: a substantial fraction of ME/CFS patients meet formal CAN criteria, and CAN burden correlates with PEM severity and exercise VO2 decrement magnitude. This reframing would shift the mechanism from “central fatigue” toward “peripheral autonomic insufficiency limiting cardiac output augmentation” — with direct implications for autonomic-targeted treatments.

Falsifiable prediction: Applying the Ewing 5-test battery to 100 ME/CFS patients will identify ≥30% with definite CAN (≥2 abnormal Ewing tests); CAN status will predict 2-day CPET VO2 drop magnitude independently of baseline severity.

Limitations: The Ewing battery was developed and validated for structural diabetic autonomic neuropathy — its performance characteristics in functional dysautonomia (POTS, autoantibody-mediated) are not established. Deconditioning alone can reduce HRV and Valsalva ratio in healthy controls, and ME/CFS patients are often deconditioned; this creates a confound. POTS, present in ~30% of ME/CFS, would generate Ewing abnormalities by definition without implying structural neuropathy. SSc CAN has structural microvascular correlates confirmed by capillaroscopy; the equivalent structural substrate in ME/CFS has not been identified, and the mechanism may be entirely functional. The battery also requires procedures (Valsalva, deep breathing, orthostatic challenge) that can trigger PEM and needs adapting for severe/very-severe patients.

Speculation: ADRA2A Vasospasm as Autonomic Tone Driver in ME/CFS Cold Sensitivity and Orthostatic Intolerance

Certainty: 0.30. ADRA2A is the primary Raynaud’s GWAS risk locus ; ME/CFS Raynaud comorbidity is documented (~20%); no ADRA2A genotyping in ME/CFS cohorts has been performed. Cross-disease inference; not yet replicated.

Two large GWAS studies establish ADRA2A (α2A-adrenoreceptor) as the strongest Raynaud’s risk locus (OR=1.26, $p < 9.6 \times 10^{-27}$) . The α2A receptor is presynaptic on sympathetic nerve terminals (limiting norepinephrine release) and postsynaptic on vascular smooth muscle (causing vasoconstriction) — gain-of-function risk variants would cause both excess peripheral vasoconstriction and impaired baroreflex feedback. Hypothesis: a subset of ME/CFS patients with Raynaud-positive history and orthostatic intolerance carry ADRA2A risk variants that contribute mechanistically to peripheral vasospasm (cold extremities, Raynaud episodes) and impaired postural blood pressure regulation simultaneously. This would link cold sensitivity, OI, and exercise intolerance in a single genetic substrate — testable via Mendelian randomization using Raynaud GWAS summary statistics and DecodeME outcome data.

Falsifiable prediction: Genotyping ME/CFS patients at ADRA2A Raynaud lead variants will show enrichment (OR≥1.5) in patients with both Raynaud-positive history and OI-positive tilt test, compared to ME/CFS patients with neither.

Limitations: ADRA2A risk variants confer modest individual effect sizes (OR≈1.26); predictions of OR≥1.5 in a doubly-selected subgroup face winner’s curse — subgroup effects typically shrink from discovery estimates. The leading ME/CFS vasospasm hypothesis invokes α2C adrenoreceptor upregulation (via cold-induced ROS/Rho-kinase) rather than genetic ADRA2A variants; if autoantibody-mediated α2C upregulation is the operative mechanism, ADRA2A germline variants may be irrelevant to ME/CFS vasospasm. The GWAS variant identified is non-coding and regulatory; its directionality (gain or loss of ADRA2A function) is inferred but not confirmed.

Open Question: Does PEM Occur in SSc? A Tractable Research Gap

Post-exertional malaise (PEM) has not been systematically assessed in SSc cohorts using validated instruments (DSQ-PEM, PEM questionnaire, 2-day CPET) despite the fact that a subset of SSc patients reports fatigue patterns consistent with ME/CFS criteria . SSc patients are followed in large, well-characterized prospective registries (SPIN, EUSTAR) with biobanked samples and serial phenotyping — making this one of the few settings where pre-onset baseline data would exist for ME/CFS converters. Key questions: (1) What fraction of SPIN/EUSTAR patients show PEM-positive fatigue profiles by DSQ-PEM? (2) Do PEM-positive SSc patients show the same immunometabolic signature (CyB deficit, elevated IFN-I score, autonomic neuropathy) as the van Eeden 2023 SSc-ME/CFS subgroup? (3) Does CCC/IOM-criterion ME/CFS develop prospectively in SSc patients who show early CyB or IFN-I signal?

Open Question: IFN-I Score plus CyB Expression as Two-Axis ME/CFS Subtype Stratifier

Two orthogonal axes — immune activation (IFN-I stimulated gene panel score) and mitochondrial deficit (CyB / Complex III expression) — define a candidate 2×2 stratification of ME/CFS: high-IFN/low-CyB (autoimmune-metabolic), high-IFN/normal-CyB (immune-dominant), low-IFN/low-CyB (post-viral metabolic), low-IFN/normal-CyB (other/unexplained). The SSc-ME/CFS data from van Eeden 2023 and the IFN signature model [NO SPECULATION] together suggest these two axes are biologically independent and reflect distinct upstream drivers. No ME/CFS cohort study has simultaneously measured both IFN-I score and CyB expression in the same patients. A cluster analysis combining these axes could yield reproducible subtypes with distinct mechanistic targets — IFN-high patients as candidates for immune-modulating strategies, CyB-low patients as candidates for mitochondrial-support approaches — though neither has been tested in subgroup-stratified trials.

Speculation: Vitiligo as a Non-Fibrotic Control for the ME/CFS TRM-IL-15 Hypothesis

Certainty: 0.25. Based on shared CD8+ TRM + IFN-γ + IL-15 biology in vitiligo and theoretical ME/CFS TRM model; no direct vitiligo-ME/CFS comparison data exists. Not yet replicated.

Vitiligo shares with LS the CD8+ TRM dominance and IL-15 trans-presentation mechanism but does NOT produce fibrosis — melanocytes are destroyed (not replaced by collagen). If ME/CFS involves TRM-driven pathology at multiple sites (Section [NO SPECULATION]), the vitiligo comparator tests a specific component: the TRM/IL-15 axis. ME/CFS patients with vitiligo (TRM-positive, fibrosis-negative) compared to those with LS (TRM-positive, fibrosis-positive) could distinguish whether ME/CFS chronicity/irreversibility correlates with fibrotic terrain, separating the TRM hypothesis from the fibrotic terrain hypothesis. Vitiligo prevalence is elevated in the same autoimmune comorbidity cluster as LS and ME/CFS .

Falsifiable prediction: In a ME/CFS registry, vitiligo prevalence will be elevated above population baseline (expected ~1–2%) at a rate comparable to or greater than other autoimmune comorbidities. ME/CFS patients with vitiligo will show higher EV miR-155 and lower NK cytotoxicity than ME/CFS patients without any skin autoimmune comorbidity.

Limitations: Vitiligo is often cosmetically minimized and under-reported in registry questionnaires. The TRM comparison requires matched biopsy data from ME/CFS patients with each condition, which is not feasible at scale.

22 Hidradenitis Suppurativa as Autoinflammatory Comparator: The IL-1beta-NLRP3-IL-17 Convergence

Hidradenitis suppurativa (HS; maladie de Verneuil) is a chronic autoinflammatory skin disease of the intertriginous areas, characterised by painful nodules, abscesses, and sinus tracts. Unlike psoriasis (which has established Th17/IL-17-centred treatment paradigms), HS is driven by a broader inflammatory architecture — TNF-alpha, IL-1beta, IL-17, and IL-23 — with the NLRP3 inflammasome as a central organising node. HS is increasingly classified as a systemic autoinflammatory rather than purely dermatological disease, and it carries a significant comorbidity burden including metabolic syndrome, inflammatory arthritis, and fibromyalgia. A single large population study (Prens 2022, Lifelines Cohort, n = 56,084) found HS associated with ME/CFS at adjusted OR 1.72 (95% CI 1.06–2.78), alongside fibromyalgia (OR 2.26, 95% CI 1.64–3.11), after adjustment for age, sex, BMI, smoking, and socioeconomic status (Prens et al. 2022). This association should be treated as provisional: the lower CI (1.06) is barely above the null; both HS and ME/CFS were self-reported (not clinically validated); depression, sleep quality, and physical activity level were not adjusted for (all shared confounds); and the finding has not been replicated in an independent cohort four years after publication. Fatigue prevalence in HS is 30–61% across studies — comparable to SLE and RA — and is independently associated with pain, depression, and kinesiophobia (fear of movement) (B. McGrath et al. 2026) (Almahdi et al. 2026) (Meral Ketenci, Meral, and Meral Obholzer 2026). This prevalence range comes predominantly from dermatology clinic populations (Hurley II–III disease disproportionately represented; tertiary referral bias) and may overestimate fatigue in the community HS population, most of whom have mild (Hurley I) disease and are undiagnosed (Prens et al. 2022). (Origin: brainstorm — evidence quality concern.)

HS is mechanistically interesting for ME/CFS because it provides a human autoinflammatory disease in which the NLRP3/IL-1beta axis — a pathway also implicated in ME/CFS PEM pathophysiology via mitochondrial danger signals — is causally implicated (scRNA-seq, in vitro, and PCR evidence) and partially targetable (biologics targeting TNF-alpha, IL-17, and IL-1beta are approved or in trials in HS, though none have validated fatigue endpoints). The HS literature therefore offers a “natural laboratory” for studying NLRP3-driven systemic inflammation and its fatigue consequences — a pathway that in ME/CFS can only be studied inferentially.

Speculation: Hidradenitis Suppurativa–NLRP3–IL-1beta Pathway as Mechanistic Bridge to ME/CFS Fatigue

Certainty: 0.35. Based on convergent indirect evidence across two independently-validated bodies of research — NLRP3 involvement in HS (5 papers, cert aggregated 0.65) and NLRP3 mediation of fatigue in animal models (2 independent KO studies, cert aggregate 0.60) — with zero direct HS–ME/CFS mechanistic studies. The epidemiological association (OR 1.72) is from a single study; the biochemical bridge is inferential. (Origin: literature synthesis.)

The NLRP3 inflammasome is robustly established as a driver of HS inflammation. scRNA-seq of HS skin lesions shows enrichment of NLRP3, IL-1beta, and IL-17 pathways, with the NLRP3 inhibitor MCC950 significantly reducing IL-1beta and IL-17A secretion from HS skin explants (Moran et al. 2023). NLRP3 mRNA is elevated in both lesional and perilesional HS skin vs healthy controls (Krajewski, Szukała, and Szepietowski 2024). Metformin, which suppresses NLRP3 via the AMPK-mTOR pathway, reduces inflammatory markers in HS patient PBMCs (Petrasca et al. 2023). In ME/CFS, PEM is proposed to be NLRP3-mediated via mitochondrial ROS/mtDNA release → IL-1beta → neuroinflammation (Jin et al. 2026), and NLRP3 KO mice show markedly reduced fatigue after LPS and repeated-swim challenges with decreased serum and brain IL-1beta (ZT Zhang et al. 2016) (Z. Zhang et al. 2017).

If the NLRP3/IL-1beta pathway is a shared mediator of fatigue in both conditions, then HS serves as a human model system in which the biochemical cascade from systemic inflammation to fatigue perception can be studied with existing disease infrastructure — biopsiable tissue, approved cytokine-targeting therapies, and dose-titrated anti-inflammatory agents — that is unavailable in ME/CFS. Existing HS biologics (adalimumab, secukinumab, bimekizumab, anakinra) targeting TNF-alpha, IL-17A/F, and IL-1beta provide natural-experiment data on whether effective anti-cytokine treatment reduces fatigue, despite the fact that fatigue has never been a primary endpoint in HS clinical trials.

Falsifiable prediction: An observational study of HS patients initiating biologics (anti-TNF, anti-IL-17, anti-IL-1beta) will show a clinically significant (>MCID) reduction in FSS or PROMIS-Fatigue scores from baseline to Week 16 in N-of-1 responder analyses, with IL-1beta-targeted therapy showing the largest fatigue effect. Falsified if anti-cytokine treatment produces no fatigue improvement beyond placebo effect despite achieving clinical HS response (HiSCR ≥ 50). Kilgour 2026 — in which the oral P2X7-NLRP3 inhibitor AZD9056 failed to produce clinical HS improvement (HiSCR primary endpoint negative) despite restoring PBMC cytokine production ex vivo (Kilgour et al. 2026) — already provides a partial falsification: the null HiSCR result makes a positive fatigue signal extremely unlikely, and the fatigue data from this trial has not been published. Two interpretations: (a) NLRP3 is necessary but not sufficient for HS inflammation (redundant inflammasomes and TLR pathways); (b) the P2X7 receptor is the wrong node within the NLRP3 pathway (ATP is not the dominant NLRP3 activator in HS, compared to K+ efflux or lysosomal cathepsins). Either interpretation constrains the expectation that single-node NLRP3 inhibition would improve fatigue in any autoinflammatory disease — and by extension, in ME/CFS. (Origin: brainstorm — evidence quality concern.)

Limitations: The entire bridge is inferential — zero studies have directly measured NLRP3 in ME/CFS patients, and zero studies have used validated fatigue instruments as an endpoint in HS biologic trials. The single epidemiological association (Prens 2022) is cross-sectional and self-reported with a lower CI grazing 1.0. HS is a distinct disease with its own pathogenic drivers (follicular occlusion, keratinocyte dysfunction, microbiome dysbiosis of the pilosebaceous unit) that have no ME/CFS counterpart. The fatigue phenotype in HS has never been characterised for PEM specifically, so it is unknown whether HS fatigue is post-exertional (Type 1, like ME/CFS) or constant inflammatory-cytokine-driven (Type 2, like active rheumatoid arthritis). The two could be mechanistically distinct, with NLRP3 driving Type 2 fatigue in HS and mitochondrial-ROS-triggered NLRP3 driving Type 1 fatigue in ME/CFS — an untested but critical distinction. The NLRP3 mouse evidence (Zhang 2016, Zhang 2017) comes from acute LPS/swim challenge models that capture acute IL-1beta-mediated sickness behaviour (resolution within 72 h), not the chronic, sensitised, relapsing-remitting PEM state that is the defining feature of ME/CFS. Extrapolating from an acute model to a chronic disease without a chronic NLRP3 activation model that includes PEM features is an untested assumption — a chronic NLRP3 gain-of-function mouse model followed for months with serial exercise challenges has never been created. (Origin: brainstorm — evidence quality concern.)

Consequence: If the HS→NLRP3→fatigue pathway is validated, HS would become a “fast-track” human model for studying cytokine-driven fatigue mechanisms — with biopsiable tissue, existing biologics, and clinical trial infrastructure — that could accelerate ME/CFS-relevant fatigue biology by a decade compared to de-novo ME/CFS studies. If the pathway does not hold (i.e., HS biologics resolve skin lesions but not fatigue), it would constrain the hypothesis that IL-1beta/IL-17 pathway inhibition is sufficient for fatigue improvement in systemic inflammatory states, redirecting ME/CFS research toward inflammasome-independent fatigue mediators.

Speculation: Prodromal HS Fatigue as a Model for Cytokine-to-Fatigue Latency in PEM

Certainty: 0.30. Based on a single questionnaire study in HS (Ring 2017, n = 72) demonstrating systemic symptoms preceding visible HS lesions by 12–24+ hours; no direct comparison to ME/CFS PEM time-course exists. (Origin: literature synthesis.)

Ring et al. found that 83.3% of HS patients experience prodromal symptoms 12–24+ hours before visible inflammatory lesions erupt; 32% of these patients report fatigue as a prodromal feature, alongside malaise (23%), headache (11%), and nausea (2%) (Ring et al. 2017). This temporal pattern — systemic symptoms emerging well in advance of a local inflammatory event — mirrors the delayed onset of PEM in ME/CFS (typically 12–48h post-exertion). In HS, the mechanism is plausibly a cytokine “spillover” from incipient follicular inflammation: local NLRP3/IL-1beta activation in the pilosebaceous unit releases systemic IL-1beta, IL-6, and TNF-alpha before the lesion becomes clinically apparent, producing a sickness-behaviour syndrome (anorexia, fatigue, hyperalgesia, social withdrawal) that is evolutionarily conserved across mammals.

If the HS prodrome and ME/CFS PEM share a common effector — systemic cytokine-mediated sickness behaviour originating from a spatially-localised inflammatory trigger — then HS provides a tractable model for studying the latency kinetics between inflammatory initiation and fatigue perception. In HS, the trigger site is visible and accessible (skin), the timing relative to a pre-existing inflammatory wave can be serially sampled, and a dose-response relationship can be studied by comparing lesion size/inflammatory burden to fatigue magnitude. This is impractical in ME/CFS, where the trigger (exertion) is not localisable to a biopsiable site and the temporal relationship cannot be studied with serial biochemical sampling in a way that isolates the initiating event.

Falsifiable prediction: Serial serum IL-6, TNF-alpha, and IL-1beta measured Q4h throughout an HS flare prodrome-to-resolution will show a cytokine peak that temporally precedes (by 4–8h) the peak of patient-reported fatigue severity. PEM-like-symptom onset latency will correlate with the cytokine peak-to-fatigue lag (r > 0.5). Falsified if fatigue rises simultaneously with or before systemic cytokines, or if no temporal dissociation between lesion and fatigue onset is detectable.

Limitations: Ring 2017 was a questionnaire study (n = 72) with no serum cytokine measurements and no validated fatigue instrument; prodrome description is entirely patient-reported. The PEM comparison is a structural analogy, not an empirical claim. The cytokine-to-fatigue latency hypothesis requires controlled serial biomarker sampling with matched HS-lesion and PEM-timepoint protocols — a study that has never been performed. HS biopsies cannot ethically be taken from non-lesional skin during the prodromal phase in a way that would confirm incipient inflammation; the cytokine source must remain inferred, not directly demonstrated. The intrafollicular anaerobic microbiome (Prevotella, Porphyromonas) may drive the inflammatory cascade in ways that have no ME/CFS parallel.

Consequence: If the latency kinetics of HS prodrome and ME/CFS PEM are shown to be mechanistically similar, a novel experimental paradigm becomes possible: using HS patients as “accelerated” human models where the inflammatory trigger is spontaneous and serial sampling is feasible — compressing PEM mechanism studies from years to months. Researchers could test anti-IL-1beta pre-treatment for aborting PEM-like symptoms without waiting for sporadic ME/CFS PEM events.

Speculation: Kinesiophobia as a Shared Behavioral Phenotype Across HS, FM, and ME/CFS

Certainty: 0.35. Based on one HS-specific study (Meral Ketenci 2026) and convergent evidence from FM and ME/CFS activity-avoidance literature. No cross-condition kinesiophobia comparison has been performed. (Origin: literature synthesis.)

Meral Ketenci et al. found that HS patients have significantly higher kinesiophobia (fear of movement) and lower physical activity compared to matched controls, with pain intensity, fatigue severity, and depressive symptoms as independent predictors in multivariable regression (Meral Ketenci, Meral, and Meral Obholzer 2026). This triad — pain → fatigue → depression → movement avoidance — is a well-documented pattern in fibromyalgia and post-exertional malaise in ME/CFS. The shared structure suggests that kinesiophobia may be a unified behavioural phenotype of chronic inflammatory conditions rather than disease-specific, mediated by a common pathway: repeated experience of symptom exacerbation after physical activity → learned avoidance → deconditioning → reduced activity tolerance → further avoidance.

In HS, kinesiophobia has an identifiable physical trigger (movement that stretches or compresses inflamed intertriginous skin) that is distinct from but behaviourally convergent with the diffuse pain-triggered avoidance of FM and the PEM-driven post-exertional avoidance of ME/CFS. The fact that three conditions with different primary pathology (skin, musculoskeletal, metabolic) produce the same behavioural end-state supports a shared neurobiological substrate — possibly IL-1beta/IL-6-mediated sickness behaviour acting on anterior insula and anterior cingulate cortex — rather than condition-specific psychological processes.

Falsifiable prediction: Using the Tampa Scale of Kinesiophobia (TSK) and validated fatigue (FSS) / pain (VAS) instruments across HS, FM, and ME/CFS cohorts, a structural equation model will show that a latent “movement-avoidance” factor, driven equally by pain and fatigue, fits the data from all three conditions with configural invariance (CFI > 0.95, RMSEA $<$ 0.06), supporting a shared phenotype rather than condition-specific kinesiophobia. Falsified if configural invariance is rejected across conditions.

Limitations: Single HS study (Meral Ketenci 2026); no cross-condition TSK data exists. Correlation ≠ causation: the fatigue-pain-depression-kinesiophobia cluster could be mediated by common third factors (sleep disturbance, socioeconomic disadvantage, diagnostic odyssey) rather than by a shared inflammatory mechanism. Kinesiophobia questionnaires (TSK-11, TSK-17) were validated in musculoskeletal pain populations, not inflammatory skin diseases; their psychometric properties in HS are unknown. The IL-1beta/insula hypothesis for kinesiophobia is itself inferential — no neuroimaging study has linked IL-1beta levels to kinesiophobia in any condition.

Consequence: If kinesiophobia is a shared target across chronic inflammatory diseases, graded-activity and pain-neuroscience-education interventions developed for FM could be adapted to HS and ME/CFS with minimal modification — accelerating treatment development. The caution is that for ME/CFS patients in whom activity triggers PEM via a metabolic mechanism (not just pain/fatigue perception), kinesiophobia reduction without addressing the underlying metabolic limit could be harmful, not helpful — distinguishing perceptual from metabolic kinesiophobia is the critical clinical challenge.

Limitation: The Autoinflammatory Disease Category — Absent from ME/CFS Literature

Certainty: 0.70. Based on a systematic search for “autoinflammatory disease” × ME/CFS returning zero results across PubMed. The concept is absent from the indexed literature, not a negative finding.

A PubMed search for the intersection of “autoinflammatory disease” AND (“chronic fatigue syndrome” OR ME/CFS) returned zero results (July 2026). Autoinflammatory diseases — a distinct category of innate-immune-driven, IL-1beta/IL-18/NLRP3-mediated conditions including familial Mediterranean fever (FMF), cryopyrin-associated periodic syndromes (CAPS), TNF receptor-associated periodic syndrome (TRAPS), and the polygenic counterparts (HS, adult-onset Still’s disease, SAPHO syndrome, Behçet’s disease) — have never been systematically compared to or studied alongside ME/CFS. This is not a null result (no study found autoinflammatory features absent in ME/CFS) but a complete absence of investigation — the category has simply not been examined.

This matters because autoinflammatory diseases:

Share the NLRP3/IL-1beta effector arm with ME/CFS PEM models,
Produce fatigue as a major symptom, typically fluctuating in a flare-remission pattern,
Are treated with therapies (anakinra, canakinumab, colchicine) that are mechanistically targeted to the same pathways hypothesised to drive ME/CFS,
Produce systemic symptoms — fever, malaise, arthralgia, myalgia — that overlap partially with ME/CFS (though fever is typically absent in ME/CFS, a differentiating feature),
Yet are clinically distinguishable from ME/CFS by their objective inflammatory markers (CRP, SAA, ESR) — markers that are typically normal or only mildly elevated in ME/CFS.

The absence of “autoinflammatory” as a research category in ME/CFS may reflect a real biological distinction — ME/CFS is not a classical autoinflammatory disease because it lacks the systemic inflammatory biomarker signature. Or it may reflect a field artefact — no one has looked. Until HS, FMF, or CAPS cohorts are systematically screened for ME/CFS criteria (and ME/CFS cohorts for autoinflammatory disease prevalence and NLRP3 biomarkers), the relationship between these two adjacent innate-immune disease categories cannot be assessed.

Consequence: This is a categorical research gap not a negative finding. Adding validated ME/CFS screening instruments (DSQ-PEM, DePaul Symptom Questionnaire) to existing autoinflammatory disease registries (Eurofever, HS ALLIANCE global registry) is a low-cost, high-yield next step that would either establish autoinflammatory-ME/CFS comorbidity rates or document their distinctness within one to two years.

Open Question: Does Anti-Cytokine Therapy for HS Reduce Fatigue Independently of Skin Improvement?

HS biologics — adalimumab (anti-TNF-alpha, approved, PIONEER I/II trials), secukinumab (anti-IL-17A, approved, SUNSHINE/SUNRISE trials), bimekizumab (anti-IL-17A/F, approved, BE HEARD I/II trials), anakinra (anti-IL-1beta, off-label, small series only) — are effective for HS skin lesions (HiSCR endpoint). Fatigue has never been a primary or key-secondary endpoint in any phase 3 HS trial: No validated fatigue instrument (FSS, PROMIS-Fatigue, FACIT-F) was used in the phase 3 trials of approved HS biologics (adalimumab PIONEER, secukinumab SUNSHINE/SUNRISE, bimekizumab BE HEARD); The DLQI, used in all phase 3 trials, includes one item (“social or leisure activities affected”) that is sometimes used as a fatigue proxy, but this conflates fatigue with dermatological embarrassment and functional limitation from pain — it is inadequate for distinguishing fatigue from skin-specific disability. (Origin: brainstorm — evidence quality concern.) Post-hoc or registry analyses could answer: does effective anti-cytokine treatment reduce fatigue, and does this reduction correlate with the skin response (HiSCR) or diverge from it? A dissociation — skin improves but fatigue persists — would constrict the NLRP3/cytokine→fatigue hypothesis. A concordant improvement would strengthen it but could still be confounded by pain reduction and improved sleep (both secondary to skin improvement and independent of direct anti-cytokine fatigue effects). IL-1beta-targeted therapy (anakinra, recombinant IL-1 receptor antagonist; canakinumab, anti-IL-1beta monoclonal antibody) is the cleanest test: IL-1beta is the most directly NLRP3-linked cytokine and has the strongest preclinical fatigue evidence via IL-1beta→sickness behaviour. Kilgour 2026 (Phase 2, AZD9056, oral P2X7-NLRP3 inhibitor) already demonstrated that NLRP3 inhibition alone had no clinical HS effect despite restoring PBMC cytokine production (Kilgour et al. 2026) — the question is whether the fatigue signal in that trial diverged from the skin signal (i.e., fatigue improved while HiSCR did not). That data has not been published. An open-label pilot of anakinra (recombinant IL-1RA) in HS patients with high baseline fatigue, using FSS/PROMIS-Fatigue as primary endpoint and HS-PGA as secondary, would provide the most direct test of the HS→IL-1beta→fatigue hypothesis within a one-year timeline.

Consequence: Answering this question is low-hanging fruit — post-hoc fatigue analyses of phase 3 HS trial data could be completed within months without new data collection, and would provide the first direct evidence for or against the general principle that anti-cytokine therapy reduces fatigue in a human autoinflammatory disease. A positive result would support repurposing IL-1beta blockade for fatigue in other conditions; a negative result would constrain the cytokine→fatigue model and redirect effort toward non-inflammatory fatigue mechanisms.

Open Question: Is the Fatiguing Subset of HS Patients an Undiagnosed ME/CFS Subgroup?

The 30–61% of HS patients with clinically significant fatigue (Almahdi et al. 2026) has never been characterised using validated ME/CFS diagnostic instruments (DSQ-PEM, CCC, IOM criteria). Cross-sectional screening of an HS registry cohort with DSQ-PEM + CCC/IOM questionnaires would answer: what proportion of fatigued HS patients meet ME/CFS criteria? Does this subgroup differ from non-ME/CFS HS patients in disease severity, inflammatory markers (CRP, ESR, IL-1beta, IL-6), treatment response, or comorbidity profile? If the ME/CFS-criteria-meeting subgroup is substantial (e.g., >10% of HS patients), HS would transition from “anecdotal comorbidity” to “identifiable at-risk population” — with implications for biologics selection (some anti-cytokine therapies may worsen fatigue in ME/CFS-predisposed patients, as seen with interferon-alpha in hepatitis C). If near zero, the epidemiological OR of 1.72 reflects a small true-risk group or a shared confound (BMI, socioeconomic status), not a biologically meaningful overlap. Either outcome is informative.

Consequence: Screening HS registries with a validated PEM questionnaire costs roughly USD 5–15 per patient and could be completed within one year using existing infrastructure. A positive finding would identify an immediately actionable at-risk population for ME/CFS prevention studies and inform biologics selection in co-affected patients. A negative finding (near-zero ME/CFS prevalence in HS) would rule out HS as a meaningful ME/CFS comorbidity despite the elevated OR, reinforcing that population-level ORs can overstate individual risk.

Limitation: Conflating HS Type 2 Inflammatory Fatigue with ME/CFS Type 1 PEM Harms Both Patient Populations

Certainty: 0.55. Based on the convergent HS-fatigue, NLRP3, and PEM literature — the evidence consistently supports HS fatigue as Type 2 (cytokine-driven sickness behaviour) rather than Type 1 (post-exertional metabolic decompensation), and the most parsimonious interpretation of the epidemiological OR 1.72 is shared confounds (BMI, smoking, depression, healthcare access) rather than shared pathophysiology. (Origin: brainstorm.)

Three distinct harms arise from conflating HS fatigue with ME/CFS before objective PEM has been demonstrated in HS:

Harm to HS patients. Attributing HS fatigue to “undiagnosed ME/CFS” medicalises a symptom that may be fully explained by modifiable factors within dermatology’s scope — chronic pain, sleep deprivation from pruritus and drainage, anaemia of chronic disease, and the psychological burden of a disfiguring skin condition (all documented in the HS literature (B. McGrath et al. 2026) (Deckers, Zee, and Prens 2016) (Almahdi et al. 2026)). Treating HS fatigue as ME/CFS would divert attention from treating the drivers (pain management, sleep hygiene, correction of anaemia) that are within dermatology’s scope, subject HS patients to pacing recommendations when their fatigue may improve with appropriately graded physical activity as inflammation is controlled, and stigmatise HS fatigue as “medically unexplained” when it is well-explained by known HS pathophysiology.

Harm to ME/CFS patients and diagnostic specificity. Diluting ME/CFS diagnostic criteria by pathologising fatigue that is proportional to objective inflammatory burden, correlates with CRP, and may resolve with anti-cytokine therapy as “ME/CFS” weakens the specificity of ME/CFS as a diagnosis. ME/CFS entered the literature in part to distinguish post-exertional, non-inflammatory fatigue from the fatigue of active inflammatory disease. Collapsing this distinction reverses decades of nosological progress. Until a 2-day CPET study demonstrates objective PEM in HS patients (proposal 2.2), and until HS patients meeting ME/CFS criteria are shown to have normal CRP/ESR (i.e., fatigue dissociated from inflammatory burden), the simpler explanation — that HS fatigue is Type 2 inflammation-driven fatigue sharing symptom surface features but not mechanisms with PEM — fits all available evidence without requiring new mechanisms.

Harm to research resource allocation. The post-hoc secondary analyses of HS biologic trials (Does Anti-Cytokine Therapy for HS Reduce Fatigue Independently of Skin Improvement?) are genuinely low-cost and worth pursuing. But the resource-intensive studies — 2-day CPET in HS, serial Q12h cytokine sampling through flare cycles — should be funded only after a threshold probability of shared pathophysiology is demonstrated (e.g., independent epidemiological replication of OR > 1.5 after depression and sleep adjustment; positive DSQ-PEM screen in an HS registry cohort). Until then, these resources are better allocated to higher-prior-probability mechanistic bridges — anti-NR2 antibodies in SLE-ME/CFS, post-infectious disease spectrum with Long COVID as counterfactual evidence, or direct measurement of NLRP3 activity in ME/CFS PBMCs (which has never been done).

Consequence: The single most important result that would resolve the conflation question — 2-day CPET in HS patients with high fatigue — has never been performed. Until it is, the HS-ME/CFS literature should frame the epidemiological signal as provisional and the mechanistic bridge as inferential, with the Type 1-vs-Type 2 distinction flagged as the critical unresolved variable. Researchers designing HS fatigue studies should pre-register whether they expect PEM or constant fatigue and include the diagnostic instruments (DSQ-PEM, CPET-2, PROMIS-Fatigue) that can distinguish them.

23 Direct Immunophenotyping Comparison: Monocyte and Dendritic Cell Signatures

Achievement: Divergent Monocyte and Dendritic Cell Profiles in ME/CFS vs Long COVID

Certainty: 0.50. The largest comparative immunophenotyping study to date (n=207 total; 103 ME/CFS, 63 long COVID, 41 healthy controls — though ‘’largest’’ is relative to a field with few comparative studies, and the 41 controls are modest) by Petrov et al. used 12-parameter flow cytometry to directly compare monocyte, dendritic cell, and T cell subsets between ME/CFS and long COVID (Petrov et al. 2026). Key findings:

Long COVID: Increased M2-like (CD206-high) monocyte polarization, elevated CD80 expression across all monocyte subsets, dendritic cell expansion, and reduced HLA-DR on dendritic cells — consistent with persistent immune activation with features of immune exhaustion (on DCs; note that T cell exhaustion has been separately reported as absent in Long COVID in some studies, creating a complex picture where exhaustion may be compartment-specific)
ME/CFS: Reduced CD80 on M1-like (CD206-low) monocytes (paradoxical given M1-polarised cells normally upregulate CD80 — suggesting a functional defect beyond polarization state), impaired CCR7 expression on both monocytes and dendritic cells, reduced dendritic cell frequency — these monocyte/DC findings suggest a predominantly suppressively skewed monocyte/DC compartment. However, the same study also found T cell activation features in ME/CFS (see below), complicating a uniform ‘’immune suppression’’ label
T cells: Long COVID showed increased CD8+ central memory and CD4+CD95+ cells; ME/CFS showed increased CD8+ effector memory/terminal differentiation (CD45RA+CD62L-) with CD4+CD95+ elevation — indicating that ME/CFS is not a purely ‘’suppressed’’ immune state but a mixed phenotype with concurrent T cell activation features
Correlation networks: Long COVID had more extensive and integrated immune interactions; ME/CFS showed less coordinated activation patterns
Clinical correlation: Inverse relationship between monocyte HLA-DR expression and psychiatric symptom severity in ME/CFS — higher HLA-DR associated with lower depression/anxiety scores. This association is cross-sectional and cannot distinguish reverse causality (patients with less severe psychiatric symptoms may have better sleep, less physical deconditioning, or fewer medications — all of which could independently affect HLA-DR). The ‘’protective association’’ framing should be interpreted cautiously

The study achieved moderate discrimination between conditions using PCA and PLS-DA (VIP discriminators: CD80 on monocytes, CCR7 on dendritic cells, M2-like polarization); AUC was not reported, and the magnitude of ‘’moderate’’ discrimination is therefore unquantified. Significant within-group heterogeneity exists (as evidenced by the moderate discrimination itself), meaning many individual patients’ immune profiles partially overlap between conditions.

Limitations: Cross-sectional design — the observed divergence could reflect different disease stages (years to decades of illness in ME/CFS vs months to ~6 years in Long COVID) rather than fundamentally distinct mechanisms, as the same process at different timepoints could produce these patterns; no functional assays (surface marker expression was measured, not functional costimulatory or chemotactic capacity); vaccination status not analysed — a significant confound given that SARS-CoV-2 vaccination alters monocyte and DC phenotypes; single-center; excluded common ME/CFS medications (LDN, antihistamines, beta-blockers) — which may bias the ME/CFS cohort toward milder or less-treated patients and may not represent the broader ME/CFS population; healthy controls significantly younger than patient groups (age-adjusted ANCOVA applied, but residual confounding possible given age-dependent CCR7 expression); not yet replicated in an independent cohort. Certainty reduced from initial 0.70 because the study is unreplicated, cross-sectional, has medication exclusion biases, and did not include functional immune assays.

Implications: These findings suggest that ME/CFS and long COVID, despite overlapping clinical features, show divergent monocyte/DC phenotypes that, if replicated, might inform different therapeutic approaches. The direction of any therapeutic implication (restorative vs resolutive) remains speculative from cross-sectional immunophenotyping alone and requires longitudinal and interventional data. Proposals for immune-restorative approaches (for ME/CFS) and immune-resolution strategies (for long COVID) are tentative and must be conditioned on replication, functional validation, and resolution of the fundamental question whether the CD80/CCR7 reductions in ME/CFS represent maladaptive suppression or adaptive tolerance (see Chapter 18 and hypothesis registry).

24 Long-COVID as Counterfactual Validation of ME/CFS Biology

The Watton–Prusty review (Watton and Prusty 2026) provides a systematic analysis of how Long-COVID research has functioned as a “decisive counterfactual” for ME/CFS. (Certainty: 0.55)

Convergent findings across post-infectious conditions. Across multiple large-scale Long-COVID cohorts, investigators have independently identified abnormalities that map directly onto pre-existing strands of ME/CFS research: persistent innate and adaptive immune dysregulation, endothelial dysfunction and impaired microvascular regulation, aberrant coagulation and fibrinolysis-resistant fibrin(ogen) aggregates, metabolic reprogramming and mitochondrial stress signatures, circulating pathogenic factors capable of inducing cellular dysfunction in vitro, and evidence of viral persistence or abortive reactivation without productive infection. Each of these domains has been investigated in ME/CFS for decades, but Long-COVID studies provided statistical power, temporal anchoring (known infection date), and institutional legitimacy that ME/CFS research historically lacked. (Watton and Prusty 2026)

Shared but not identical. The Shahbaz et al. comparison of Long-COVID patients meeting ME/CFS criteria vs. idiopathic ME/CFS found both overlap and divergence: shared immune activation was evident, but Long-COVID-associated cases showed more pronounced T-cell exhaustion, NK-cell alteration, MAIT/gamma-delta T-cell depletion, and inflammatory monocyte skewing. Concurrently, Petrov et al. identified divergent monocyte/DC profiles (Section Divergent Monocyte and Dendritic Cell Profiles in ME/CFS vs Long COVID). These findings support placement of ME/CFS and Long-COVID within a broader post-infectious disease framework while arguing against unqualified mechanistic equivalence. The specific clinical expression — ME/CFS, Long-COVID, or post-infectious fatigue — may depend on the combination of pre-existing vulnerabilities and the nature of the trigger, consistent with the terrain-plus-trigger model. (Watton and Prusty 2026)

Research governance implications. The pace of progress on Long-COVID demonstrates what is achievable when disease burden, funding, and methodological ambition align. For ME/CFS, this convergence validates long-reported patient experiences and provides an emerging basis for translation of mechanistic insights into targeted therapeutic approaches — while also underscoring the consequences of decades of underinvestment. (Watton and Prusty 2026)

Hypothesis: Post-Infectious Disease Spectrum Model

ME/CFS, Long-COVID, and related post-infectious syndromes represent points on a disease spectrum sharing core mechanisms of impaired physiological resilience, with individual clinical expression determined by trigger-specific factors, pre-existing vulnerability, and disease duration. (Certainty: 0.45)

Evidence: Convergent findings across multiple biological domains (immune, vascular, metabolic, virological) in both conditions; divergent features in immune cell phenotypes suggesting trigger-specific or duration-specific effects. The Shahbaz and Petrov studies support both overlap and divergence simultaneously.

The Davis 2023 landmark review as a synthetic anchor. The Patient-Led Research Collaborative review (H. E. Davis et al. 2023) — a landmark patient-led synthesis (Nature Reviews Microbiology, 2023) — independently corroborates this spectrum framing and adds a formal temporal dimension. It documents that a substantial fraction of Long COVID patients meet ME/CFS criteria, that the majority report post-exertional malaise when it is specifically measured (H. E. Davis et al. 2023), and that Long COVID and ME/CFS share hypothesized mechanisms spanning immune dysregulation (with or without herpesvirus reactivation), persistent viral reservoirs, autoimmunity, microvascular clotting with endothelial dysfunction, and dysfunctional brainstem/vagus signalling (H. E. Davis et al. 2023). The review’s explicit framing of these as “multiple, potentially overlapping, causes” aligns with the multi-hypothesis rather than single-cause interpretation developed throughout this chapter.

Open Question: Long COVID Trajectory: Temporal Evidence for Post-Infectious Chronicity

Certainty: 0.60 (Long COVID evidence; ME/CFS extrapolation uncertain).

The Davis 2023 review compiles temporal data that bear directly on the question of whether post-infectious syndromes become chronic. Among patients symptomatic at 2 months after infection, 85% remained symptomatic at 1 year (H. E. Davis et al. 2023). Cognitive impairment prevalence progressed from 16% at 2 months to 26% at 12 months (H. E. Davis et al. 2023) — an increasing trajectory rather than a resolving one. Neurological symptoms often have a delayed onset (43% of cognitive-symptom cases onset ≥1 month after infection), and parosmia shows a characteristic ~3-month onset. By contrast, gastrointestinal and respiratory symptoms are more likely to resolve (H. E. Davis et al. 2023).

Interpretation. The high 1-year persistence and the increasing cognitive-impairment trajectory argue that post-infectious syndromes can become self-sustaining rather than resolving — consistent with the paper’s model of a sustained post-infectious state (see Post-Infectious Disease Spectrum Model and Chapter Epidemiological and Outcomes Research). ME/CFS is generally considered lifelong once established (H. E. Davis et al. 2023), so the Long COVID trajectory toward chronicity supports the contention that at least a substantial subset of Long COVID converges on an ME/CFS-like chronic state.

How observation would answer this. A prospective cohort tracking symptom and cognitive-impairment prevalence from acute infection to ≥24 months would resolve whether post-infectious symptoms plateau into a chronic state or continue to accrue: the question is supported if cognitive-impairment prevalence keeps rising past 12 months without a plateau, and would be answered in the negative if prevalence plateaus or declines once duration is accounted for (i.e., the observed 16%→26% rise reflects delayed ascertainment rather than true progression).

Severity applicability: unknown — the reviewed studies are not stratified by severity.

Consequence: If post-infectious syndromes commonly become self-sustaining and even worsen over the first year, this argues for early, mechanism-targeted intervention in the window before chronicity locks in — a rationale that applies equally to Long COVID and to post-infectious ME/CFS onset.

Open Question: Vaccination and Established Long COVID: Heterogeneous Symptom Response

Certainty: 0.45.

The Davis 2023 review reports heterogeneous effects of vaccination on established Long COVID symptoms: 16.7% of patients experienced symptom relief, 21.4% experienced worsening, and the remainder unchanged (Tsuchida et al. 2022) (H. E. Davis et al. 2023). Vaccination reduces the risk of developing new Long COVID (partial protection, ~15-41%), but its effect on established symptoms is mixed and unpredictable.

Interpretation. The heterogeneous response underscores that Long COVID (and by extension ME/CFS) is not a single uniform process — the same intervention can improve or worsen depending on the underlying mechanism (e.g., immune-amplification vs. immune-deficit subtypes). It cautions against blanket claims about vaccination or immunomodulation in post-infectious illness.

How observation would answer this. A stratified cohort study measuring pre-vaccination immune status (e.g., baseline cytokine/antibody profiles) alongside symptom response after vaccination would resolve whether the relief-vs-worsening split tracks a measurable baseline immune subtype — supported if a specific baseline profile (e.g., low-inflammatory/immune-deficit) predicts the 16.7% who improve and a contrasting profile predicts the 21.4% who worsen.

Severity applicability: unknown — not stratified.

Consequence: A treatment that helps some post-infectious patients and harms others signals mechanistic heterogeneity; this argues for careful individualised titration and mechanism-based patient stratification rather than uniform protocols.

Additional Long COVID findings with ME/CFS relevance (Davis 2023 review). Beyond the temporal and vaccine findings above, the Davis 2023 review (H. E. Davis et al. 2023) documents several organ-system findings that parallel or inform ME/CFS pathophysiology: (i) fungal translocation — elevated markers of fungal translocation from the gut/lung epithelium in Long COVID that induce NF-$\kappa$B signalling and cytokine production (Giron et al. 2022), a gut-lung-immune mechanism consistent with the corpus’s gut-microbiome and immune-activation emphasis; (ii) ocular/retinal microvascular involvement — retinal haemorrhages, cotton wool spots, retinal vein occlusion, and corneal small-nerve-fibre loss (Sen et al. 2022) (Bitirgen et al. 2021), echoing the small-fibre neuropathy and microvascular findings documented in ME/CFS; (iii) fragmented QRS on ECG as a low-cost cardiac-injury marker in COVID-19 recovery (Stavileci et al. 2022), relevant to the cardiac-involvement discussion; and (iv) male reproductive consequences — impaired sperm count, motility, and morphology correlated with elevated cytokines and seminal caspases (Maleki and Tartibian 2021). These are presented as corroborating the multisystem post-infectious character shared with ME/CFS rather than as ME/CFS-specific mechanisms in their own right; each awaits direct ME/CFS-cohort replication. Severity applicability: unknown for all (not stratified).

Pycnogenol. A pilot study reported that Pycnogenol (French maritime pine bark extract) improved oxidative-stress markers, microcirculation/endothelial function, and quality of life in COVID-19 recovery (Belcaro et al. 2022), prompting the Davis review (H. E. Davis et al. 2023) to list it among candidate endothelial-supportive treatments. Given the microclot/endothelial pathology shared with ME/CFS, this is a candidate (not validated) intervention; see the pharmacodiagnostic and treatment chapters for how endothelial-targeted options are triaged. Severity applicability: unknown; general COVID-recovery population (not ME/CFS-specific).

Limitation: Long COVID-ME/CFS Extrapolation: Limits of the Counterfactual

Certainty: 0.55.

The Davis 2023 review (H. E. Davis et al. 2023) is a powerful counterfactual for ME/CFS, but extrapolating its Long COVID findings to ME/CFS carries several limits that must temper the confidence of the overlap argument:

Narrative, not systematic. The review is a narrative synthesis; a subset of its findings rest on preprints and small cohorts, so per-claim certainty is lower than the review’s landmark status suggests (H. E. Davis et al. 2023).
Acute-to-chronic gap. Long COVID is largely studied in the first 1-2 years post-infection, while ME/CFS is typically chronic and lifelong. A substantial fraction of Long COVID cases improve over time (GI/respiratory symptoms often resolve), whereas ME/CFS persists — so Long COVID data are an acute-to-chronic extrapolation, not proof of identity with established ME/CFS.
Criteria overlap. The ≈50% of Long COVID meeting ME/CFS criteria may partly reflect overlapping, self-referential symptom criteria rather than shared underlying biology; the 2-day CPET divergence (no Day-2 deterioration in Long COVID vs. consistent ME/CFS deterioration) argues against full mechanistic equivalence.
Cohort overlap. Several Long COVID findings derive from overlapping cohorts or the same research groups; treating them as independent replications inflates apparent certainty.

Consequence: Long COVID evidence is best used as a strong suggestive bridge and a source of research hypotheses for ME/CFS, not as proof that the two are mechanistically identical — direct head-to-head mechanistic and longitudinal studies are still required before Long COVID mechanisms are treated as established ME/CFS pathophysiology.

Synthesis: Post-Infectious Chronicity as the Convergent Long COVID-ME/CFS Signal

The Davis 2023 landmark review (H. E. Davis et al. 2023), integrated across this chapter, collectively argues that post-infectious syndromes can become self-sustaining rather than resolving: a substantial majority of those symptomatic at 2 months remain symptomatic at 1 year, cognitive impairment increases over the first year (Long COVID Trajectory: Temporal Evidence for Post-Infectious Chronicity), and an insufficient early antibody response predicts progression to chronicity (Does Insufficient Early Antibody Response Gate Post-Infectious Chronicity?) — all consistent with ME/CFS being a lifelong post-infectious outcome (see Post-Infectious Disease Spectrum Model). The strongest constraint is that this remains an acute-to-chronic extrapolation: Long COVID is largely studied early in its course, a meaningful fraction improves, and the overlap may be partly criteria-driven (Long COVID-ME/CFS Extrapolation: Limits of the Counterfactual). The most actionable implication is that the acute-to-subacute window after infection may be the decisive therapeutic intervention point before chronicity locks in — a rationale that applies equally to Long COVID and to post-infectious ME/CFS onset. The open question is whether early immune-supportive or viral-clearing intervention can shift the trajectory away from the self-sustaining state.

Recent single-cell and functional immunology studies (Watton and Prusty 2026) have sharpened the understanding of how specific immune cell populations can exert disproportionate effects on whole-organism physiology in post-infectious disease. Rather than reflecting uniform immune activation, ME/CFS and Long-COVID increasingly appear to involve selective, stable reprogramming of particular immune subsets — most notably monocytes and T cells. Studies have identified distinct monocyte transcriptional and metabolic states in Long-COVID, characterized by altered interferon signaling, inflammatory tone, and metabolic gene expression. Such monocytes influence vascular function, coagulation, and tissue oxygenation through cytokine release, endothelial interaction, and modulation of redox balance.

In parallel, Kumar et al. demonstrated persistent alterations in T-cell activation and exhaustion profiles following viral infection, with consequences for immune regulation, mitochondrial signaling, and systemic energy balance. Complementing these findings, studies report functional impairment of antiviral T-cell responses in both ME/CFS and Long-COVID, including CD8+ T-cell dysfunction with exhaustion-like features, alongside expansion of atypical CD4+CD8+ “double-positive” T-cell populations. (Watton and Prusty 2026)

Conceptual advance. These observations support a model in which post-infectious disease is maintained by maladaptively reprogrammed immune populations. In this framework, immune cells function not merely as effectors of inflammation, but as mobile regulators of metabolic and vascular physiology, capable of shaping organism-level responses to stress. This conceptual shift integrates observations of monocyte alterations, T-cell exhaustion, endothelial vulnerability, and mitochondrial dysfunction into a single immunological framework. (Certainty: 0.50)

Open Question: Immune Reprogramming Durability and Reversibility

How stable is the reprogrammed immune state in post-infectious conditions? Is selective immune cell reprogramming an actively maintained process (requiring ongoing antigenic or inflammatory drive) or an epigenetically self-sustaining state that persists after the initial trigger resolves? The Bergemann hyperthermia study’s finding (now replicated mechanistically by Hochecker et al., who demonstrated WBH reduces autophagy and improves mitochondrial respiration in ME/CFS PBMCs (Hochecker et al. 2025)) that some stress signatures are dynamically modifiable suggests at least partial reversibility, but the durability and scope of immune reprogramming remain unknown. Longitudinal studies tracking immune cell states from acute infection through chronicity, combined with perturbation-based assays (heat, exercise, pharmacological challenge), are needed. (Watton and Prusty 2026)

Speculation: Intrapulmonary Shunt as Post-Viral Mechanism of Tissue Hypoxia

(Certainty: 0.35 — inferred from Long COVID data; zero direct ME/CFS measurements; mechanistic plausibility supported by pre-existing anatomical substrate and documented HPV impairment.)

Intrapulmonary right-left shunt — blood bypassing alveolar oxygenation via pre-existing arteriovenous anastomoses (IPAVAs) — may be a shared post-viral mechanism contributing to tissue hypoxia in both Long COVID and post-infectious ME/CFS. The evidence chain supporting this inference comes from five independent lines, all from post-COVID populations, with no direct ME/CFS data:

Shunt persistence: Qs/Qt elevated to ~8% (normal 3–5%) at 6 months post-COVID, despite normal spirometry and DLCO (Farrow et al. 2023). Independently replicated (Sandhu et al. 2026).
Anatomical substrate: Intrapulmonary arteriovenous anastomoses are present in ~30% of adults at rest and open in nearly all during exercise (Lyne, Camporota, and Montgomery 2024). Intrapulmonary bronchopulmonary anastomoses serve as microvascular shunt channels (Bush, Abman, and Galambos 2022).
Impaired protective reflex: Hypoxic pulmonary vasoconstriction (HPV) — the mechanism that normally reduces shunt by constricting vessels in poorly ventilated lung regions — is impaired after SARS-CoV-2 infection (Li et al. 2024). If other viral triggers similarly impair HPV, the mechanism generalizes beyond COVID.
Functional shunt regulation: IPAVA flow is actively regulated by oxygen tension — hyperoxia reduces shunt flow, while exercise and hypoxia may increase it (J. T. Davis et al. 2023). This dynamic regulation suggests a mechanism for exertional worsening.
Clinical phenotype match: The case of platypnea-orthodeoxia (positional hypoxemia) after COVID with no cardiac shunt (Espejo et al. 2025) demonstrates that post-viral intrapulmonary shunt can be clinically significant — severe enough to cause measurable desaturation on standing.

(Origin: integrative — literature-derived; no ME/CFS data exist.)

If this mechanism operates in ME/CFS: Post-exertional tissue hypoxia — already inferred from metabolic phenotypes (Hoel et al. 2021), lactate elevation, and impaired oxygen extraction (Walitt et al. 2024) — would have an identifiable proximal cause upstream of the microcirculatory and mitochondrial dysfunction documented in Chapter Energy Metabolism and Mitochondrial Function. The shunt would be a source of reduced arterial oxygen content, distinct from the delivery bottlenecks (endothelial dysfunction, RBC deformability, capillary remodeling) already described.

Falsifiable prediction: ME/CFS patients undergoing 100% FiO₂ shunt testing (A-a gradient measurement) would show Qs/Qt significantly above healthy controls (>5%), with the shunt fraction correlating with 2-day CPET measures of exertional desaturation and PEM severity. Falsified if Qs/Qt is normal (3–5%) in ME/CFS patients with documented exertional hypoxemia.

Consequence: If confirmed, intrapulmonary shunt would provide a treatable upstream target — supplemental oxygen during exertion, or interventions to restore HPV (pulmonary vasoconstriction) — for a subset of ME/CFS patients whose hypoxia is driven at the pulmonary level rather than the microcirculatory level. Until measured, this remains a plausible but untested extrapolation from Long COVID.

Non-generalizability caveat: SARS-CoV-2 directly infects pulmonary vascular endothelium via ACE2 — a property not shared by EBV, enteroviruses, influenza, or other classic ME/CFS triggers. If HPV impairment is ACE2-mediated, shunt is a COVID-restricted mechanism with no relevance to the ~75% of ME/CFS cases triggered by non-COVID pathogens. This is the single most important caveat: until HPV impairment is demonstrated after non-COVID viral triggers, the entire shunt→ME/CFS extrapolation is COVID-specific.

Clinical effect-size uncertainty: The measured shunt increase (~5 percentage points above normal, from ~3% to ~8%) reduces PaO₂ from ~100 mmHg to ~80–85 mmHg — well within the flat portion of the oxygen-hemoglobin dissociation curve where SaO₂ remains >95%. Healthy individuals with equivalent shunt from hepatopulmonary syndrome are often asymptomatic at rest. Whether this degree of chronic mild hypoxemia is sufficient to drive tissue-level metabolic consequences in ME/CFS — or whether it is a minor contributor dwarfed by microcirculatory and mitochondrial dysfunction — is unknown.

Measurement caveat: The 100% O₂ shunt measurement has known limitations: hyperoxia-induced absorption atelectasis may artifactually increase shunt; the non-physiological FiO₂ means values may not generalize to room-air or exercise conditions; and arterial cannulation selects against the most severely affected ME/CFS patients who cannot tolerate invasive procedures.

Evidence-base caveat: All post-COVID shunt studies were conducted in respiratory follow-up clinic populations (Farrow 2023) or hospitalized patients (Harbut 2023) — selected for dyspnea or severe initial infection. The shunt prevalence in unselected post-COVID and general ME/CFS populations is unknown. Sample sizes are small (n=11–26 per group in the primary studies), and none have pre-illness baseline Qs/Qt measurements. The anatomical attribution of shunt to IPAVAs specifically — as opposed to bronchopulmonary anastomoses, diffusion limitation at the capillary level, or multiple shunt channels — is semi-quantitative and subject to measurement imprecision.

Speculation: Intrapulmonary Shunt as Driver of HIF Isoform Imbalance in ME/CFS

(Certainty: 0.25 — HIF-2α endothelial deficiency model established in this paper (cert 0.55); shunt provides a parsimonious upstream driver; bidirectional causality also possible. Origin: brainstorm.)

This paper proposes that post-viral ME/CFS involves a HIF isoform imbalance: sustained HIF-2α activation in endothelial cells (HIF-2$\alpha$ Sustained Activation as the Molecular Basis for Post-Viral Endothelial Dysfunction, cert 0.50) paired with blunted HIF-1α inducibility in other tissues (HIF Pathway Inertia: Failure of Hypoxic Adaptation in ME/CFS, cert 0.50). The paper further proposes a β2AR-autoantibody-negative subtype where HIF-2α-mediated endothelial dysfunction drives tissue hypoxia independently of vasoconstriction (Sustained HIF-2$\alpha$ as a Mechanistic Explanation for $\beta_2$AR-Autoantibody-Negative Endothelial Dysfunction, cert 0.55). Intrapulmonary shunt offers a unifying upstream mechanism for this HIF imbalance: if shunt produces chronic low-grade arterial hypoxemia (PaO₂ persistently ~75–85 mmHg rather than 95–100 mmHg), this signal profile would differentially activate the endothelial HIF-2α system (which responds to chronic, mild hypoxia) over the ubiquitous HIF-1α system (which responds to acute, severe hypoxia). The shunt would thus bias the HIF isoform ratio toward HIF-2α dominance in endothelium while chronic mild hypoxia desensitizes HIF-1α signaling in other tissues — exactly the pattern proposed. (Bidirectional causality is also plausible — HIF-2α-driven endothelial dysfunction could impair HPV, worsening shunt — creating a positive feedback loop. However, the unidirectional hypothesis is testable: if shunt precedes HIF imbalance, correcting shunt should normalize HIF ratios. If the causal direction is reversed or absent, this prediction fails.)

Falsifiable prediction: ME/CFS patients with elevated Qs/Qt (>6%) show a lower nuclear HIF-2α : HIF-1α protein ratio in PBMCs compared to patients with normal Qs/Qt and healthy controls — with the direction being HIF-2α-skewed in shunt-positive patients. Correlate with PaO₂ on room air. Falsified if HIF isoform ratios do not differ by shunt status, or if the direction of skew is opposite (HIF-1α-dominant despite shunt).

Consequence: If shunt drives HIF isoform imbalance, correcting shunt (via HPV restoration or O₂ supplementation) could normalize HIF signaling across multiple tissues — providing a single upstream intervention that addresses both the endothelial and metabolic arms of ME/CFS HIF pathology.

24.1 Post-COVID ME/CFS Risk: Critical Caveats

Limitation: Four-Year Persistence May Reflect Diagnostic Delay Rather Than Sustained New-Onset Pathophysiology

The post-COVID finding that new-onset ME/CFS risk remains elevated to four years (Hadidchi et al. 2025) may partly reflect the diagnostic pathway rather than a true four-year window of newly arising disease. ME/CFS diagnosis frequently lags symptom onset by 2–5 years (diagnostic odyssey), and clinicians often code post-viral fatigue or long COVID rather than ME/CFS; a patient developing ME/CFS at 3–6 months post-infection may only receive a first ICD code years later. The EHR design cannot distinguish new onset from the delayed coding of earlier onset, nor from the unmasking of pre-existing undiagnosed mild ME/CFS. Residual surveillance bias (hospitalized patients having more follow-up encounters and thus more chance of receiving a new code) is also possible, although the observed ordering — hospitalized HR 1.46 lower than non-hospitalized HR 1.56 — argues against a pure “more healthcare contact” artifact. (Certainty of the diagnostic-delay alternative: 0.40.) The clinical implication is not “extend surveillance for four years” but “diagnose earlier at 3–6 months and shorten the diagnostic odyssey.”

Consequence: If true, part of the apparent four-year “new-onset” risk is actually delayed diagnosis of earlier-onset disease — so the public-health response should emphasize faster diagnosis (months, not years) so patients stop losing years without a label and appropriate management. (Origin: brainstorm.)

Limitation: Vaccination-Era Data Are Heavily Confounded and Must Not Be Over-Read

Post-COVID ME/CFS incidence data spanning the vaccine rollout cannot separate vaccination effects from era and selection: early rollout prioritised the elderly, healthcare workers, and the comorbid — the same groups at higher baseline ME/CFS risk — and testing/coding intensity changed across pandemic waves. Any reported relation between vaccination and ME/CFS risk in such a cohort is therefore confounded and should never be cited as evidence either that vaccination increases or decreases ME/CFS risk. The honest interpretation is that these data cannot distinguish vaccination effects from era, selection, and baseline-risk confounds. (Certainty: n/a – an evidence-quality caution, not a hypothesis.)

Consequence: Helps prevent misusing an era-confounded statistic to argue either for or against vaccination — the data are simply unable to answer the vaccination-relationship question on their own. (Origin: brainstorm.)

Limitation: Single-System Cohort Limits Generalisability of Effect Magnitudes

The four-year risk estimate rests on a single integrated health system in the Bronx (Hadidchi et al. 2025), a population with distinct socioeconomic and comorbidity profiles (high NAFLD, poverty, and chronic-disease burden). The qualitative finding (elevated risk, direction consistent with RECOVER and INSPIRE) is robust, but the specific hazard ratios (1.46–1.56) should not be treated as universal constants. (Certainty: n/a – external-validity caveat.)

Consequence: Guards against using one health system’s precise risk numbers as global forecasts — the direction of increased risk is reliable, but the exact magnitude will differ from population to population. (Origin: brainstorm.)

Open Question: Liver Function as a Risk-Stratification Variable in Post-Infectious ME/CFS

Liver disease is a reported risk modifier for post-COVID ME/CFS conversion (Hadidchi et al. 2025), and the liver is the body’s hepcidin factory – a plausible node linking hepatic dysfunction to the iron-regulatory impairment implicated in post-viral ME/CFS. Existing post-COVID and post-EBV cohort datasets (RECOVER, INSPIRE, Dubbo, Katz) already contain baseline liver-function measures; a zero-cost retrospective re-analysis could determine whether ALT/AST or FIB-4 at the time of infection predicts ME/CFS conversion, potentially yielding an immediately available risk-stratification marker. (Certainty: 0.55 – methodologically straightforward, no new data collection.)

Consequence: If liver function at the time of infection predicts who later develops ME/CFS, a routine already-available blood test could immediately flag higher-risk patients for monitoring — no new test and no extra cost. (Origin: brainstorm.)

Open Question: Does a Non-Inflammatory Acute Mechanism Trigger Post-Viral ME/CFS?

The null that acute inflammatory markers (ferritin, CRP, D-dimer, LDH) do not predict post-COVID ME/CFS conversion (Hadidchi et al. 2025) raises the question of whether the triggering mechanism is non-inflammatory — e.g. autoimmunity via molecular mimicry, viral persistence in immune-privileged sites, or a metabolic switch — rather than an acute cytokine storm. If the acute-biomarker null replicates, anti-inflammatory interventions given during acute infection would be predicted NOT to reduce ME/CFS incidence, shifting research toward post-acute mechanisms (autoantibody screening, viral-persistence testing, metabolic support). Robust to the alternative that a broader marker panel or tissue-level (not serum) inflammation is the true trigger. (Certainty of the non-inflammatory read: 0.30 – single-study null.)

Consequence: If the trigger turns out to be non-inflammatory, money spent on anti-inflammatory prevention during acute infection may be wasted — research would pivot to the post-acute window where autoimmunity, viral persistence, and metabolism are the targets. (Origin: brainstorm.)

25 Cross-Disease Research Directions

Open Question: Long-COVID ATG13 Signature Comparison

Does Long-COVID share the ATG13 circulating DAMP signature identified in ME/CFS (Gottschalk et al.), or are ATG13 elevations specific to idiopathic ME/CFS? Comparing ATG13 levels across Long-COVID patients with and without ME/CFS criteria will clarify whether abortive reactivation DAMP signaling is a shared post-infectious mechanism or specific to ME/CFS pathophysiology. Probability of clinical utility: 0.15. (Watton and Prusty 2026)

Open Question: Fibromyalgia TRPM3 Autoantibody Overlap

Fibromyalgia shares central sensitisation and autonomic dysfunction with ME/CFS (~30–50% overlap). If TRPM3 autoantibodies are present in fibromyalgia patients, this would implicate a shared ion-channel-autoantibody mechanism explaining hyperalgesia and dysautonomia across both conditions. Probability that TRPM3 autoantibodies explain fibromyalgia: 0.05. Falsifiable: fibromyalgia patients will show elevated TRPM3 autoantibodies correlating with pain sensitivity (QST thresholds). (Watton and Prusty 2026)

Open Question: POTS Endothelial Stress Response vs ME/CFS Endothelial Dysfunction

POTS is highly comorbid with ME/CFS and involves endothelial dysfunction. Are endothelial stress responses (provocative oxidation/immune challenge assays) abnormal in both POTS and ME/CFS, or is endothelial adaptive capacity specifically impaired in ME/CFS? Answering this would clarify whether endothelial dysfunction is a shared vascular substrate or a defining feature of ME/CFS pathophysiology distinct from dysautonomia alone. Probability that endothelial provocative testing distinguishes POTS from ME/CFS-POTS: 0.20. (Watton and Prusty 2026)

Open Question: Autoimmune Dysautonomia GPCR Autoantibody Overlap

Autoimmune autonomic ganglionopathy (AAG) involves GPCR autoantibodies ($\beta$-adrenergic, muscarinic) similar to those found in ME/CFS. Does ME/CFS represent a mild or early form of AAG, or a distinct dysautonomia with unique autoantibody targets (TRPM3, other ion channels)? Autoantibody repertoire comparison (protein microarray, cell-based assays) between AAG and ME/CFS cohorts would clarify whether a spectrum disorder exists. Probability that ME/CFS is an AAG-spectrum condition: 0.08. (Watton and Prusty 2026)

Speculation: Complex Regional Pain Syndrome GPCR Autoantibody Parallel

Certainty: 0.45. CRPS involves functionally active autoantibodies against $\beta_2$-adrenergic and M2 muscarinic receptors that show agonistic-like properties in cardiomyocyte bioassays (Blaes, Wallukat, et al. 2011). CRPS shares several features with ME/CFS: small fiber neuropathy, autonomic dysregulation, hyperalgesia, and post-traumatic/inflammatory onset. The key difference is that CRPS $\beta_2$/M2 AAbs are functionally agonistic (causing hyperhidrosis and sympathetic overactivity) in validated functional assays, while ME/CFS $\beta_2$ AAbs are detected by binding assays (CellTrend ELISA) and correlate with sympathovagal imbalance ($r=0.45$, $p=0.001$ (N. Azcue et al. 2026)) — leaving their functional status (agonistic, antagonistic, or non-pathogenic) undetermined. CRPS demonstrates that GPCR autoantibodies can cause autonomic and sensory dysfunction in humans (an existence proof), but does not validate the specific ME/CFS autoantibody findings because: (a) CRPS autoantibody pathogenicity was established via functional cardiomyocyte bioassays, not ELISA binding, (b) agonistic activation and sympathovagal imbalance are opposing functional effects, and (c) the same receptor target (β2-AR) could be a convergent readout of different upstream pathologies rather than a shared causal node. Both conditions have shown clinical response to immunoadsorption (Goebel 2018 for CRPS, Stein 2025 for ME/CFS), but this is equally consistent with shared GPCR-AAb pathogenicity, non-specific IgG removal benefiting any immune-mediated condition, or expectation effects in open-label designs.

Speculation: Multiple Sclerosis Sphingolipid Mechanistic Bridge

Certainty: 0.45. Probability of therapeutic cross-application: 0.03. Multiple sclerosis involves sphingolipid metabolism abnormalities — FTY720 (fingolimod) is FDA-approved for MS precisely because it modulates sphingosine-1-phosphate receptors. ME/CFS SMPDL3B and lipid raft dysfunction may share sphingolipid pathology with MS, suggesting FTY720 or related modulators could address ME/CFS lipid raft dysfunction. However, FTY720’s lymphocyte-sequestering mechanism (inducing lymphopenia) may be contraindicated in ME/CFS where immune dysfunction is already present. The cross-disease bridge is mechanistically interesting but the therapeutic translation is obstructed by the divergent immune requirements of MS (overactive autoimmunity) vs ME/CFS (impaired immune clearance). Falsifiable: ME/CFS patients will show altered sphingolipid profiles overlapping with MS-specific species, and SMPDL3B expression will correlate with sphingolipid imbalance.

Speculation: Mast Cell Activation Syndrome SMPDL3B-Lipid Overlap

Certainty: 0.40. Probability of shared mechanism: 0.04. MCAS shares autonomic, inflammatory, and endothelial features with ME/CFS. SMPDL3B-mediated lipid raft dysfunction could underlie mast cell dysregulation in both conditions — lipid raft organization governs IgE receptor (Fc$\epsilon$RI) clustering and activation thresholds, directly controlling mast cell degranulation. If SMPDL3B dysregulation is present in MCAS patients, lipid-targeted interventions (sphingolipid modulators, membrane stabilizers) could address both conditions simultaneously. However, MCAS SMPDL3B has never been measured. Falsifiable: MCAS patients will show altered SMPDL3B expression and exaggerated mast cell degranulation in response to lipid raft stressors.

26 IgM-Mediated Neuropathy Parallels

Speculation: CIDP IgM Paraproteinemic Neuropathy as a Model for ME/CFS Microvascular Neuropathy

(Certainty: 0.35.) Chronic inflammatory demyelinating polyneuropathy with IgM paraprotein targets myelin-associated glycoprotein causing complement-mediated nerve damage. The IgM-dominant autoantibody profile in long COVID (Tatai et al. 2026) suggests a potentially analogous mechanism: IgM binding to neural or microvascular endothelial antigens triggers classical complement activation, causing small fibre nerve or microvascular endothelial injury. Small fibre neuropathy is documented in a subset of ME/CFS patients. IgM paraproteinemic neuropathy treatments (rituximab, complement inhibition) show efficacy in CIDP—the same mechanisms under investigation for ME/CFS. Key prediction: ME/CFS patients with IgM-dominant profiles and small fibre neuropathy will show IgM binding to MAG and sulfatide in ELISA at higher rates than ME/CFS patients without neuropathy. Falsified if no MAG/sulfatide binding is detected. (Tatai et al. 2026).

27 Iron Redox Polarity: ME/CFS vs Long COVID Diagnostic Bifurcation

Hypothesis: Functional Iron Deficiency as a Diagnostic Differentiator Between ME/CFS and Long COVID

Certainty: 0.55. ME/CFS and Long COVID show divergent iron handling profiles with direct clinical consequences for treatment selection. In ME/CFS, the pattern is functional iron deficiency: elevated ferritin with low serum iron and low transferrin saturation (TSAT) (Świątczak et al. 2022), (Baklund et al. 2021). Serum ferritin is elevated by an average of +28.13 µg/L in ME/CFS compared to controls (Baklund et al. 2021). In Long COVID, a broader multi-compartment iron dysregulation is present: Hanson et al. demonstrated iron maldistribution with monocyte iron loading concurrent with lymphocyte iron starvation at single-cell resolution, plus stress erythropoiesis at 2 weeks post-onset that predicted PASC at 3–5 months with 72% accuracy (Hanson et al. 2024). Sonnweber et al. documented that 35% of LC patients have hyperferritinaemia at 60 days, while 24% have genuine iron deficiency and 9% have anaemia (Sonnweber et al. 2022). A direct comparative study by the Scheibenbogen group found that post-COVID ME/CFS patients showed elevated ferritin associated with hand grip strength, while post-COVID patients without ME/CFS did not (Kedor et al. 2022). Yamamoto et al. found ferritin levels of 193.0 µg/L in post-COVID patients meeting ME/CFS criteria versus 98.2 µg/L in those who did not (Yamamoto et al. 2023). (Evidence source: serum/blood — Inference target: systemic iron handling. Link is direct for serum iron parameters; indirect for tissue-level iron maldistribution — single-cell data (Hanson et al. 2024) provides the only direct evidence of monocyte-lymphocyte iron redistribution.)

The hepcidin paradox. A critical finding distinguishes ME/CFS from classic inflammatory anemia. Kavyani et al. measured hepcidin in ME/CFS — the only published hepcidin measurement in this population — and found it decreased by nearly half (Kavyani et al. 2024). This is the opposite of what the IL-6→STAT3→hepcidin pathway (documented in Hepcidin-Inflammation Axis as Endocrine-Immune Bridge) would predict for a condition with elevated IL-6. Three competing explanations exist: (1) chronic hypoxia-driven HIF-mediated hepcidin suppression overriding IL-6 signalling, (2) a compensatory attempt to mobilize sequestered iron from macrophages, or (3) a distinct ME/CFS-specific hepcidin biology where chronic upregulation exhausts the HAMP transcription machinery. In Long COVID, Gietl et al. found that patients who recovered at 12 months had higher hepcidin than those with persistent symptoms (Gietl et al. 2024) — suggesting adequate hepcidin response may be protective. The polarity is stark: low hepcidin in ME/CFS (Kavyani2023) vs variable but predictively associated with recovery in LC (Gietl2024).

Therapeutic bifurcation. This creates a dangerous therapeutic fork where treating one condition with the other’s approach may cause harm. In ME/CFS functional iron deficiency, the body already has iron — it is trapped in macrophages and unavailable to tissues. Oral iron supplementation carries risk: if ferroportin is blocked (hepcidin paradox notwithstanding), oral iron further elevates hepcidin and worsens the iron sequestration trap Rensburg et al. (2001). In Long COVID, 24% of patients have genuine iron deficiency (Sonnweber et al. 2022) and may benefit from iron repletion, but should be phenotyped first: measure full iron panel (serum Fe, ferritin, TSAT, sTfR) plus CRP before deciding between iron supplementation versus anti-inflammatory versus iron chelation. The most dangerous clinical error: administering iron to an ME/CFS patient with high ferritin (>150 µg/L) and low TSAT (<20%) because “iron is low” — this is functional iron deficiency. Iron supplementation here may fuel ferroptosis risk, as iron provides the catalytic substrate for lipid peroxidation (Sousa, Yehia, and Abulseoud 2023).

Evidence type. Observational cohort studies (clinical, not interventional). All data are human. ME/CFS-specific (Swiatczak2022, Baklund2021, Kavyani2023) and Long COVID-specific (Hanson2024, Sonnweber2022, Gietl2024, KronsteinWiedemann2024). No RCT of iron supplementation or chelation exists for either condition. (Severity coverage: mostly mild-moderate ambulatory cohorts; severe/very-severe ME/CFS iron profiles not studied.)

Falsifiable predictions. (1) Head-to-head iron panel comparison (serum Fe, ferritin, TSAT, sTfR, hepcidin, CRP, IL-6) in ME/CFS vs Long COVID patients measured in the same lab will show: ME/CFS — high ferritin, low TSAT, low hepcidin; LC — variable ferritin, normal-high TSAT, normal-high hepcidin. (2) Falsified if hepcidin levels do not differ between groups or if the direction reverses (high hepcidin in ME/CFS, low in LC). (3) Iron supplementation in ME/CFS patients with ferritin >150 µg/L and TSAT <20% increases lipid peroxidation markers (4-HNE, MDA) at 4 weeks compared to placebo. Falsified if no increase in oxidative stress markers occurs. (4) Iron chelation (deferiprone) improves fatigue scores in ME/CFS patients with ferritin >200 µg/L and TSAT <15% more than placebo. Falsified if no difference or if chelation worsens fatigue via anaemia induction.

Falsifiability. Testable. All predictions require prospective clinical studies with iron panel + hepcidin + oxidative stress measurement. Hepcidin assays are research-grade (not yet routine clinical) but commercially available. The therapeutic predictions are the strongest test: if iron supplementation helps ME/CFS functional iron deficiency patients rather than harms them, the entire model is wrong.

Consequence. If validated, this would give clinicians a simple blood test (ferritin + TSAT) and a clear decision rule — don’t give iron for functional iron deficiency — that could prevent iatrogenic harm from inappropriate iron supplementation in ME/CFS. Conversely, it would promote targeted iron repletion for the ~24% of Long COVID patients with genuine deficiency who are currently missed because clinicians assume all post-viral fatigue is the same. The polarity may also provide a laboratory tool for distinguishing ME/CFS from other post-infectious fatigue syndromes when clinical presentation alone is ambiguous.

Limitations. No head-to-head iron panel comparison study between ME/CFS and Long COVID exists — all comparisons are indirect across different studies with different labs, methods, and timepoints. Kavyani2023 is the only hepcidin measurement in ME/CFS; Gietl2024 is the only in LC (n=34). Kedor2022 grouped post-COVID ME/CFS vs classic ME/CFS but did not control for iron-specific confounders. The functional iron deficiency pattern described here (high ferritin, low TSAT) is based on ME/CFS studies that predate the hepcidin finding — whether low hepcidin paradoxically still results in functional iron deficiency via ferroportin trafficking defects or IRP/IRE regulation is unknown. (Translation gap: all evidence is human clinical, no animal-to-human gap. Competing mechanism: simple iron deficiency without anaemia — ferritin <30 µg/L — must be ruled out before applying the functional iron deficiency model. Menorrhagia, low dietary intake, and GI malabsorption are common in ME/CFS and produce genuine iron deficiency indistinguishable from functional iron deficiency on ferritin alone — TSAT distinguishes these: TSAT >20% in simple iron deficiency even at low ferritin vs TSAT <20% in functional iron deficiency at normal-high ferritin.)

(Origin: brainstorm from cynaera-gaps investigation; literature synthesis.)

Hypothesis: Post-Acute (Not Acute) Iron Dynamics as the Temporal Predictor of Post-COVID ME/CFS

Certainty: 0.40. The post-COVID epidemiological finding that none of the acute-phase biomarkers (ferritin, D-dimer, LDH, CRP) predicts which patients later develop ME/CFS (Hadidchi et al. 2025) is not a negative result — it constrains the predictive window to the post-acute transition. Under the iron-redox temporal model above, acute inflammatory hypoferremia (IL-6 → hepcidin spike) is universal and non-discriminatory, while the divergence between future ME/CFS converters and recoverers occurs in the weeks-to-months transition (functional iron deficiency emerging as hepcidin falls). The Hadidchi finding is the first large-cohort epidemiological confirmation that acute iron parameters are indeed non-predictive, narrowing the search for a converter-discriminating blood test to the post-acute period.

Alternative reading. The acute null is also consistent with iron parameters being non-discriminatory at every timepoint — i.e. iron biology may be off the causal pathway to ME/CFS conversion entirely. The post-acute prediction stands only because the iron-redox model independently predicts a post-acute divergence (iron redox polarity diagnostic bifurcation), not because the acute null itself favors a later window. This is why the falsification condition below requires 3-month (not 1-month) discrimination to succeed and why the certainty is capped at 0.40.

Falsifiable prediction. A longitudinal study measuring ferritin, TSAT, and hepcidin at 1, 3, 6, and 12 months post-COVID must show that 3-month (not 1-month) iron trajectories separate future ME/CFS converters (IOM-2015 criteria) from recoverers, e.g. AUC > 0.70. Falsified if all post-acute timepoints are non-predictive, indicating the iron divergence occurs later (>12 months) or is not predictive at all.

Severity applicability: Unknown — post-COVID incidence cohorts are not severity-stratified.

Consequence: If validated, an approx. $20 blood test (ferritin + TSAT) at the standard 3-month post-COVID follow-up could flag which patients need ME/CFS monitoring, shifting post-COVID care from passive surveillance to risk-stratified follow-up – testable within 2–3 years in a single prospective cohort. (Origin: brainstorm.)

Hypothesis: Erythrocyte Dysfunction as a Long-COVID-Specific Oxygen Transport Defect

Certainty: 0.45. Long COVID may have a distinct erythrocyte-level defect not described in ME/CFS. Kronstein-Wiedemann et al. demonstrated impaired O₂ binding to haemoglobin with increased CO binding, decreased plasma iron and TSAT, and elevated MCH in Long COVID patients compared to controls (Kronstein-Wiedemann et al. 2024). This erythrocyte dysfunction would directly impair oxygen delivery to tissues independent of mitochondrial defects — a mechanism distinct from the mitochondrial ATP production bottleneck documented in ME/CFS (Chapter Energy Metabolism and Mitochondrial Function). If validated, this provides a second axis of divergence: ME/CFS fatigue is primarily mitochondrial (ATP production failure despite adequate O₂ delivery), while Long COVID fatigue may additionally involve impaired O₂ delivery at the erythrocyte level. (Evidence source: RBC analysis — Inference target: whole-body oxygen delivery. Link is direct.)

Falsifiable prediction. ME/CFS patients will show normal O₂-Hb dissociation curves and normal CO-Hb binding, while Long COVID patients with similar fatigue severity will show impaired O₂ binding and elevated CO-Hb. Falsified if ME/CFS patients also show RBC dysfunction, or if LC patients show normal RBC parameters.

Falsifiability. Testable. Requires co-oximetry and O₂ dissociation curve measurement in matched cohorts. A single-center study with n=30 per group is sufficient.

Consequence. A distinct erythrocyte-level defect in Long COVID but not ME/CFS would open a second treatment pathway — interventions targeting RBC oxygen delivery (e.g., pyruvate, which modulates Hb-O₂ affinity) — that would help Long COVID patients but could harm ME/CFS patients if misapplied.

Limitations. KronsteinWiedemann2024RBC is a single study with moderate sample size; independent replication is needed. No ME/CFS erythrocyte dysfunction data exist. The CO-Hb finding may reflect smoking status or environmental exposure rather than pathophysiological CO production. (Origin: literature integration.)

Speculation: Ferroptosis Susceptibility as a Shared Mechanism with Divergent Triggers

Certainty: 0.35. Ferroptosis — iron-dependent lipid peroxidation cell death — is a candidate shared mechanism in both ME/CFS and Long COVID, but triggered by different upstream pathology. In Long COVID, the trigger is inflammatory iron overload: COVID-induced oxidative stress combined with monocyte iron loading (Hanson et al. 2024) creates an iron-rich, high-ROS environment that depletes GPX4 and triggers lipid peroxidation (Sousa, Yehia, and Abulseoud 2023). In ME/CFS, the trigger may instead be chronic low-grade oxidative stress plus impaired antioxidant capacity (including glutathione depletion, documented in Chapter Energy Metabolism and Mitochondrial Function) in the context of functional iron deficiency — the iron is trapped in macrophages but ferroptosis-competent iron may be redistributed to susceptible tissues via non-transferrin-bound iron (NTBI) during oxidative bursts. The net result (ferroptotic cell death) is shared; the upstream supply of catalytic iron differs. (Evidence source: serum iron parameters + ferroptosis biomarkers in LC — Inference target: tissue-level ferroptosis. Link is indirect — no tissue-level ferroptosis measurement in either condition.)

Falsifiable prediction. Lipid peroxidation markers (4-HNE, MDA, F₂-isoprostanes) will be elevated in both ME/CFS and Long COVID compared to controls, but will correlate with different iron parameters: with serum iron/TSAT elevation in LC, with IL-6/CRP elevation in ME/CFS. Falsified if neither group shows elevated lipid peroxidation, or if the iron parameter correlations are absent/null in both.

Falsifiability. Testable but requires ferroptosis-specific biomarkers (4-HNE, MDA, GPX4 activity, ACSL4 expression) which are research-grade. The mechanistic specificity (lipid peroxidation product identity distinguishing ferroptosis from generic oxidative stress) is currently weak.

Consequence. Ferroptosis as a shared mechanism with divergent triggers would explain why both conditions show oxidative stress and why anti-ferroptosis interventions (CoQ10, vitamin E, GPX4 enhancers) are already in the ME/CFS treatment armamentarium with anecdotal benefit — but would also caution against iron supplementation in LC patients with hyperferritinaemia who may already be at ferroptosis risk.

Limitations. The ferroptosis-ME/CFS connection is completely inferential — no ME/CFS study has measured ferroptosis-specific biomarkers. The LC ferroptosis literature is review-level (Sousa, Yehia, and Abulseoud 2023), not primary interventional data. Mantle 2025 (Mantle et al. 2025) reviews ferroptosis across multiple conditions but provides no ME/CFS-specific evidence. The connection between functional iron deficiency and ferroptosis is mechanistically paradoxical — trapped iron should reduce ferroptosis risk, not increase it — and requires the NTBI redistribution hypothesis for coherence. (Origin: literature synthesis + brainstorm.)

Open Question: Iron Chelation vs Iron Supplementation: When to Use Which and in Whom

Central question. The therapeutic bifurcation model predicts that ME/CFS patients with functional iron deficiency (ferritin >150 µg/L, TSAT <20%) should NOT receive iron supplementation and may benefit from iron chelation, while Long COVID patients with genuine iron deficiency (ferritin <30 µg/L or TSAT <16%) should receive iron. This model is entirely mechanistic — zero interventional data exist. The following questions require prospective clinical trials:

Does oral iron supplementation increase or decrease fatigue in ME/CFS patients stratified by baseline ferritin and TSAT? (2) Does iron chelation (deferiprone, deferasirox) improve fatigue and physical function in ME/CFS patients with functional iron deficiency? (3) Do the ~24% of Long COVID patients with genuine iron deficiency (Sonnweber et al. 2022) benefit from iron repletion, and does the ~35% with hyperferritinaemia worsen? (4) Does the hepcidin level predict response to iron supplementation in either condition — specifically, do low-hepcidin patients (ME/CFS, (Kavyani et al. 2024)) respond differently than normal-high hepcidin patients (recovering LC, (Gietl et al. 2024))?

Until these questions are answered with interventional data, clinicians should: (a) measure full iron panel (Fe, ferritin, TSAT, sTfR, Hb, CRP) before prescribing iron for post-viral fatigue; (b) treat genuine iron deficiency (ferritin <30 µg/L or TSAT <16%) with iron regardless of MC/CFS or LC diagnosis; (c) do NOT supplement iron in patients with ferritin >150 µg/L and TSAT <20% (functional iron deficiency pattern) without clear evidence of benefit; (d) monitor ferritin and TSAT at 4 weeks if supplementing — no rise in ferritin suggests hepcidin blockade and may warrant IV iron consideration (per existing sec-12 Iron entry).

Consequence. Answering this question would transform post-viral fatigue management: a simple blood test (ferritin + TSAT) would tell clinicians, patients, and researchers whether iron should be supplemented, withheld, or chelated — replacing the current default of “try iron, it might help” with evidence-based decision rules. Until trials exist, the precautionary principle applies: don’t supplement iron in functional iron deficiency.

(Origin: brainstorm from cynaera-gaps investigation.)

28 Additional Iron Redox Speculations (Brainstorm Origin)

Speculation: Hepcidin-Independent Ferroportin Blockade: Four Candidate Mechanisms Resolving the Hepcidin Paradox

Certainty: 0.40. The hepcidin paradox — low hepcidin (Kavyani et al. 2024) yet persistent functional iron deficiency (high ferritin, low serum iron, low TSAT) (Świątczak et al. 2022), (Baklund et al. 2021) — requires a hepcidin-independent explanation for why iron remains trapped in macrophages. Four candidate mechanisms, none mutually exclusive and all untested in ME/CFS, could resolve the paradox:

Ferroportin post-translational trafficking defect. Even with low hepcidin, ferroportin may fail to traffic to the cell membrane. IRP1/IRP2 (iron regulatory proteins) control ferroportin translation via the iron-responsive element (IRE) in ferroportin’s 5’ untranslated region. If IRP1 is aconitase-inactive — as the ACO1/IRP1-telomere correlation in fibrotic Long COVID suggests (Mulet et al. 2026) — ferroportin translation could be suppressed independently of hepcidin. In this model, the IRP/IRE iron-sensing system that sits upstream of hepcidin is the primary defect; hepcidin reduction is a compensatory response to intracellular iron deficiency.
Ceruloplasmin/ferroxidase deficiency. Iron export via ferroportin requires concurrent ferroxidation (Fe²⁺ → Fe³⁺) by ceruloplasmin (GPI-anchored or soluble) for loading onto transferrin. If ferroxidase activity is impaired — due to copper deficiency, ceruloplasmin autoantibodies, or oxidative inactivation — iron exits the cell but cannot bind transferrin, re-entering via non-specific pathways. The iron is “exported” but not “delivered,” producing low TSAT despite functioning ferroportin.
Macrophage iron retention via ferritinophagy blockade. NCOA4-mediated ferritinophagy releases stored iron from ferritin for export. If NCOA4 is suppressed (e.g., by chronic mTOR activation, documented in ME/CFS immune cells), ferritin-sequestered iron cannot be mobilized for export even with open ferroportin channels. The macrophage is simultaneously iron-rich (ferritin-loaded) and iron-poor (unavailable for export).
Lipocalin-2 (LCN2/NGAL) iron scavenging. LCN2 binds siderophore-iron complexes and delivers them to cells via the 24p3R receptor, bypassing the transferrin system entirely. Dufrusine et al. found LCN2 altered in Long COVID proteomics (Dufrusine et al. 2022). If LCN2 is elevated in ME/CFS, it could scavenge any iron that does get exported, delivering it right back to macrophages — a futile iron cycle maintaining functional deficiency.

Under this model, low hepcidin in ME/CFS is not a paradox — it is the body’s appropriate compensatory response to hepcidin-independent iron trapping. The liver senses tissue iron deficiency (via HFE/TfR2/HJV) and appropriately suppresses hepcidin. But the iron remains trapped because the downstream export machinery is broken. The thermostat (hepcidin) is set correctly; the furnace (iron export) is what fails.

Falsifiable predictions. (a) ME/CFS monocytes will show normal or elevated ferroportin protein (western blot) but reduced cell-surface ferroportin (biotinylation/flow cytometry) — trafficking defect. (b) ME/CFS serum ceruloplasmin ferroxidase activity (pPD oxidation assay) will be reduced and correlate inversely with TSAT. (c) ME/CFS monocyte NCOA4 protein will be reduced (western blot), ferritin will be elevated, and labile iron pool (calcein-AM) will be low — ferritinophagy blockade. (d) ME/CFS serum LCN2 will be elevated and correlate positively with ferritin and negatively with TSAT. Falsified if: all four candidate mechanisms are normal in ME/CFS — then the hepcidin paradox has no molecular resolution and the measured low hepcidin may be a measurement artifact or transient state.

Falsifiability. Testable but requires research-grade assays (IRP activity gels, ferroportin surface biotinylation, ceruloplasmin ferroxidase assay, NCOA4 western blot). None are routine clinical tests. A research study with n=30 ME/CFS vs matched controls is sufficient.

Consequence. The body’s iron-regulating thermostat is correctly calling for iron release — but the release machinery is broken. If validated, this shifts treatment from “lower hepcidin to release iron” (danazol, per Danazol/Hepcidin Antagonism for Iron Redistribution) to “fix the export machinery downstream of hepcidin.” Clinically, this makes deferiprone (iron chelator, see deferiprone functional iron deficiency) a more rational target than hepcidin antagonism — release iron that is trapped, rather than trying to change the setpoint of a thermostat that is already calling for release.

Limitations. All four candidate mechanisms are untested in ME/CFS. The model is entirely inferential from general iron biology and Long COVID proteomics. Ferroportin trafficking defects due to IRP dysfunction are precedented in hereditary hyperferritinemia-cataract syndrome but not in ME/CFS. Ceruloplasmin deficiency produces a known clinical syndrome (aceruloplasminemia) with neurodegeneration and diabetes — features not prominent in ME/CFS. NCOA4 regulation in macrophages is incompletely understood. The mTOR→NCOA4 link is plausible but untested. (Translation gap: IRP/ferroportin biology is well-established in model systems; human ME/CFS macrophage data are absent. Competing mechanism: the hepcidin measurement itself may be artefactual — Kavyani2023 is the only measurement and needs independent replication before the paradox is accepted as real.)

(Origin: brainstorm.)

Speculation: Temporal Phase-Shift Hypothesis: ME/CFS as Late-Stage Long COVID Iron Phenotype

Certainty: 0.35. The “polarity” between ME/CFS and Long COVID iron handling may not reflect disease-specific divergence but a shared trajectory sampled at different timepoints. Under this model, both conditions traverse the same iron dysregulation trajectory:

Phase	Time post-trigger	Iron phenotype	Dominant mechanism
0 (acute)	Days 0–30	Hypoferremia (IL-6 → hepcidin spike → iron withholding)	Innate immune iron sequestration
I (early post-acute)	1–6 months	Stress erythropoiesis, iron maldistribution, ferritin elevation, variable hepcidin	Hanson2024 pattern: compensatory erythroid iron demand + inflammatory trapping
II (established chronic)	6–24 months	Functional iron deficiency: high ferritin, low TSAT, declining hepcidin	IRP/ferroportin post-translational defects, ferritinophagy blockade
III (late chronic, > 2 years)	> 24 months	Low hepcidin, persistent functional iron deficiency, ferroptosis susceptibility	Exhausted hepcidin machinery, established NTBI redistribution

In this framework, Long COVID research disproportionately samples Phases 0–I (1–6 month follow-up, per Sonnweber2022, Hanson2024, Gietl2024), while ME/CFS research samples Phases II–III (years-to-decades of established disease, per Swiatczak2022, Baklund2021, Kavyani2023). The “polarity” (hepcidin-dependent in LC, hepcidin-independent in ME/CFS) is temporal, not diagnostic. The therapeutic bifurcation persists but for a different reason: Phase I patients may benefit from iron supplementation (if genuinely deficient per TSAT), while Phase II–III patients require iron redistribution rather than supplementation — regardless of whether the diagnosis is ME/CFS or late-stage Long COVID.

Falsifiable prediction. A longitudinal iron panel study in a post-COVID cohort followed for ≥2 years should show: hepcidin declining progressively from elevated (months 1–3) to normal (months 6–12) to low (months 18–24), while ferritin remains elevated and TSAT remains depressed as patients transition from LC to post-COVID ME/CFS. Falsified if: hepcidin trajectory does not show this progressive decline, or if a subset of long-duration LC patients maintain elevated hepcidin beyond 2 years.

Falsifiability. Testable with existing or planned LC cohort data (LIINC, RECOVER, PHOSP-COVID) — iron panels and hepcidin measured at multiple timepoints to ≥24 months. A dedicated prospective study with n≥100, sampled every 3 months for 24 months, is sufficient to distinguish temporal from diagnostic hypotheses.

Consequence. If the temporal model is correct, the iron phenotype at a given timepoint tells you how long the disease has been active — not which disease it is. This collapses the “ME/CFS vs LC treatment bifurcation” into a single trajectory where treatment choice depends on iron phase, not diagnosis: supplement iron in Phase I deficiency, chelate or redistribute in Phase II–III functional trapping.

Limitations. Longitudinal iron data in post-COVID cohorts is sparse beyond 12 months. Kavyani2023 is cross-sectional (established ME/CFS), not longitudinal; the low-hepcidin state’s onset time is unknown. The temporal model predicts that all LC patients will eventually develop functional iron deficiency with low hepcidin — a strong claim untested by any study beyond 12 months. Most LC recovery occurs within 12 months, reducing the cohort size available for Phase II–III measurement. (Origin: brainstorm.)

Speculation: Monocyte Epigenetic Reprogramming as a Ferroptosis Amplifier in ME/CFS

Certainty: 0.30. Hanson et al. demonstrated that monocytes in Long COVID are iron-loaded while lymphocytes are iron-starved (Hanson et al. 2024). If this monocyte iron loading persists chronically — and the same pattern occurs in ME/CFS — the iron-loaded monocyte may undergo epigenetic reprogramming toward a pro-ferroptotic phenotype. Iron-loaded monocytes, on entering tissues and differentiating into macrophages, could deliver concentrated iron to tissue microenvironments, sensitizing resident cells to ferroptosis. The reprogramming would be self-reinforcing: monocyte iron loading → tissue iron deposition → ferroptosis → release of DAMPs (HMGB1, S100A8/A9) → further monocyte recruitment → more iron delivery. This constitutes a monocyte-mediated ferroptosis amplification loop that links the peripheral iron phenotype to tissue-level damage.

Falsifiable prediction. ME/CFS circulating monocytes (CD14⁺) will show elevated labile iron pool (calcein-AM quenching) and elevated ferroportin mRNA but reduced ferroportin surface protein (consistent with post-translational blockade). Tissue biopsies (skin, muscle) in ME/CFS will show elevated iron deposition (Perls’ stain) and 4-HNE (lipid peroxidation adduct) co-localizing with CD68⁺ macrophages, in regions adjacent to CD31⁺ microvasculature. Falsified if: ME/CFS monocytes show normal or low labile iron pool, or if tissue biopsies show no iron deposition or ferroptosis markers.

Falsifiability. Testable but requires invasive tissue biopsies (skin punch, optional muscle) for the tissue-level component and research-grade flow cytometry for the monocyte iron component. Existing ME/CFS biopsy studies (SFN skin biopsies) could be re-stained for iron and ferroptosis markers without new collection — using archived tissue.

Consequence. For researchers: explains how a “blood-only” iron finding (high serum ferritin) translates to tissue damage — monocytes are the couriers, delivering trapped iron to tissues. For drug developers: identifies monocyte iron unloading as a therapeutic target (deferiprone monocyte-penetrant, iron chelators targeting the circulating monocyte pool before tissue entry) — conceptually analogous to the NETosis pathway where extracellular traps deliver autoantigens to tissues.

Limitations. Monocyte epigenetic reprogramming in response to iron loading is demonstrated in hemochromatosis and thalassemia, not in ME/CFS. The DAMP→monocyte recruitment loop is established in atherosclerosis (iron-loaded foam cells → ferroptosis → plaque rupture) but not in ME/CFS-affected tissues. No ME/CFS study has co-localized iron, ferroptosis markers, and macrophage markers in any tissue. The link to PEM (post-exertional triggering of the amplification loop) is entirely speculative. (Origin: brainstorm.)

Prediction: Ferritin:TSAT Ratio as a Diagnostic Differentiator and Treatment-Stratification Biomarker

Certainty: 0.40. (This is NOT a validated biomarker — it is a testable prediction, reinforced by a large, replicable internal computation.) For a computationally tractable clinical tool usable in any primary care setting, the ferritin-to-TSAT ratio (FTR, ferritin in µg/L divided by TSAT in %) may distinguish functional iron deficiency (FTR > 10, i.e., ferritin 150 µg/L with TSAT <15%) from genuine iron deficiency (FTR < 5, ferritin 30 µg/L with TSAT > 6%) with sufficient accuracy for pragmatic clinical decision-making. Intermediate FTR values (5–10) would indicate mixed or indeterminate phenotype.

The rationale is computational: ferritin and TSAT move in opposite directions in functional iron deficiency (ferritin up, TSAT down) vs genuine iron deficiency (ferritin down, TSAT variable). Their ratio amplifies the signal compared to either measure alone. In ME/CFS, the predicted FTR is > 10 (high ferritin, low TSAT). In post-COVID patients without ME/CFS, predicted FTR is variable and driven by the ferritin elevation pattern (hyperferritinaemia with preserved TSAT → FTR intermediate).

Falsifiable prediction. In a head-to-head comparison of ME/CFS (diagnosed $>$ 2 years) vs post-COVID-without-ME/CFS (6–24 months) vs healthy controls, measured in the same laboratory: (a) Median FTR in ME/CFS $>$ 10 (ferritin $>$ 100 µg/L, TSAT $<$ 15%), vs post-COVID $<$ 8 (ferritin variable, TSAT $>$ 15%), vs controls $<$ 5. (b) Receiver operating characteristic (ROC) analysis: FTR discriminates ME/CFS from post-COVID without ME/CFS with AUC $>=$ 0.75. (c) FTR discriminates ME/CFS from healthy controls with AUC $>=$ 0.80. Falsified if: FTR distributions overlap extensively between groups (AUC $<$ 0.65), or if ferritin and TSAT both trend in the same direction in ME/CFS (both elevated or both depressed).

Falsifiability. Testable. Requires iron panel (Fe, ferritin, TSAT, sTfR) in the three comparator groups. A single-center study with n=50 per group provides power ≥ 0.80 to detect an AUC ≥ 0.75. The computation (ferritin/TSAT) is trivial — any clinician or spreadsheet can perform it. The barrier is not measurement complexity but the absence of head-to-head comparative data.

Consequence. A simple blood test ratio (FTR) tells clinicians whether a post-viral fatigue patient has functional iron deficiency (ferritin high, TSAT low → don’t supplement iron) or genuine iron deficiency (ferritin low → supplement iron) — replacing the dangerous default “iron might help.” The ratio requires only a standard iron panel ($20–50), not hepcidin, making it deployable in any primary care setting worldwide. If validated, this changes iron prescribing for every post-viral fatigue patient.

Limitations. No study has computed or validated FTR in any post-viral fatigue population. The cutoffs (FTR > 10, < 5) are inferred from published ferritin and TSAT means and are not empirically established. Acute illness, recent blood transfusion, and recent iron supplementation all confound FTR interpretation. The ratio is sensitive to lab-specific ferritin reference ranges. Sex-specific cutoffs (women have lower baseline ferritin) have not been determined. FTR is a second-generation refinement of an untested first-order claim — the “no clinical action” choice is simpler but loses the actionable information the ratio provides. (Origin: brainstorm.)

Speculation: Deferiprone for Functional Iron Deficiency in ME/CFS: Rationale and Caution

Certainty: 0.30. Deferiprone (3-hydroxy-1,2-dimethylpyridin-4-one) is a membrane-permeable oral iron chelator approved for transfusional iron overload in thalassemia. Its properties make it a candidate for ME/CFS functional iron deficiency where hepcidin antagonism may be misdirected (see hepcidin independent ferroportin blockade): deferiprone chelates intracellular labile iron (the iron pool available for ferroptosis catalysis) and is small and lipophilic, penetrating cell membranes without requiring ferroportin-mediated export. Unlike hepcidin antagonists (danazol, Danazol/Hepcidin Antagonism for Iron Redistribution), deferiprone does not depend on functional hepcidin or ferroportin machinery — it removes iron directly from the cytosol, bypassing the entire hepcidin-ferroportin axis.

Rationale for ME/CFS. If the hepcidin-independent ferroportin blockade model is correct, the trapped iron is inside macrophages and enterocytes, inaccessible to oral supplementation and resistant to hepcidin modulation. Deferiprone enters those cells directly and chelates the trapped labile iron pool, forming a deferiprone-iron complex that is excreted in urine. This would reduce macrophage iron loading (lowering ferroptosis substrate), reduce the labile iron pool available for lipid peroxidation, and potentially improve iron availability by reducing the futile LCN2 iron-scavenging cycle if present. The endpoint is reduced oxidative stress and ferroptosis risk, not increased serum iron (deferiprone chelates and excretes, it does not redistribute).

Safety concerns (MANDATORY caveats). Deferiprone carries black-box warnings for agranulocytosis and neutropenia — severe, potentially fatal infections. Weekly ANC monitoring is mandatory during treatment. Liver function requires monitoring. Deferiprone is teratogenic — pregnancy contraindicated. It chelates zinc as well as iron, requiring zinc supplementation. It is FDA-approved only for transfusional iron overload in thalassemia, not for ME/CFS or any post-viral fatigue condition. There is ZERO human data on deferiprone in ME/CFS. The dosages used in thalassemia (75–100 mg/kg/day) produce substantial iron removal; whether lower doses could produce milder iron redistribution without neutropenia risk is unknown.

Hypothetical pilot design. A small open-label safety pilot (n=10–15 ME/CFS patients with ferritin >150 µg/L and TSAT <20%, confirmed functional iron deficiency, 4-week treatment at the lowest effective thalassemia dose, with weekly ANC and monthly liver/kidney function monitoring, plus pre/post iron panel, lipid peroxidation markers (4-HNE, MDA), and fatigue scales. The primary endpoint is safety (no grade ≥3 adverse events). The secondary endpoint is evidence of biological activity (reduction in ferritin and lipid peroxidation markers). This is a safety-only pilot — efficacy cannot be determined from an open-label n=10.

Falsifiable prediction. A 4-week deferiprone course in ME/CFS patients with functional iron deficiency will reduce serum ferritin (by ≥30%), reduce 4-HNE (by ≥20%), and reduce monocyte labile iron pool without inducing anaemia (Hb decrease < 1 g/dL). Falsified if: ferritin does not decrease (drug ineffective at iron chelation in this population), or Hb decreases by > 1 g/dL (iron removal causes anaemia — net iron deficit, not just redistribution), or ≥1 patient develops grade ≥3 neutropenia (unacceptable safety profile).

Falsifiability. Testable in a small pilot but ethically complex — the agranulocytosis risk requires careful risk-benefit analysis and institutional ethics board approval. The null result (no ferritin decrease) would close the deferiprone hypothesis but leave the broader iron chelation question open (deferasirox or deferoxamine may have different properties).

Consequence. For researchers: establishes whether direct iron chelation — rather than hepcidin manipulation — is the right approach for functional iron deficiency in ME/CFS. For patients: this is a research-only drug with serious risks (black-box warning for fatal infections). Self-experimentation with deferiprone is DANGEROUS. Weekly blood monitoring by a physician is mandatory. This is not a supplement — it is a prescription iron chelator with potentially fatal adverse effects.

Limitations. Zero human data in any post-viral fatigue condition. The theoretical rationale (hepcidin-independent ferroportin blockade → deferiprone bypasses the block) depends on a model that is itself entirely untested in ME/CFS. Deferiprone’s membrane permeability is advantageous for intracellular iron chelation but also means it can penetrate the blood-brain barrier — CNS iron chelation is therapeutically untested in ME/CFS and could have unanticipated neurological effects. The drug is expensive ($50,000-100,000/year at thalassemia doses). Lower dosing for investigational ME/CFS use is unexplored. (Translation gap: deferiprone is a prescription drug with established safety/efficacy in thalassemia; the translation gap is thalassemia → ME/CFS, not animal→human. The safety profile in ME/CFS is entirely unknown.)

(Origin: brainstorm.)

Limitation: Null Hypotheses Constraining the Iron Redox Polarity Model

The iron redox polarity model rests on several assumptions that may fail empirical testing. The null hypotheses constraining the model are:

N1 — Measurement artifact null. The Kavyani2023 low-hepcidin finding (Kavyani et al. 2024) is a single measurement from one lab; hepcidin assays are notoriously variable (inter-assay CV 10–20% for ELISA, batch effects in mass spectrometry). If independent replication finds normal or elevated hepcidin in ME/CFS, the entire hepcidin paradox disappears, and the classic IL-6→hepcidin→ferroportin→functional iron deficiency model is restored. Certainty of N1: 0.55 — this is the most plausible null because independent replication of the single measurement has the highest prior probability of changing the model.

N2 — Patient selection null. The ME/CFS iron phenotype (high ferritin, low TSAT) reflects selection bias: the patients who are well enough to enroll in observational studies are a biased subset (ambulatory, mild-moderate). Severely affected ME/CFS patients — bedbound or housebound — have not had iron panels measured. If severe ME/CFS patients have normal iron profiles, the pattern is severity-specific, not universally diagnostic. Certainty: 0.25 (plausible but untested).

N3 — Confounding-by-comorbidity null. The elevated ferritin in ME/CFS reflects undiagnosed comorbid autoimmune conditions (Sjögren’s, SLE, RA) or chronic infections (EBV reactivation, tick-borne) that independently raise ferritin as an acute-phase reactant, rather than reflecting a primary iron handling defect. The high ferritin is not “functional iron deficiency” — it is standard inflammatory anaemia with undiagnosed comorbidity as the driver. Certainty: 0.30 (plausible — comorbidity is widespread in ME/CFS and underdiagnosed).

N4 — Temporal-recovery null. The ME/CFS iron phenotype is transient and resolves with clinical improvement, independent of any iron-specific intervention. The high ferritin/low TSAT pattern is a state marker (like CRP in acute infection), not a trait marker (like genetics). When patients recover, iron normalizes. This is consistent with the temporal phase-shift model (temporal phase shift iron) but collapses the “therapeutic bifurcation” into “wait and monitor.” Certainty: 0.20 (no longitudinal ME/CFS iron data to test this).

N5 — Therapeutic-indifference null. Neither iron supplementation nor iron chelation affects fatigue outcomes in ME/CFS or Long COVID when measured in an adequately powered RCT. The ferritin/TSAT pattern is a biomarker with no clinical consequence — it reflects underlying biology but targeting it with iron-modifying interventions does not change outcomes. Certainty: 0.40 (the prior probability that any single intervention improves ME/CFS fatigue in an RCT is low, based on the history of negative ME/CFS treatment trials).

N6 — Common-pathway null. The iron phenotype is downstream of a shared pre-iron mechanism (chronic inflammation, mitochondrial dysfunction, NAD depletion) and is therefore a biomarker of upstream pathology, not a distinct disease mechanism. Fixing the iron problem without fixing the upstream driver produces no sustained benefit. Certainty: 0.50 — this is the consensus mechanistic model of anaemia of chronic disease, and the prior probability of a downstream iron-specific defect producing sustained clinical benefit is low without addressing the upstream cause.

Limitation. Of these six nulls, N1 (measurement artifact) is the most immediate and testable. A second hepcidin measurement in ME/CFS — ideally with mass spectrometry rather than ELISA, in an independent cohort with matched controls — would take <6 months and cost \<$50,000. Until N1 is resolved, all downstream mechanistic models (hepcidin-independent blockade, temporal phase shift, ferroptosis amplification) are contingent claims built on an un-replicated single measurement. The evidence quality concerns detailed in the brainstorm (Q1–Q9 in ops/brainstorms/brainstorm-iron-redox-polarity-2026-07-26.md) apply to the entire iron redox polarity model.

Consequence. For researchers: resolve the hepcidin measurement question first — it is the most tractable, highest-priority uncertainty. For clinicians: use the iron panel (ferritin + TSAT) for clinical decisions regardless of hepcidin level — ferritin+TSAT tells you whether the patient has functional iron deficiency without needing hepcidin. The polarity model functions at the clinical-decision level (don’t supplement iron for high ferritin + low TSAT) even if the hepcidin mechanism is wrong.

(Origin: brainstorm — null hypotheses and evidence quality concerns, critical categories 11 and 12.)

Synthesis: The Iron Redox Polarity Model — A Convergent Diagnostic Bifurcation Across Four Independent Levels

Iron handling in ME/CFS and Long COVID is not a shared mechanism — it is a diagnostic bifurcation with potentially opposite therapeutic implications, supported by four independent levels of evidence. At the clinical epidemiology level, serum ferritin distinguishes post-COVID ME/CFS from non-ME/CFS Long COVID in two independent cohorts (Kedor2022: ferritin 193 vs 98 µg/L, (Kedor et al. 2022); Yamamoto2023: ferritin correlated with fatigue scores, (Yamamoto et al. 2023)), though the Okayama group’s second paper ((Morita2024ME?)-CFSphase) shares the same patient stream and is not an independent replication. At the molecular regulatory level, the Kavyani2023 hepcidin measurement — the only published in ME/CFS — challenges the classic hepcidin model: hepcidin is low (~50% reduction (Kavyani et al. 2024)), not high, yet functional iron deficiency persists (hepcidin independent ferroportin blockade, cert 0.40). This paradox drives the search for hepcidin-independent trapping mechanisms (ferroportin trafficking, ferritinophagy blockade, ceruloplasmin deficiency, LCN2 scavenging). At the single-cell compartment level, Hanson2024 demonstrates monocyte iron loading concurrent with lymphocyte iron starvation in Long COVID (Hanson et al. 2024) — a maldistribution pattern with no equivalent measurement in ME/CFS. At the erythrocyte functional level, Kronstein-Wiedemann2024 documents impaired RBC oxygen binding in LC (Kronstein-Wiedemann et al. 2024) — a defect not described in ME/CFS (erythrocyte dysfunction lc specific, cert 0.45). The convergent clinical implication is a simple, testable rule: measure ferritin + TSAT before prescribing iron for post-viral fatigue. Functional iron deficiency (ferritin $>$ 150 µg/L, TSAT $<$ 20%) = do NOT supplement iron — the body has iron, it’s just trapped (ferritin tsat ratio diagnostic, cert 0.40). Genuine iron deficiency (ferritin $<$ 30 µg/L or TSAT $<$ 16%) = supplement iron regardless of diagnosis. The four hepcidin-independent trapping mechanisms and the iron chelation hypothesis (deferiprone functional iron deficiency, cert 0.30) remain purely mechanistic and untested. The single most tractable uncertainty is N1 — whether the Kavyani2023 low-hepcidin finding replicates in an independent ME/CFS cohort (iron redox null hypotheses). A second measurement using mass spectrometry (not ELISA) in an independent cohort with matched controls would cost <$50,000 and take $<$ 6 months — and would settle whether the hepcidin paradox is biology or artifact. Until then, the clinical bifurcation model functions at the blood-test level regardless of mechanism: ferritin + TSAT tells you what to do, even if we don’t yet know why the thermostat is broken.

Consequence. A simple, cheap blood test (ferritin + TSAT, $20–50, available worldwide) can prevent iatrogenic harm from inappropriate iron supplementation in post-viral fatigue — while simultaneously identifying the ~24% of Long COVID patients who have genuine iron deficiency and will benefit from targeted repletion. This changes the default clinical reflex from “try iron, it might help” to “measure first, treat the phenotype.”

References

Aditi, Thomas Blackwell, Xiang Fang, Surendra Sharma, Matthew Mendoza, Alan Landay, Georgiy Golovko, and Parimal Samir. 2026. “Long-Term Risk of Dementia Following Encephalitis: A Large-Scale Retrospective Cohort Study of Electronic Health Records.” Journal of Neurology 273 (9): 511. https://doi.org/10.1007/s00415-026-14017-3.

Almahdi, MA, A Mashi, AY Shaikh, MM Alahmadi, MA Alalyani, MS Alghamdi, and SA Hamzi. 2026. “Palliative and Supportive Care Needs in Advanced Hidradenitis Suppurativa: A Systematic Review.” Cureus 18 (5): e108573. https://doi.org/10.7759/cureus.108573.

Almutairi, Mohammed M, Abdulrahman Althekair, Fahad Almutairi, Mohammed Alatabani, and Abdulaziz Alsaikhan. 2024. “Mitochondrial Dysfunction and Mitophagy in ADHD: Cellular and Molecular Mechanisms.” Saudi Pharmaceutical Journal 32 (12): 102212. https://doi.org/10.1016/j.jsps.2024.102212.

Andreoli, Laura, and Angela Tincani. 2017. “Undifferentiated Connective Tissue Disease, Fibromyalgia and the Environmental Factors.” Current Opinion in Rheumatology 29 (4): 380–86. https://doi.org/10.1097/BOR.0000000000000392.

Arcani, Robin, Elisabeth Jouve, Laurent Chiche, and Noémie Jourde-Chiche. 2023. “Categorization of Patients with Systemic Lupus Erythematosus Using Disease Activity, Patient-Reported Outcomes, and Transcriptomic Signatures.” Clinical Rheumatology 42: 2061–70. https://doi.org/10.1007/s10067-023-06525-8.

Arendt, Christof T., Sven Klinsing, Elena Becke, Mena Fahim, et al. 2026. “Subacute Frontoinsular-Cingulate Tract Changes in Unvaccinated COVID-19 Survivors: A Tract-Based Spatial Statistics Study of Diffusion Tensor Imaging.” Brain Research Bulletin 242: 111986. https://doi.org/10.1016/j.brainresbull.2026.111986.

Asherson, Philip, and Jessica Agnew-Blais. 2019. “Annual Research Review: Does Late-Onset Attention-Deficit/Hyperactivity Disorder Exist?” Journal of Child Psychology and Psychiatry 60 (4): 333–52. https://doi.org/10.1111/jcpp.13020.

Aslan, Cihan, Bahadır Konuşkan, Burçin Şener, and Fatih Ünal. 2021. “Comparison of Serum Anti-Neuronal Antibody Levels in Patients Having Autism Spectrum Disorder with and Without Regression.” The Turkish Journal of Pediatrics 63 (5): 780–89. https://doi.org/10.24953/turkjped.2021.05.006.

Azcue, Néstor et al. 2025. “Small Fiber Neuropathy in the Post-COVID Condition and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Clinical Significance and Diagnostic Challenges.” European Journal of Neurology 32 (2): e70016. https://doi.org/10.1111/ene.70016.

Azcue, Néstor, Rocio Del Pino, Miriam Acera, Teresa Fernández-Valle, Mayte Ayo-Gonzalez, Juan Carlos Gómez-Esteban, Beatriz Ibañez, David Otaegui, María Sanchez-Rodriguez, and Iñigo Gabilondo. 2023. “Dysautonomia and Small Fiber Neuropathy in Post-COVID Condition and Chronic Fatigue Syndrome.” Journal of Translational Medicine 21 (1): 814. https://doi.org/10.1186/s12967-023-04678-3.

Azcue, N., A. Prada, R. Del Pino, M. Acera, T. Fernández-Valle, N. Ayo-Mentxakatorre, T. Pérez-Concha, et al. 2026. “Involvement of Autoantibodies Against G Protein-Coupled Receptors in Post-COVID Condition and Chronic Fatigue Syndrome.” Scientific Reports 16. https://doi.org/10.1038/s41598-026-49131-9.

Azcue, N., A. Sisto, L. Guglieri, and et al. 2022. “Brain Fog of Post-COVID-19 Condition and Chronic Fatigue Syndrome, Same Medical Disorder?” Journal of Translational Medicine 20 (1): 569. https://doi.org/10.1186/s12967-022-03764-2.

Baklund, Inger H., Toril Dammen, Torbjørn Å. Moum, Wenche Kristiansen, Diogo S. Duarte, Jesús Castro-Marrero, Ingrid B. Helland, and Elin B. Strand. 2021. “Evaluating Routine Blood Tests According to Clinical Symptoms and Diagnostic Criteria in Individuals with Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” Journal of Clinical Medicine 10 (14): 3105. https://doi.org/10.3390/jcm10143105.

Barjasteh-Askari, Fatemeh, Mohsen Davoudi, Homayoun Amini, Mahsa Ghorbani, Mehdi Yaseri, Masud Yunesian, et al. 2020. “Relationship Between Suicide Mortality and Lithium in Drinking Water: A Systematic Review and Meta-Analysis.” Journal of Affective Disorders 264: 234–41. https://doi.org/10.1016/j.jad.2019.12.027.

Bartley, Christopher M, and David A Ross. 2020. “Schizophrenia: A Homecoming.” Biological Psychiatry 88 (4): e15–17. https://doi.org/10.1016/j.biopsych.2020.06.013.

Belcaro, G. et al. 2022. “Preventive Effects of Pycnogenol on Cardiovascular Risk Factors (Including Endothelial Function) and Microcirculation in Subjects Recovering from Coronavirus Disease 2019 (COVID-19).” Minerva Medica 113 (2): 300–308. https://doi.org/10.23736/S0026-4806.21.07487-3.

Berthier, J, Francky Teddy Endomba, Michel Lecendreux, et al. 2025. “Cerebral Blood Flow in Attention Deficit Hyperactivity Disorder: A Systematic Review.” Neuroscience 567: 67–76. https://doi.org/10.1016/j.neuroscience.2024.11.075.

Bitirgen, G. et al. 2021. “Corneal Confocal Microscopy Identifies Corneal Nerve Fibre Loss and Increased Dendritic Cells in Patients with Long COVID.” British Journal of Ophthalmology. https://doi.org/10.1136/bjophthalmol-2021-319450.

Blaes, Franz, Gerd Wallukat, et al. 2011. “Autoimmunity Against the β2 Adrenergic Receptor and Muscarinic-2 Receptor in Complex Regional Pain Syndrome.” PAIN 152 (12): 2690–700. https://doi.org/10.1016/j.pain.2011.06.012.

Blockmans, Daniel, Philippe Persoons, Boudewijn Van Houdenhove, and Herman Bobbaers. 2006. “Does Methylphenidate Reduce the Symptoms of Chronic Fatigue Syndrome?” American Journal of Medicine 119 (2): 167.e23–30. https://doi.org/10.1016/j.amjmed.2005.07.047.

Blundell, Sally, Kausik K Ray, Matthew Buckland, and Peter D White. 2015. “Chronic Fatigue Syndrome and Circulating Cytokines: A Systematic Review.” Brain, Behavior, and Immunity 50: 186–95. https://doi.org/10.1016/j.bbi.2015.07.004.

Braunschweig, Daniel, Paula Krakowiak, Paul Duncanson, Ryan Boyce, Robin L Hansen, Paul Ashwood, Irva Hertz-Picciotto, Isaac N Pessah, and Judy Van de Water. 2013. “Autism-Specific Maternal Autoantibodies Recognize Critical Proteins in Developing Brain.” Translational Psychiatry 3 (7): e277. https://doi.org/10.1038/tp.2013.50.

Breda, V, Luis Augusto Rohde, Ana M B Menezes, Luciana Anselmi, Arthur Caye, Diego L Rovaris, Eugenia S Vitola, Claiton H D Bau, and Eugenio H Grevet. 2020. “Revisiting ADHD Age-of-Onset in Adults: To What Extent Should We Rely on the Recall of Childhood Symptoms?” Psychological Medicine 50 (5): 857–66. https://doi.org/10.1017/S003329171900076X.

Buiting, Karin, Charles Williams, and Bernhard Horsthemke. 2016. “Angelman Syndrome – Insights into a Rare Neurogenetic Disorder.” Nature Reviews Neurology 12 (10): 584–93. https://doi.org/10.1038/nrneurol.2016.133.

Bush, D., S. H. Abman, and C. Galambos. 2022. “Microvascular Shunts in COVID-19 Pneumonia.” American Journal of Respiratory and Critical Care Medicine 206 (2): 227–28. https://doi.org/10.1164/rccm.202111-2627LE.

Bynke, Anna, Per Julin, Carl-Gerhard Gottfries, Harald Heidecke, Carmen Scheibenbogen, and Jonas Bergquist. 2020. “Autoantibodies to Beta-Adrenergic and Muscarinic Cholinergic Receptors in Myalgic Encephalomyelitis (ME) Patients—a Validation Study in Plasma and Cerebrospinal Fluid from Two Swedish Cohorts.” Brain, Behavior, & Immunity - Health 7: 100107. https://doi.org/10.1016/j.bbih.2020.100107.

Campbell, T. Mark, Bahareh Bahram Ghaedi, Elizabeth Tanjong Ghogomu, and Vivian Welch. 2019. “Shoe Lifts for Leg Length Discrepancy in Adults with Common Painful Musculoskeletal Conditions: A Systematic Review of the Literature.” Archives of Physical Medicine and Rehabilitation 100 (2): 346–57. https://doi.org/10.1016/j.apmr.2017.10.027.

Campen, Claire L. M. van, Peter C. Rowe, and F. C. Visser. 2020. “Two-Day Cardiopulmonary Exercise Testing in Females with a Severe Grade of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Comparison with Patients with Mild and Moderate Disease.” Healthcare (Basel) 8 (3): 192. https://doi.org/10.3390/healthcare8030192.

Cañadas, P., L. Gonzalez-Vides, M. Alberquilla García-Velasco, P. Arriola, N. Guemes-Villahoz, J. A. Gegúndez-Fernández, C. D. Méndez-Hernández, et al. 2023. “Neuroinflammatory Findings of Corneal Confocal Microscopy in Long COVID-19 Patients, 2 Years After Acute SARS-CoV-2 Infection.” Diagnostics 13 (22): 3469. https://doi.org/10.3390/diagnostics13223469.

Casanova, Jean-Laurent, and Laurent Abel. 2022. “From Rare Disorders of Immunity to Common Determinants of Infection: Following the Mechanistic Thread.” Cell 185 (17): 3086–3103. https://doi.org/10.1016/j.cell.2022.07.004.

Cesaroni, Carlo Alberto, Giulia Pisanò, Susanna Rizzi, Agnese Pantani, Daniele Frattini, and Carlo Fusco. 2026. “PANDAS Syndrome: A Narrative Review of the Diagnostic Conundrum in Children with Acute Neuropsychiatric Symptoms.” International Journal of Molecular Sciences 27 (10): 4612. https://doi.org/10.3390/ijms27104612.

Chang, Kiki, Jennifer Frankovich, Michael Cooperstock, Madeleine W Cunningham, M Elizabeth Latimer, Tanya K Murphy, Mark Pasternack, et al. 2015. “Clinical Evaluation of Youth with Pediatric Acute-Onset Neuropsychiatric Syndrome (PANS): Recommendations from the 2013 PANS Consensus Conference.” Journal of Child and Adolescent Psychopharmacology 25 (1): 3–13. https://doi.org/10.1089/cap.2014.0084.

Chang, Xiao, Yichuan Liu, Frank Mentch, Joseph Glessner, Huiqi Qu, Kenny Nguyen, Patrick M A Sleiman, and Hakon Hakonarson. 2020. “Mitochondrial DNA Haplogroups and Risk of Attention Deficit and Hyperactivity Disorder in European Americans.” Translational Psychiatry 10: 370. https://doi.org/10.1038/s41398-020-01064-1.

Cherneva, Dilyana I., Gabriela Kehayova, Simona Dimitrova, and Stela Dragomanova. 2025. “The Central Nervous System Modulatory Activities of N-Acetylcysteine: A Synthesis of Two Decades of Research.” Current Issues in Molecular Biology 47 (9): 710. https://doi.org/10.3390/cimb47090710.

Chiarello, Francesca, Silvia Spitoni, Eric Hollander, Marco Matucci Cerinic, and Stefano Pallanti. 2017. “An Expert Opinion on PANDAS/PANS: Highlights and Controversies.” International Journal of Psychiatry in Clinical Practice 21 (2): 91–98. https://doi.org/10.1080/13651501.2017.1285941.

Cohen, Jack, Aparna Mathew, Katerina D. Dourvetakis, Estefania Sanchez-Guerrero, Rajendra P. Pangeni, Narasimman Gurusamy, Kristina K. Aenlle, et al. 2024. “Recent Research Trends in Neuroinflammatory and Neurodegenerative Disorders.” Cells 13 (6): 511. https://doi.org/10.3390/cells13060511.

Connery, Kathleen, Marie Tippett, Leanna M Delhey, Shannon Rose, John C Slattery, Stephen G Kahler, Juergen Hahn, et al. 2018. “Intravenous Immunoglobulin for the Treatment of Autoimmune Encephalopathy in Children with Autism.” Translational Psychiatry 8 (1): 148. https://doi.org/10.1038/s41398-018-0214-7.

Cox, E R, K F M Marwick, R W Hunter, J Priller, and S M Lawrie. 2020. “Dialysis and Plasmapheresis for Schizophrenia: A Systematic Review.” Psychological Medicine 50 (8): 1233–40. https://doi.org/10.1017/S0033291720001324.

Croen, Lisa A, Daniel Braunschweig, Lori Haapanen, Cathleen K Yoshida, Bruce Fireman, Judith K Grether, Martin Kharrazi, Robin L Hansen, Paul Ashwood, and Judy Van de Water. 2008. “Maternal Mid-Pregnancy Autoantibodies to Fetal Brain Protein: The Early Markers for Autism Study.” Biological Psychiatry 64 (7): 583–88. https://doi.org/10.1016/j.biopsych.2008.05.006.

Cross-Disorder Group of the Psychiatric Genomics Consortium. 2013. “Genetic Relationship Between Five Psychiatric Disorders Estimated from Genome-Wide SNPs.” Nature Genetics 45 (9): 984–94. https://doi.org/10.1038/ng.2711.

Cysique, Lucette A., David Jakabek, Stephanie G. Bracken, Yana Allen-Davidian, Benjamin Heng, Simone Chow, Mona Dehhaghi, et al. 2023. “The Kynurenine Pathway Relates to Post-Acute COVID-19 Objective Cognitive Impairment and PASC.” Annals of Clinical and Translational Neurology 10 (8): 1338–52. https://doi.org/10.1002/acn3.51825.

Dalmau, Josep, Amy J Gleichman, Ethan G Hughes, Jeffrey E Rossi, Xiaoyu Peng, Meizan Lai, Scott K Dessain, Myrna R Rosenfeld, Rita Balice-Gordon, and David R Lynch. 2008. “Anti-NMDA-Receptor Encephalitis: Case Series and Analysis of the Effects of Antibodies.” The Lancet Neurology 7 (12): 1091–98. https://doi.org/10.1016/S1474-4422(08)70224-2.

Davis, Hannah E., Lisa McCorkell, Julia Moore Vogel, and Eric J. Topol. 2023. “Long COVID: Major Findings, Mechanisms and Recommendations.” Nature Reviews Microbiology 21 (3): 133–46. https://doi.org/10.1038/s41579-022-00846-2.

Davis, J. T., J. E. Elliott, J. W. Duke, A. Cristobal, and A. T. Lovering. 2023. “Hyperoxia-Induced Stepwise Reduction in Blood Flow Through Intrapulmonary, but Not Intracardiac, Shunt During Exercise.” American Journal of Physiology: Regulatory, Integrative and Comparative Physiology 324 (5): R689–700. https://doi.org/10.1152/ajpregu.00014.2023.

Davis, Leah, Megan Higgs, Alice Snaith, Tiffany A. Lodge, Jessica Strong, José Antonio Espejo-Oltra, Slawomir Kujawski, et al. 2025. “Dysregulation of Lipid Metabolism, Energy Production, and Oxidative Stress in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome, Gulf War Syndrome and Fibromyalgia.” Frontiers in Neuroscience 19: 1498981. https://doi.org/10.3389/fnins.2025.1498981.

Deckers, IE, HH van der Zee, and EP Prens. 2016. “Severe Fatigue Based on Anaemia in Patients with Hidradenitis Suppurativa: Report of Two Cases and a Review of the Literature.” Journal of the European Academy of Dermatology and Venereology 30 (1): 174–75. https://doi.org/10.1111/jdv.12680.

DecodeME Consortium, Chris P Ponting, et al. 2025. “Initial Findings from the DecodeME Genome-Wide Association Study of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” medRxiv. https://doi.org/10.1101/2025.08.06.25333109v1.

Demontis, D., G. B. Walters, G. Athanasiadis, R. Walters, K. Therrien, T. T. Nielsen, L. Farajzadeh, et al. 2023. “Genome-Wide Analyses of ADHD Identify 27 Risk Loci, Refine the Genetic Architecture and Implicate Several Cognitive Domains.” Nature Genetics 55 (2): 198–208. https://doi.org/10.1038/s41588-022-01285-8.

Dufrusine, Beatrice, Silvia Valentinuzzi, Sandra Bibbò, Verena Damiani, Paola Lanuti, Damiana Pieragostino, Piero Del Boccio, et al. 2022. “Iron Dyshomeostasis in COVID-19: Biomarkers Reveal a Functional Link to 5-Lipoxygenase Activation.” International Journal of Molecular Sciences 24 (1): 15. https://doi.org/10.3390/ijms24010015.

Dunn, Geoffrey A, Krithi Bhanu, Srinivasan Bhanu, John F Cryan, Krithi Bhanu, and Srinivasan Bhanu. 2019. “Neuroinflammation as a Risk Factor for Attention Deficit Hyperactivity Disorder.” Pharmacology Biochemistry and Behavior 182: 22–34. https://doi.org/10.1016/j.pbb.2019.05.005.

Eckey, Macy, Peng Li, Brett Morrison, Jonas Bergquist, Ronald W. Davis, and Wenzhong Xiao. 2025. “Patient-Reported Treatment Outcomes in ME/CFS and Long COVID.” Proceedings of the National Academy of Sciences 122 (28): e2426874122. https://doi.org/10.1073/pnas.2426874122.

Endres, D, V Maier, F Leypoldt, K P Wandinger, B Lennox, T A Pollak, K Nickel, et al. 2020. “Autoantibody-Associated Psychiatric Syndromes: A Systematic Literature Review Resulting in 145 Cases.” Psychological Medicine 52 (6): 1135–46. https://doi.org/10.1017/S0033291720002895.

Endres, Dominique, Thomas A Pollak, Karl Bechter, Dominik Denzel, Karoline Pitsch, Kathrin Nickel, Kimon Runge, et al. 2022. “Immunological Causes of Obsessive-Compulsive Disorder: Is It Time for the Concept of an “Autoimmune OCD” Subtype?” Translational Psychiatry 12 (1): 5. https://doi.org/10.1038/s41398-021-01700-4.

Espejo, T., L. Kiakouama, R. Etter, and S. Condrau. 2025. “Sudden-Onset Platypnea-Orthodeoxia Syndrome in a Patient with Lung Adenocarcinoma and COVID-19 Without a Patent Foramen Ovale.” European Journal of Case Reports in Internal Medicine 12 (11): 005732. https://doi.org/10.12890/2025_005732.

Ezeoke, A, A Mellor, P Buckley, and B Miller. 2013. “A Systematic, Quantitative Review of Blood Autoantibodies in Schizophrenia.” Schizophrenia Research 150 (1): 245–51. https://doi.org/10.1016/j.schres.2013.07.029.

Fanet, Haïl, Lucile Capuron, Nathalie Castanon, Frédéric Calon, and Sylvie Vancassel. 2021. “Tetrahydrobiopterin (BH4) Pathway: From Metabolism to Neuropsychiatry.” Current Neuropharmacology 19 (5): 591–609. https://doi.org/10.2174/1570159X18666200729103529.

Farrow, C. E., R. A. Robles, G. K. Prisk, P. Harbut, A. Malhotra, T. C. Amis, and P. D. Wagner. 2023. “Increased Intrapulmonary Shunt and Alveolar Dead Space Post-COVID-19.” Journal of Applied Physiology 135 (5): 1012–22. https://doi.org/10.1152/japplphysiol.00267.2023.

Feng, Yuqing, and Yun Lu. 2021. “Immunomodulatory Effects of Dopamine in Inflammatory Diseases.” Frontiers in Immunology 12: 663102. https://doi.org/10.3389/fimmu.2021.663102.

Ferrara, Francesco, Eleno Zavaleta, Antonio Vitiello, J. Antonio Villalobos, Andrea Zovi, Rosa Langella, Bianca Serrano, et al. 2023. “The Direct Correlation Between Covid-19 and Attention Deficit/Hyperactivity Disorder (ADHD).” European Neuropsychopharmacology 73: 62–64. https://doi.org/10.1016/j.euroneuro.2023.04.010.

Fesharaki-Zadeh, Arman, Timothy Belliveau, Robert H. Pietrzak, and Amy Arnsten. 2025. “A Novel Multimodal Pharmacologic Approach Using Guanfacine, N-Acetylcysteine, and Donepezil in Severe TBI: A Case Series.” Frontiers in Rehabilitation Sciences 6: 1648002. https://doi.org/10.3389/fresc.2025.1648002.

Fesharaki-Zadeh, Arman, Naomi Lowe, and Amy F. T. Arnsten. 2023. “Clinical Experience with the $\alpha$2A-Adrenoceptor Agonist, Guanfacine, and N-Acetylcysteine for the Treatment of Cognitive Deficits in Long-COVID19.” Neuroimmunology Reports 3: 100154. https://doi.org/10.1016/j.nerep.2022.100154.

Fluder, Natalia, Morgane Humbel, Emeline Recazens, Alexis A Jourdain, Camillo Ribi, et al. 2026. “Mitochondrial Dysfunction Drives Natural Killer Cell Dysfunction in Systemic Lupus Erythematosus.” JCI Insight. https://doi.org/10.1172/jci.insight.195170.

Fluge, Øystein, Ingrid Gurvin Rekeland, Kristin Sørland, et al. 2025. “Plasma Cell Targeting with the Anti-CD38 Antibody Daratumumab in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome—a Clinical Pilot Study.” Frontiers in Medicine 12: 1607353. https://doi.org/10.3389/fmed.2025.1607353.

Freitag, Helma et al. 2020. “Autoantibodies to Beta-Adrenergic and Muscarinic Cholinergic Receptors in ME/CFS.” Scientific Reports 10: 21456. https://doi.org/10.1038/s41598-020-78634-2.

Freitag, Helma, Milena Szklarski, Sibylle Lorenz, Franziska Sotzny, Stephan Bauer, Anke Philippe, Claudia Kedor, et al. 2021. “Autoantibodies to Vasoregulative G-Protein-Coupled Receptors Correlate with Symptom Severity, Autonomic Dysfunction and Disability in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” Journal of Clinical Medicine 10 (16): 3675. https://doi.org/10.3390/jcm10163675.

George, Chelsea T., Biji T. Kurien, and R. Hal Scofield. 2023. “The Potential Utility of Salivary and Tear Proteomics to Discriminate Sjögren’s Disease from Non-Sjögren’s Sicca.” International Journal of Molecular Sciences 24 (24): 17497. https://doi.org/10.3390/ijms242417497.

Germain, Arnaud, Jillian R Jaycox, Christopher J Emig, Aaron M Ring, and Maureen R Hanson. 2025. “An in-Depth Exploration of the Autoantibody Immune Profile in ME/CFS Using Novel Antigen Profiling Techniques.” International Journal of Molecular Sciences 26 (6): 2799. https://doi.org/10.3390/ijms26062799.

Ghirardi, L., I. Brikell, R. Kuja-Halkola, C. M. Freitag, B. Franke, P. Asherson, P. Lichtenstein, and H. Larsson. 2018. “The Familial Co-Aggregation of ASD and ADHD: A Register-Based Cohort Study.” Molecular Psychiatry 23 (2): 257–63. https://doi.org/10.1038/mp.2017.17.

Ghirardi, L., E. Pettersson, M. J. Taylor, C. M. Freitag, B. Franke, P. Asherson, H. Larsson, R. Kuja-Halkola, and P. Lichtenstein. 2019. “Genetic and Environmental Contributions to the Overlap Between ADHD and ASD Trait Dimensions in Young Adults: A Twin Study.” Psychological Medicine 49 (14): 2366–73. https://doi.org/10.1017/S003329171800243X.

Gietl, Marlene, Francesco Burkert, Stefanie Hofer, Johanna M. Gostner, Thomas Sonnweber, Ivan Tancevski, Alex Pizzini, et al. 2024. “Laboratory Parameters Related to Disease Severity and Physical Performance After Reconvalescence of Acute COVID-19 Infection.” Scientific Reports 14 (1): 10388. https://doi.org/10.1038/s41598-024-57448-6.

Gimbach, Silke, Daniel Vogel, Rainer Fried, Stephen V. Faraone, Tobias Banaschewski, Jan Buitelaar, Manfred Döpfner, and Ruth Ammer. 2024. “ADHD Medicine Consumption in Europe After COVID-19: Catch-up or Trend Change?” BMC Psychiatry 24 (1): 112. https://doi.org/10.1186/s12888-024-05505-9.

Giron, L. B. et al. 2022. “Markers of Fungal Translocation Are Elevated During Post-Acute Sequelae of SARS-CoV-2 and Induce NF-kB Signaling.” JCI Insight. https://doi.org/10.1172/jci.insight.160989.

Godlewska, Beata R., Amy L. Sylvester, Uzay E. Emir, Ann L. Sharpley, William T. Clarke, Stephen R. Williams, Ana Jorge Gonçalves, and Philip J. Cowen. 2025. “Brain and Muscle Chemistry in Myalgic Encephalitis/Chronic Fatigue Syndrome (ME/CFS) and Long COVID: A 7T Magnetic Resonance Spectroscopy Study.” Molecular Psychiatry 30: 5215–26. https://doi.org/10.1038/s41380-025-03108-8.

Granerod, Julia, Nicholas W. S. Davies, Parashar P. Ramanuj, Ava Easton, David W. Brown, and Sara L. Thomas. 2017. “Increased Rates of Sequelae Post-Encephalitis in Individuals Attending Primary Care Practices in the United Kingdom: A Population-Based Retrospective Cohort Study.” Journal of Neurology 264 (2): 407–15. https://doi.org/10.1007/s00415-016-8316-8.

Grotmol, Tomasz, Freddie Bray, Harald Holte, Ottar Klepp, Foss Åsmund, and Odd O. Aalen. 2011. “Frailty Modeling of the Bimodal Age-Incidence of Hodgkin Lymphoma in the Nordic Countries.” Cancer Epidemiology, Biomarkers & Prevention 20 (7): 1350–57. https://doi.org/10.1158/1055-9965.EPI-10-1014.

Grove, J., S. Ripke, T. D. Als, M. Mattheisen, R. K. Walters, H. Won, J. Pallesen, et al. 2019. “Identification of Common Genetic Risk Variants for Autism Spectrum Disorder.” Nature Genetics 51 (3): 431–44. https://doi.org/10.1038/s41588-019-0344-z.

Guo, Cheng, Xiaoyu Che, Thomas Briese, Amit Alber, Taylor Alpert, Lia Zhang, Yilin Yu, et al. 2023. “Deficient Butyrate-Producing Capacity in the Gut Microbiome Is Associated with Bacterial Network Disturbances and Fatigue Symptoms in ME/CFS.” Cell Host and Microbe 31 (2): 288–304. https://doi.org/10.1016/j.chom.2023.01.004.

Guo, Y., T. Wei, N. Hu, and X. Zhou. 2021. “Disrupted Homeostasis of Synovial Hyaluronic Acid and Its Associations with Synovial Mast Cell Proteases of Rheumatoid Arthritis Patients and Collagen-Induced Arthritis Rats.” Immunologic Research 69 (6): 584–93. https://doi.org/10.1007/s12026-021-09231-2.

Hadidchi, Ramin, Bonny Patel, Jay Madan, Andrew Liu, Sean Henry, and Tim Q. Duong. 2025. “Elevated Risk of New-Onset Chronic Fatigue Syndrome/Myalgic Encephalomyelitis up to Four Years After SARS-CoV-2 Infection.” Journal of Translational Medicine 23 (1): 815. https://doi.org/10.1186/s12967-025-06625-w.

Hansen, N, A Neyazi, D Lüdecke, A Hasan, J Wiltfang, and B Malchow. 2023. “Repositioning Synthetic Glucocorticoids in Psychiatric Disease Associated with Neural Autoantibodies: A Narrative Review.” Journal of Neural Transmission 130 (8): 1029–38. https://doi.org/10.1007/s00702-022-02578-2.

Hanson, Aimee L., Matthew P. Mulè, Hélène Ruffieux, Federica Mescia, Laura Bergamaschi, Victoria S. Pelly, Lorinda Turner, et al. 2024. “Iron Dysregulation and Inflammatory Stress Erythropoiesis Associates with Long-Term Outcome of COVID-19.” Nature Immunology 25 (3): 471–82. https://doi.org/10.1038/s41590-024-01754-8.

Hartwig, C. et al. 2020. “ME/CFS IgG Fails to Activate the Β2-Adrenergic Receptor.” Brain, Behavior, and Immunity Health.

He, Ji-Ye, Xin-Feng Zheng, and Lei-Sheng Jiang. 2014. “Autonomic Control of Bone Formation: Its Clinical Relevance.” In Handbook of Clinical Neurology, 117:161–71. Elsevier. https://doi.org/10.1016/B978-0-444-53491-0.00014-6.

Heneka, Michael T., Wiesje M. van der Flier, Frank Jessen, Jeroen Hoozemans, Dietmar R. Thal, Delphine Boche, Frederic Brosseron, Charlotte Teunissen, Henrik Zetterberg, et al. 2025. “Neuroinflammation in Alzheimer Disease.” Nature Reviews Immunology 25 (5): 321–52. https://doi.org/10.1038/s41577-024-01104-7.

Heneka, Michael T., Markus P. Kummer, Andrea Stutz, Andrea Delekate, Stephanie Schwartz, Ana Vieira-Saecker, Angelika Griep, et al. 2013. “NLRP3 Is Activated in Alzheimer’s Disease and Contributes to Pathology in APP/PS1 Mice.” Nature 493 (7434): 674–78. https://doi.org/10.1038/nature11729.

Hindley, G., O. Frei, A. A. Shadrin, W. Cheng, K. S. O’Connell, R. Icick, N. Parker, et al. 2022. “Charting the Landscape of Genetic Overlap Between Mental Disorders and Phenotypes Using Large-Scale Genomic Data.” The American Journal of Psychiatry 179 (11): 833–42. https://doi.org/10.1176/appi.ajp.21101051.

Hirsch, A. G., A. E. Justice, A. Poissant, J. N. Williams, J. L. Weissfeld, K. Alderink, T. Gordon, et al. 2025. “A Comparison of Genome-Wide Association Analyses of Persistent Symptoms After Lyme Disease, Fibromyalgia, and Myalgic Encephalomyelitis - Chronic Fatigue Syndrome.” BMC Infectious Diseases 25: 194. https://doi.org/10.1186/s12879-024-10238-x.

Hochecker, Barbara, Katja Matt, Melanie Scherer, Alica Meßmer, Alexander von Ardenne, and Jörg Bergemann. 2025. “Heat Vs. Fatigue: Hyperthermia as a Possible Treatment Option for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS).” International Journal of Molecular Sciences 26 (11): 5339. https://doi.org/10.3390/ijms26115339.

Hoel, F., A. Hoel, I. K. N. Pettersen, I. G. Rekeland, K. Risa, K. Alme, K. Sørland, et al. 2021. “A Map of Metabolic Phenotypes in Patients with Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” JCI Insight 6 (16): e149217. https://doi.org/10.1172/jci.insight.149217.

Imanaka, B. et al. 2026. “Splicing Alterations Associated with Multiple Homozygous TNXB Variants in a Patient with Suspected Classical-Like Ehlers-Danlos Syndrome.” Biochemical Genetics. https://doi.org/10.1007/s10528-026-11416-z.

Ising, Christina, Carmen Venegas, Shuangshuang Zhang, Hannah Scheiblich, Susanne V. Schmidt, Ana Vieira-Saecker, Stephanie Schwartz, et al. 2019. “NLRP3 Inflammasome Activation Drives Tau Pathology.” Nature 575 (7784): 669–73. https://doi.org/10.1038/s41586-019-1769-z.

Janicki, J. S., G. L. Brower, J. D. Gardner, M. F. Forman, J. A. Jr. Stewart, D. B. Murray, and A. L. Chancey. 2006. “Cardiac Mast Cell Regulation of Matrix Metalloproteinase-Related Ventricular Remodeling in Chronic Pressure or Volume Overload.” Cardiovascular Research 69 (3): 657–65. https://doi.org/10.1016/j.cardiores.2005.10.020.

Jason, Leonard A, Jasmine A Dorri, Benjamin H Natelson, Lucinda Bateman, and Suzanne D Vernon. 2025. “Incidence and Prevalence of Post-COVID-19 Myalgic Encephalomyelitis: A RECOVER-Adult Study.” Journal of General Internal Medicine. https://doi.org/10.1007/s11606-024-09290-9.

Javonillo, Dominic I., S. Furman, L. Le, K. Fernandez, J. Mulford, V. Singla, R. Jha, et al. 2026. “Characterization of Effects of a Neurotropic Murine Coronavirus Infection on Alzheimer’s Disease Neuropathology of 5xFAD Mice.” bioRxiv. https://doi.org/10.64898/2026.02.23.707587.

Jin, H, Y An, J Huang, T Luo, and X Wu. 2026. “Pathophysiological Mechanisms of Post-Exertional Malaise: An Integrative Analysis Based on the Metabolism-Immune-Neuro Interaction Model.” Frontiers in Immunology 17: 1774310. https://doi.org/10.3389/fimmu.2026.1774310.

Jubelt, Bruk. 2004. “Post-Polio Syndrome.” Current Treatment Options in Neurology 6 (2): 105–13. https://doi.org/10.1007/s11940-004-0018-3.

Kavyani, Bahar, Seong Beom Ahn, Daniel Missailidis, Sarah J. Annesley, Paul R. Fisher, Richard Schloeffel, Gilles J. Guillemin, David B. Lovejoy, and Benjamin Heng. 2024. “Dysregulation of the Kynurenine Pathway, Cytokine Expression Pattern, and Proteomics Profile Link to Symptomology in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS).” Molecular Neurobiology 61 (7): 3771–87. https://doi.org/10.1007/s12035-023-03784-z.

Kedor, Claudia, Helma Freitag, Lil Meyer-Arndt, Kirsten Wittke, Leif G. Hanitsch, Thomas Zoller, Fridolin Steinbeis, et al. 2022. “A Prospective Observational Study of Post-COVID-19 Chronic Fatigue Syndrome Following the First Pandemic Wave in Germany and Biomarkers Associated with Symptom Severity.” Nature Communications 13 (1): 5104. https://doi.org/10.1038/s41467-022-32507-6.

Kendler, K. S., J. G. M. Rosmalen, H. Ohlsson, J. Sundquist, and K. Sundquist. 2023. “A Distinctive Profile of Family Genetic Risk Scores in a Swedish National Sample of Cases of Fibromyalgia, Irritable Bowel Syndrome, and Chronic Fatigue Syndrome Compared to Migraine and Anxiety.” Psychological Medicine 53 (7): 3027–36. https://doi.org/10.1017/S0033291722000526.

Kerrebijn, Isabel, Gyda Bjornsdottir, Keon Arbabi, Lea Urpa, Hele Haapaniemi, Gudmar Thorleifsson, Lilja Stefansdottir, et al. 2026. “The Genetic Architecture of Fibromyalgia Across 2.5 Million Individuals.” Nature Medicine 32 (8): 3060–70. https://doi.org/10.1038/s41591-026-04492-6.

Khasnavis, Siddharth, Timothy Belliveau, Amy Arnsten, and Arman Fesharaki-Zadeh. 2024. “Combined Use of Guanfacine and N-Acetylcysteine for the Treatment of Cognitive Deficits After Traumatic Brain Injury.” Neurotrauma Reports 5 (1): 226–31. https://doi.org/10.1089/neur.2023.0124.

Kilgour, E. et al. 2026. “A Phase II Trial of the P2X7-NLRP3 Inhibitor AZD9056 in Hidradenitis Suppurativa.” Journal of the American Academy of Dermatology.

Killian, John T., Jane B. Lane, Hye-Seung Lee, Steve A. Skinner, Walter E. Kaufmann, Daniel G. Glaze, Jeffrey L. Neul, and Alan K. Percy. 2017. “Scoliosis in Rett Syndrome: Progression, Comorbidities, and Predictors.” Pediatric Neurology 70: 20–25. https://doi.org/10.1016/j.pediatrneurol.2017.01.032.

Kim, Dong-Yun, Jiwoo Youn, Nayoung Kang, Sun-Il Cho, and In-Hyuk Ha. 2026. “Potential Application of Brain-Gut Axis-Based Treatments in Long COVID and ME/CFS: A Case-Based Systematic Review.” Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-06589-x.

Kirvan, Christine A, Susan E Swedo, Janet S Heuser, and Madeleine W Cunningham. 2003. “Mimicry and Autoantibody-Mediated Neuronal Cell Signaling in Sydenham Chorea.” Nature Medicine 9 (7): 914–20. https://doi.org/10.1038/nm892.

Kirvan, Christine A, Susan E Swedo, David Kurahara, and Madeleine W Cunningham. 2006. “Streptococcal Mimicry and Antibody-Mediated Cell Signaling in the Pathogenesis of Sydenham’s Chorea.” Autoimmunity 39 (1): 21–29. https://doi.org/10.1080/08916930500484757.

Klein, Jon, Jamie Wood, Jillian R. Jaycox, Rahul M. Dhodapkar, Peiwen Lu, Jeff R. Gehlhausen, Alexandra Tabachnikova, et al. 2023. “Distinguishing Features of Long COVID Identified Through Immune Profiling.” Nature 623 (7985): 139–48. https://doi.org/10.1038/s41586-023-06651-y.

Koonce, Rachel M., and Bradley C. Martin. 2024. “The Association Between Suspected Long-COVID and Stimulant Prescribing in the United States.” Psychiatry Research 331: 115686. https://doi.org/10.1016/j.psychres.2023.115686.

Kovacevic, Miro, Paul Grant, and Susan E Swedo. 2015. “Use of Intravenous Immunoglobulin in the Treatment of Twelve Youths with Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal Infections.” Journal of Child and Adolescent Psychopharmacology 25 (1): 65–69. https://doi.org/10.1089/cap.2014.0067.

Krajewski, PK, W Szukała, and JC Szepietowski. 2024. “The NLRP3 Inflammasome Gene Is Overexpressed in Hidradenitis Suppurativa Lesions: A Preliminary Study on the Role of Pyroptosis in Disease Pathogenesis.” Current Issues in Molecular Biology 46 (3): 2544–52. https://doi.org/10.3390/cimb46030161.

Krishnan, Kalpana, Ying Lin, Kim R. M. Prewitt, and David A. Potter. 2022. “Multidisciplinary Approach to Brain Fog and Related Persisting Symptoms Post COVID-19.” Journal of Health Service Psychology 48 (1): 31–38. https://doi.org/10.1007/s42843-022-00056-7.

Kronstein-Wiedemann, Romy, Kristin Tausche, Martin Kolditz, Maria Teichert, Julia Thiel, Dirk Koschel, Torsten Tonn, and Stephan R. Künzel. 2024. “Long-COVID Is Associated with Impaired Red Blood Cell Function.” Hormone and Metabolic Research 56 (4): 318–23. https://doi.org/10.1055/a-2186-8108.

Kuring, J. K., J. L. Mathias, L. Ward, and G. Tachas. 2023. “Inflammatory Markers in Persons with Clinically-Significant Depression, Anxiety or PTSD: A Systematic Review and Meta-Analysis.” Journal of Psychiatric Research 168: 279–92. https://doi.org/10.1016/j.jpsychires.2023.10.009.

———. 2026. “Inflammatory Markers Associated with Dementia: A Systematic Review and Meta-Analysis.” Journal of Psychiatric Research 197: 26–40. https://doi.org/10.1016/j.jpsychires.2026.02.044.

Kurmashev, Rustem. 2026. “UBE3A Dosage Imbalance as a Molecular Framework Linking Angelman Syndrome and Dup15q-Associated Autism Phenotypes.” American Journal of Medical Genetics Part B: Neuropsychiatric Genetics. https://doi.org/10.1002/ajmg.b.70025.

La Bella, Saverio, Giovanna Scorrano, Marta Rinaldi, Armando Di Ludovico, Francesca Mainieri, Marina Attanasi, Alberto Spalice, Francesco Chiarelli, and Luciana Breda. 2023. “Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal Infections (PANDAS): Myth or Reality? The State of the Art on a Controversial Disease.” Microorganisms 11 (10): 2549. https://doi.org/10.3390/microorganisms11102549.

Lai, Meng-Chuan, Caroline Kassee, J. Besney, S. Bonato, L. Hull, W. Mandy, P. Szatmari, and S. H. Ameis. 2019. “Prevalence of Co-Occurring Mental Health Diagnoses in the Autism Population: A Systematic Review and Meta-Analysis.” The Lancet Psychiatry 6 (10): 819–29. https://doi.org/10.1016/S2215-0366(19)30289-5.

Latimer, M Elizabeth, Nathan L’Etoile, Jakob Seidlitz, and Susan E Swedo. 2015. “Therapeutic Plasma Apheresis as a Treatment for 35 Severely Ill Children and Adolescents with Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal Infections.” Journal of Child and Adolescent Psychopharmacology 25 (1): 70–75. https://doi.org/10.1089/cap.2014.0080.

Le, T., H. J. Kim, Y. Chen, Q. Zhang, X. Wang, Y. Liu, M. Zhou, J. Wu, S. Li, and Z. Xu. 2025. “The eIF2alpha-ATF4 Axis Regulates Mitochondrial Dynamics and Bioenergetics During ER Stress.” Cell Metabolism 35 (4): 678–695.e8. https://doi.org/10.1016/j.cmet.2025.01.012.

Lépine, Marylène, Marie-Claude Robert, and Lekha Sleno. 2024. “Discovery and Verification of Sjögren’s Syndrome Protein Biomarkers in Tears by Targeted LC-MRM.” Journal of Proteome Research 23 (6): 2219–29. https://doi.org/10.1021/acs.jproteome.4c00163.

Li, M., X. Duan, J. Zhang, and D. Yang. 2024. “Repercussions of SARS-CoV-2 Infection on Intrapulmonary Shunt in Patients Undergoing One-Lung Ventilation.” Journal of Cardiothoracic Surgery 19 (1): 522. https://doi.org/10.1186/s13019-024-03037-7.

Lim, Eun Ju, Eun Bok Kang, Eun Sol Jang, and Chae Gon Son. 2020. “The Prospects of the Two-Day Cardiopulmonary Exercise Test (CPET) in ME/CFS Patients: A Meta-Analysis.” Journal of Clinical Medicine 9 (12): 4040. https://doi.org/10.3390/jcm9124040.

Lind, T., F. R. Melo, A. M. Gustafson, A. Sundqvist, X. O. Zhao, A. Moustakas, H. Melhus, and G. Pejler. 2022. “Mast Cell Chymase Has a Negative Impact on Human Osteoblasts.” Matrix Biology 112: 1–19. https://doi.org/10.1016/j.matbio.2022.07.005.

Liu, Philip Z., David M. Raizen, Carsten Skarke, Thomas G. Brooks, and Ron C. Anafi. 2025. “Genetic Variants Associated with Chronic Fatigue Syndrome Predict Population-Level Fatigue Severity and Actigraphic Measurements.” Sleep 48 (2): zsae243. https://doi.org/10.1093/sleep/zsae243.

Loebel, Madlen, Patricia Grabowski, Harald Heidecke, Stephan Bauer, Leif G. Hanitsch, Kirsten Wittke, Christian Meisel, et al. 2016. “Antibodies to Beta Adrenergic and Muscarinic Cholinergic Receptors in Patients with Chronic Fatigue Syndrome.” Brain, Behavior, and Immunity 52: 32–39. https://doi.org/10.1016/j.bbi.2015.09.013.

Lu, Yuanqing, Xiang Li, Daoying Geng, Nan Mei, et al. 2020. “Cerebral Micro-Structural Changes in COVID-19 Patients – an MRI-Based 3-Month Follow-up Study.” EClinicalMedicine 25: 100484. https://doi.org/10.1016/j.eclinm.2020.100484.

Luykx, J J, R Visscher, I Winter-van Rossum, P Waters, L D de Witte, W W Fleischhacker, B D Lin, et al. 2024. “Clinical Symptoms and Psychosocial Functioning in Patients with Schizophrenia Spectrum Disorders Testing Seropositive for Anti-NMDAR Antibodies: A Case-Control Comparison with Patients Testing Negative.” The Lancet Psychiatry 11 (10): 828–38. https://doi.org/10.1016/S2215-0366(24)00249-9.

Luzu, Justine, Antoine Labbé, Annabelle Réaux-Le Goazigo, Guillaume Rabut, Hong Liang, Bénédicte Dupas, and Christophe Baudouin. 2022. “In Vivo Confocal Microscopic Study of Corneal Innervation in Sjögren’s Syndrome with or Without Small Fiber Neuropathy.” The Ocular Surface 25: 155–62. https://doi.org/10.1016/j.jtos.2022.07.003.

Lyne, T., L. Camporota, and H. Montgomery. 2024. “Contribution of Intrapulmonary Shunt to the Pathogenesis of Profound Hypoxaemia in Viral Infection: A Mechanistic Discussion with an Illustrative Case.” Journal of the Intensive Care Society 25 (4): 427–31. https://doi.org/10.1177/17511437241267745.

Maccallini, P. 2026. “Biological Insights from Genome-Wide Association Studies and Whole Genome Sequencing of Myalgic Encephalomyelitis/ Chronic Fatigue Syndrome.” Research Square [Preprint], June. https://doi.org/10.21203/rs.3.rs-9702020/v1.

Magarinos, N. J., K. J. Bryant, A. J. Fosang, R. Adachi, R. L. Stevens, and H. P. McNeil. 2013. “Mast Cell-Restricted, Tetramer-Forming Tryptases Induce Aggrecanolysis in Articular Cartilage by Activating Matrix Metalloproteinase-3 and -13 Zymogens.” Journal of Immunology 191 (3): 1404–12. https://doi.org/10.4049/jimmunol.1300856.

Maksoud, Reem, Stanley du Preez, Natalie Eaton-Fitch, Kiran Thapaliya, Leighton Barnden, Hélène Cabanas, Donald Staines, and Sonya Marshall-Gradisnik. 2020. “A Systematic Review of Neurological Impairments in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome Using Neuroimaging Techniques.” PLoS One 15 (4): e0232475. https://doi.org/10.1371/journal.pone.0232475.

Maleki, B. H., and B. Tartibian. 2021. “COVID-19 and Male Reproductive Function: A Prospective, Longitudinal Cohort Study.” Reproduction 161 (3): 319–31. https://doi.org/10.1530/REP-20-0554.

Mantle, D., J. C. Domingo, B. A. Golomb, and J. Castro-Marrero. 2025. “Gulf War Illness, Fibromyalgia, Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Long COVID Overlap in Common Symptoms and Underlying Biological Mechanisms: Implications for Future Therapeutic Strategies.” International Journal of Molecular Sciences 26 (19): 9044. https://doi.org/10.3390/ijms26199044.

Martin, A. 2019. “An Acquired or Heritable Connective Tissue Disorder? A Review of Hypermobile Ehlers Danlos Syndrome.” European Journal of Medical Genetics 62 (7): 103672. https://doi.org/10.1016/j.ejmg.2019.103672.

Masterson, Erin E, Kate Miles, Noelle Schlenk, Cindy Manko, Meiqian Ma, Bahare Farhadian, Kiki Chang, Melissa Silverman, Margo Thienemann, and Jennifer Frankovich. 2025. “Defining Clinical Course of Patients Evaluated for Pediatric Acute-Onset Neuropsychiatric Syndrome: Phenotypic Classification Based on 10 Years of Clinical Data.” Developmental Neuroscience 47 (4): 270–86. https://doi.org/10.1159/000545598.

Matheny, M. V., T. Craig, and T. Al-Shaikhly. 2025. “Systematic Review of Omalizumab for Refractory Clonal and Non-Clonal Mast Cell Activation Syndrome.” Allergy and Asthma Proceedings 46 (1): 11–18. https://doi.org/10.2500/aap.2025.46.240076.

McGrath, BM, D Atherton, P Brady, MF Bru-Daprés, JL Hsiao, S Lobo, L Palmer, et al. 2026. “Exploring the Impact of Fatigue and Sleep Disturbance in Hidradenitis Suppurativa: Perspectives from Patients and Healthcare Professionals.” Acta Dermato-Venereologica 106: 0179. https://doi.org/10.2340/actadv.v106.adv-2025-0179.

McGrath, Simon J., Charlie B. Hillier, Joshua J. Dibble, Trude Schei, Arild Angelsen, and Audrey A. Ryback. 2026. “Incidence Age Is Bimodal for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome, with Higher Severity Burden for Early Onset Disease.” Oxford Open Immunology 7 (1): iqag007. https://doi.org/10.1093/oxfimm/iqag007.

Meltzer, Amory, and Judy Van de Water. 2017. “The Role of the Immune System in Autism Spectrum Disorder.” Neuropsychopharmacology 42 (1): 284–98. https://doi.org/10.1038/npp.2016.158.

Memon, Anjum, Imogen Rogers, S M D D Fitzsimmons, Ben Carter, Rebecca Strawbridge, Diego Hidalgo-Mazzei, and Allan H Young. 2020. “Association Between Naturally Occurring Lithium in Drinking Water and Suicide Rates: Systematic Review and Meta-Analysis of Ecological Studies.” British Journal of Psychiatry 217 (6): 667–78. https://doi.org/10.1192/bjp.2020.128.

Meral Ketenci, B, HB Meral, and Z Meral Obholzer. 2026. “Determinants of Kinesiophobia in Patients with Hidradenitis Suppurativa: The Role of Pain, Fatigue, Physical Activity and Depression.” BMC Psychology. https://doi.org/10.1186/s40359-026-05165-2.

Mertz, Philippe, Aurélien Schlencker, Matthias Schneider, Pierre-Edouard Gavand, Thierry Martin, et al. 2020. “Towards a Practical Management of Fatigue in Systemic Lupus Erythematosus.” Lupus Science & Medicine 7: e000441. https://doi.org/10.1136/lupus-2020-000441.

Merzon, Eugene, Michael Weiss, Brian Krone, Shira Cohen, Galia Ilani, Shlomo Vinker, Avivit Cohen-Golan, et al. 2022. “Clinical and Socio-Demographic Variables Associated with the Diagnosis of Long COVID Syndrome in Youth: A Population-Based Study.” International Journal of Environmental Research and Public Health 19 (10): 5993. https://doi.org/10.3390/ijerph19105993.

Moffitt, Terrie E, Renate Houts, Philip Asherson, Daniel W Belsky, David L Corcoran, et al. 2015. “Is Adult ADHD a Childhood-Onset Neurodevelopmental Disorder? Evidence from a Four-Decade Longitudinal Cohort Study.” American Journal of Psychiatry 172 (10): 967–77. https://doi.org/10.1176/appi.ajp.2015.14101266.

Montoya, Jose G, Tyson H Holmes, Jill N Anderson, Holden T Maecker, Yael Rosenberg-Hasson, Ian J Valencia, Lily Chu, Jarred W Younger, Cristina M Tato, and Mark M Davis. 2017. “Cytokine Signature Associated with Disease Severity in Chronic Fatigue Syndrome Patients.” Proceedings of the National Academy of Sciences 114 (34): E7150–58. https://doi.org/10.1073/pnas.1710519114.

Moran, B, CM Smith, A Zaborowski, M Ryan, J Karman, RW Dunstan, KM Smith, et al. 2023. “Targeting the NLRP3 Inflammasome Reduces Inflammation in Hidradenitis Suppurativa Skin.” British Journal of Dermatology 189 (4): 447–58. https://doi.org/10.1093/bjd/ljad184.

Mueller, Chantal, James C. Lin, Sulaiman Sheriff, Andrew A. Maudsley, and Jarred W. Younger. 2020. “Evidence of Widespread Metabolite Abnormalities in Myalgic encephalomyelitis/chronic fatigue syndrome: Assessment with Whole-Brain Magnetic Resonance Spectroscopy.” Brain Imaging and Behavior 14 (2): 562–72. https://doi.org/10.1007/s11682-018-0029-4.

Mulet, A., J. Signes-Costa, E. Fernández-Fabrellas, J. A. Ros, J. A. Rodrı́guez-Portal, A. L. Andreu, F. Pallardó, and P. González-Cabo. 2026. “Mitochondrial Dysfunction and Telomeric Shortening as Long-Term Complications After COVID-19.” Archivos de Bronconeumologı́a. https://doi.org/10.1016/j.arbres.2026.03.018.

Murrough, James W., Xiangling Mao, Katherine A. Collins, Crystal Kelly, George Andrade, Paul Nestadt, Susan M. Levine, Sanjay J. Mathew, and Dikoma C. Shungu. 2010. “Increased Ventricular Lactate in Chronic Fatigue Syndrome Measured by 1H MRS Imaging at 3.0 t. II: Comparison with Major Depressive Disorder.” NMR in Biomedicine 23 (6): 643–50. https://doi.org/10.1002/nbm.1512.

Nakatomi, Yasuhito, Kei Mizuno, Akemi Ishii, Yasuhiro Wada, Masaaki Tanaka, Shusaku Tazawa, Kayo Onoe, et al. 2014. “Neuroinflammation in Patients with Chronic Fatigue Syndrome/Myalgic Encephalomyelitis: An ¹¹C-(R)-PK11195 PET Study.” Journal of Nuclear Medicine 55 (6): 945–50. https://doi.org/10.2967/jnumed.113.131045.

Natelson, B. H. et al. 2017. “Elevations of Ventricular Lactate Levels Occur in Both Chronic Fatigue Syndrome and Fibromyalgia.” Fatigue: Biomedicine, Health & Behavior 5 (1): 15–20. https://doi.org/10.1080/21641846.2017.1280114.

Nemani, K, J R Jaycox, U Akcan, B M Schuman, S M Yeon, N Harano, K Qin, et al. 2026. “High Prevalence of CNS-Directed Autoantibodies in Patients with Schizophrenia.” bioRxiv [Preprint]. https://doi.org/10.64898/2026.05.04.722731.

Pajuelo, D., M. Dezortova, M. Hajek, M. Ibrahimova, and I. Ibrahim. 2024. “Metabolic Changes Assessed by 1H MR Spectroscopy in the Corpus Callosum of Post-COVID Patients.” MAGMA 37 (5): 937–46. https://doi.org/10.1007/s10334-024-01171-w.

Panov, Josif, Liron Simchi, Yaron Feuermann, and Hanoch Kaphzan. 2020. “Bioinformatics Analyses of the Transcriptome Reveal Ube3a-Dependent Effects on Mitochondrial-Related Pathways.” International Journal of Molecular Sciences 21 (11): 4156. https://doi.org/10.3390/ijms21114156.

Parodis, Ioannis, Julius Lindblom, Leonardo Palazzo, Alexander Tsoi, Nursen Cetrez, et al. 2025. “Poor Health-Related Quality of Life Despite Lupus Low Disease Activity State or Definitions of Remission in Systemic Lupus Erythematosus (SLE) Remission in Patients with SLE: Results from a Clinical Trial Setting.” RMD Open 11: e006061. https://doi.org/10.1136/rmdopen-2025-006061.

Pearlman, D M, and S Najjar. 2014. “Meta-Analysis of the Association Between N-Methyl-D-Aspartate Receptor Antibodies and Schizophrenia, Schizoaffective Disorder, Bipolar Disorder, and Major Depressive Disorder.” Schizophrenia Research 157 (1-3): 249–58. https://doi.org/10.1016/j.schres.2014.05.001.

Peng, Madeleine S., Jialin Jiang, Lucia Polizzi, Tiancheng Shi, Rakshitha Ramkumar, Victor O. Anosike, Giulia Guasoni, et al. 2026. “Dysregulated Adult Hippocampal Neurogenesis in Major Depressive Disorders.” Nature Medicine 32. https://doi.org/10.1038/s41591-026-04571-8.

Peppercorn, Katie, Sayan Sharma, Christina D. Edgar, Peter A. Stockwell, Euan J. Rodger, Aniruddha Chatterjee, and Warren P. Tate. 2025. “Comparing DNA Methylation Landscapes in Peripheral Blood from Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Long COVID Patients.” International Journal of Molecular Sciences 26 (14): 6631. https://doi.org/10.3390/ijms26146631.

Petrasca, A, R Hambly, N Kearney, CM Smith, EK Pender, J Mac Mahon, AM O’Rourke, et al. 2023. “Metformin Has Anti-Inflammatory Effects and Induces Immunometabolic Reprogramming via Multiple Mechanisms in Hidradenitis Suppurativa.” British Journal of Dermatology 189 (6): 730–40. https://doi.org/10.1093/bjd/ljad305.

Petrov, Steliyan, Martina Bozhkova, Mariya Ivanovska, Teodora Kalfova, Dobrina Dudova, Ralitsa Nikolova, Kalina Vaseva, et al. 2025. “Comparable Immune Alterations and Inflammatory Signatures in ME/CFS and Long COVID.” Biomedicines 13 (12): 3001. https://doi.org/10.3390/biomedicines13123001.

Petrov, Steliyan, Martina Bozhkova, Mariya Ivanovska, Teodora Kalfova, Dobrina Dudova, Yana Todorova, Radostina Dimitrova, et al. 2026. “Comprehensive Immunophenotyping of Monocytes and Dendritic Cells Suggests Distinct Pathophysiology in Chronic Fatigue Syndrome and Long COVID.” International Journal of Molecular Sciences 27 (10): 4488. https://doi.org/10.3390/ijms27104488.

Plioplys, A V. 1998. “Intravenous Immunoglobulin Treatment of Children with Autism.” Journal of Child Neurology 13 (2): 79–82. https://doi.org/10.1177/088307389801300207.

Pluvinage, John V, Thomas Ngo, Camille Fouassier, Meghan McDonagh, Brandon B Holmes, Christopher M Bartley, Srikirti Kondapavulur, et al. 2024. “Transcobalamin Receptor Antibodies in Autoimmune Vitamin B12 Central Deficiency.” Science Translational Medicine 16 (753): eadl3758. https://doi.org/10.1126/scitranslmed.adl3758.

Polderman, T. J. C., R. A. Hoekstra, D. Posthuma, and H. Larsson. 2014. “The Co-Occurrence of Autistic and ADHD Dimensions in Adults: An Etiological Study in 17,770 Twins.” Translational Psychiatry 4: e435. https://doi.org/10.1038/tp.2014.84.

Pollak, T A, R McCormack, M Peakman, T R Nicholson, and A S David. 2014. “Prevalence of Anti-N-Methyl-D-Aspartate (NMDA) Receptor Antibodies in Patients with Schizophrenia and Related Psychoses: A Systematic Review and Meta-Analysis.” Psychological Medicine 44 (12): 2475–87. https://doi.org/10.1017/S003329171300295X.

Prens, LM, K Bouwman, LD Troelstra, EP Prens, BZ Alizadeh, and B Horváth. 2022. “New Insights in Hidradenitis Suppurativa from a Population-Based Dutch Cohort: Prevalence, Smoking Behaviour, Socioeconomic Status and Comorbidities.” British Journal of Dermatology 186 (5): 814–22. https://doi.org/10.1111/bjd.20954.

Proessl, Felix, Nathaniel B. Ketelhut, and Thorsten Rudroff. 2018. “No Association of Leg Strength Asymmetry with Walking Ability, Fatigability, and Fatigue in Multiple Sclerosis.” International Journal of Rehabilitation Research 41 (1): 81–86. https://doi.org/10.1097/MRR.0000000000000278.

Qu, Shen, Jinjing Wang, Xiaoqian Guan, Caijun Song, and Xiaoyun Wang. 2024. “Sleep Disturbance in Angelman Syndrome Patients.” Orphanet Journal of Rare Diseases 19 (1): 146. https://doi.org/10.1186/s13023-024-03154-5.

Quadt, Lisa, Jenny Csecs, Robert Bond, et al. 2024. “Childhood Neurodivergent Traits, Inflammation and Chronic Disabling Fatigue in Adolescence: A Longitudinal Case-Control Study.” BMJ Open 14 (7): e084203. https://doi.org/10.1136/bmjopen-2024-084203.

Raijmakers, R, M Roerink, S Keijmel, L Joosten, M Netea, J van der Meer, H Knoop, H Klein, C Bleeker-Rovers, and J Doorduin. 2021. “No Signs of Neuroinflammation in Women with Chronic Fatigue Syndrome or q Fever Fatigue Syndrome Using TSPO Ligand [(11)c]-PK11195.” Neurology Neuroimmunology and Neuroinflammation 9 (1): e1113. https://doi.org/10.1212/NXI.0000000000001113.

Rassoulpour, Asma, Hui-Qiu Wu, Sergi Ferré, and Robert Schwarcz. 2005. “Nanomolar Concentrations of Kynurenic Acid Reduce Extracellular Dopamine Levels in the Striatum.” Journal of Neurochemistry 93 (3): 762–65. https://doi.org/10.1111/j.1471-4159.2005.03134.x.

Rensburg, S. J. van, F. C. Potocnik, T. Kiss, F. Hugo, P. van Zijl, E. Mansvelt, M. E. Carstens, et al. 2001. “Serum Concentrations of Some Metals and Steroids in Patients with Chronic Fatigue Syndrome with Reference to Neurological and Cognitive Abnormalities.” Brain Research Bulletin 55 (2): 319–25. https://doi.org/10.1016/s0361-9230(01)00478-6.

Ring, HC, P Theut Riis, K Zarchi, IM Miller, DM Saunte, and GB Jemec. 2017. “Prodromal Symptoms in Hidradenitis Suppurativa.” Clinical and Experimental Dermatology 42 (3): 261–65. https://doi.org/10.1111/ced.13025.

Rubio, Jose, and Vasileios C Kyttaris. 2023. “Undifferentiated Connective Tissue Disease: Comprehensive Review.” Current Rheumatology Reports 25: 98–106. https://doi.org/10.1007/s11926-023-01099-5.

Rücker, Martin. 2026. “Breaking the Vicious Cycle: How Two German Scientists Seek to Solve ME.” The Sick Times. https://thesicktimes.org/2026/05/14/breaking-the-vicious-cycle-how-two-german-scientists-seek-to-solve-me/.

Saarinen, J., N. Kalkkinen, H. G. Welgus, and P. T. Kovanen. 1994. “Activation of Human Interstitial Procollagenase Through Direct Cleavage of the Leu83-Thr84 Bond by Mast Cell Chymase.” Journal of Biological Chemistry 269 (27): 18134–40.

Sandhu, D., S. Magor-Elliott, N. Petousi, N. Talbot, A. Bennett, and P. A. Robbins. 2026. “Alveolar Deadspace and Intrapulmonary Shunt in Healthy Individuals and in Individuals Who Have Recovered from COVID-19 Infection.” Experimental Physiology 111 (2): 556–67. https://doi.org/10.1113/EP092971.

Sarkar, Chinmoy, Basabi Basu, Debanjan Chakroborty, Partha S. Dasgupta, and Sujit Basu. 2010. “The Immunoregulatory Role of Dopamine: An Update.” Brain, Behavior, and Immunity 24 (4): 525–28. https://doi.org/10.1016/j.bbi.2009.10.015.

Schou, M, S G Sæther, K Borowski, B Teegen, D Kondziella, W Stoecker, A Vaaler, and S K Reitan. 2016. “Prevalence of Serum Anti-Neuronal Autoantibodies in Patients Admitted to Acute Psychiatric Care.” Psychological Medicine 46 (16): 3303–13. https://doi.org/10.1017/S0033291716002038.

Schulte, Tobias L., Bastian Mester, Denise Oberdiek, Nani Osada, Ulf Liljenqvist, Timm J. Filler, Martin Marziniak, and Viola Bullmann. 2010. “Approach-Related Lesions of the Sympathetic Chain in Anterior Correction and Instrumentation of Idiopathic Scoliosis.” European Spine Journal 19 (9): 1558–67. https://doi.org/10.1007/s00586-010-1455-1.

Schwarting, Andreas, Tamara Möckel, Freya Lütgendorf, Konstantinos Triantafyllias, Sophia Grella, et al. 2019. “Fatigue in SLE: Diagnostic and Pathogenic Impact of Anti-N-Methyl-D-Aspartate Receptor (NMDAR) Autoantibodies.” Annals of the Rheumatic Diseases 78: 1226–33. https://doi.org/10.1136/annrheumdis-2019-215098.

Sen, S. et al. 2022. “Retinal Manifestations in Patients with SARS-CoV-2 Infection and Pathogenetic Implications: A Systematic Review.” International Ophthalmology 42 (1): 323–36. https://doi.org/10.1007/s10792-021-01949-0.

Shankar, Vishnu, Julie Wilhelmy, Ellis J. Curtis, Basil Michael, Layla Cervantes, Vamsee A. Mallajosyula, Ronald W. Davis, et al. 2025. “Oxidative Stress Is a Shared Characteristic of ME/CFS and Long COVID.” Proceedings of the National Academy of Sciences U.S.A. 122 (28): e2426564122. https://doi.org/10.1073/pnas.2426564122.

Shkalim Zemer, Vered, Iris Manor, Abraham Weizman, Herman A. Cohen, Moshe Hoshen, Nehama M. Caspi, Shlomo Cohen, Stephen V. Faraone, and Nily Shahar. 2024. “The Influence of COVID-19 on Attention-Deficit/Hyperactivity Disorder Diagnosis and Treatment Rates Across Age, Gender, and Socioeconomic Status: A 20-Year National Cohort Study.” Psychiatry Research 339: 116077. https://doi.org/10.1016/j.psychres.2024.116077.

Sibley, Margaret H, Luis Augusto Rohde, James M Swanson, Lily T Hechtman, Brooke S G Molina, John T Mitchell, L Eugene Arnold, et al. 2018. “Late-Onset ADHD Reconsidered with Comprehensive Repeated Assessments Between Ages 10 and 25.” American Journal of Psychiatry 175 (2): 140–49. https://doi.org/10.1176/appi.ajp.2017.17030298.

Sigra, Sofia, Eva Hesselmark, and Susanne Bejerot. 2018. “Treatment of PANDAS and PANS: A Systematic Review.” Neuroscience & Biobehavioral Reviews 86: 51–65. https://doi.org/10.1016/j.neubiorev.2018.01.001.

Simchi, Liron, Pratiksha K. Gupta, Yaron Feuermann, and Hanoch Kaphzan. 2023. “Elevated ROS Levels During the Early Development of Angelman Syndrome Alter the Apoptotic Capacity of the Developing Neural Precursor Cells.” Molecular Psychiatry 28 (6): 2382–97. https://doi.org/10.1038/s41380-023-02038-7.

Sonnweber, Thomas, Philipp Grubwieser, Sabina Sahanic, Anna K. Böhm, Alex Pizzini, Anna Luger, Christoph Schwabl, et al. 2022. “The Impact of Iron Dyshomeostasis and Anaemia on Long-Term Pulmonary Recovery and Persisting Symptom Burden After COVID-19: A Prospective Observational Cohort Study.” Metabolites 12 (6): 546. https://doi.org/10.3390/metabo12060546.

Sousa, Ricardo Augusto Leoni, Amr Yehia, and Osama A. Abulseoud. 2023. “Attenuation of Ferroptosis as a Potential Therapeutic Target for Neuropsychiatric Manifestations of Post-COVID Syndrome.” Frontiers in Neuroscience 17: 1237153. https://doi.org/10.3389/fnins.2023.1237153.

Spanoghe, Martin, Tomaso Antonacci, Nicole Schneider, and Thomas H. J. Molmans. 2026. “Viewpoint: Linking Long Covid and AD(H)D Through Neuroimmune Dysfunction: A Translational Framework Proposal for Precision Medicine.” Brain, Behavior, and Immunity 131: 106181. https://doi.org/10.1016/j.bbi.2025.106181.

Spinelli, Francesca Romana, Riccardo Berti, Gabriele Farina, Fulvia Ceccarelli, and Fabrizio Conti. 2023. “Exercise-Induced Modulation of Interferon-Signature: A Therapeutic Route Toward Management of Systemic Lupus Erythematosus.” Autoimmunity Reviews 22: 103412. https://doi.org/10.1016/j.autrev.2023.103412.

Stavileci, B., E. Özdemir, B. Özdemir, E. Ereren, and M. Cengiz. 2022. “De-Novo Development of Fragmented QRS During a Six-Month Follow-up Period in Patients with COVID-19 Disease and Its Cardiac Effects.” Journal of Electrocardiology 72: 44–48. https://doi.org/10.1016/j.jelectrocard.2022.03.006.

Steen, O. D., H. Ohlsson, S. L. van Ockenburg, K. S. Kendler, J. Sundquist, and K. Sundquist. 2026. “Shared Genetic Risk Between Functional Somatic Syndromes, Internalizing Disorders, and Immune-Mediated Diseases: A Twin-Sibling Study.” Brain, Behavior, and Immunity 117: 1–10. https://doi.org/10.1016/j.bbi.2026.106837.

Stein, Elisa, Cornelia Heindrich, Kirsten Wittke, Claudia Kedor, Rebekka Rust, Helma Freitag, Franziska Sotzny, et al. 2025. “Efficacy of Repeated Immunoadsorption in Patients with Post-COVID Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Elevated Beta2-Adrenergic Receptor Autoantibodies: A Prospective Cohort Study.” The Lancet Regional Health - Europe 48: 101161. https://doi.org/10.1016/j.lanepe.2024.101161.

Strawbridge, Rebecca, Maria Laura Sartor, Fiona Scott, and Anthony J Cleare. 2019. “Inflammatory Proteins Are Altered in Chronic Fatigue Syndrome–a Systematic Review and Meta-Analysis.” Neuroscience and Biobehavioral Reviews 107: 69–83. https://doi.org/10.1016/j.neubiorev.2019.08.011.

Su, Hui, Wei Fan, Pinar E. Coskun, Jouni Vesa, Jeffrey A. Gold, Yong-hui Jiang, Prasanth Potluri, Douglas C. Wallace, Jane Gibson, and Virginia E. Kimonis. 2011. “Mitochondrial Dysfunction in CA1 Hippocampal Neurons of the UBE3A Deficient Mouse Model for Angelman Syndrome.” Neuroscience Letters 487 (2): 129–33. https://doi.org/10.1016/j.neulet.2009.06.079.

Swedo, Susan E, Henrietta L Leonard, Marjorie Garvey, Barbara Mittleman, Albert J Allen, Susan Perlmutter, Lorraine Lougee, Sara Dow, Jason Zamkoff, and Billie K Dubbert. 1998. “Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal Infections: Clinical Description of the First 50 Cases.” The American Journal of Psychiatry 155 (2): 264–71. https://doi.org/10.1176/ajp.155.2.264.

Świątczak, Michał, Karolina Sikorska, Mateusz Ciężarek, Andrzej Białczyk, Anita Poniatowska, Justyna Gołębiewska, Ewa Bryl, et al. 2022. “Chronic Fatigue Syndrome in Patients with Deteriorated Iron Metabolism.” Diagnostics 12 (9): 2057. https://doi.org/10.3390/diagnostics12092057.

Tanaka, Masaru, Ákos Szabó, Eleonóra Spekker, Helga Polyák, Fanni Tóth, and László Vécsei. 2022. “Mitochondrial Impairment: A Common Motif in Neuropsychiatric Presentation? The Link to the Tryptophan–Kynurenine Metabolic System.” Cells 11 (16): 2607. https://doi.org/10.3390/cells11162607.

Tatai, Ottó, Szilárd Nagy, Trai Huynh Thanh Nguyen, Beáta Lajszné Tóth, Péter Antal-Szalmás, Ivetta Mányiné Siket, Tamás Bence Pintér, et al. 2026. “Tissue-Specific Autoantibody Signatures Reveal Immune Alterations Undetected by Routine Serology in Long COVID.” GeroScience, May. https://doi.org/10.1007/s11357-026-02286-9.

Taylor, M. J., J. Martin, Y. Lu, I. Brikell, S. Lundström, H. Larsson, and P. Lichtenstein. 2019. “Association of Genetic Risk Factors for Psychiatric Disorders and Traits in a Swedish National Twin Sample.” JAMA Psychiatry 76 (3): 280–89. https://doi.org/10.1001/jamapsychiatry.2018.3652.

Tebartz van Elst, L, K Runge, P T Meyer, H Urbach, N Venhoff, and H Prüss. 2025. “The Neuropsychiatric Checklist for Autoimmune Psychosis: A Narrative Review.” Biological Psychiatry 98 (9): 654–69. https://doi.org/10.1016/j.biopsych.2025.02.889.

Thapa, R., M. R. Kjaer, E. Mignot, J. Zou, et al. 2026. “A Multimodal Sleep Foundation Model for Disease Prediction.” Nature Medicine. https://doi.org/10.1038/s41591-025-04133-4.

Thapaliya, Kiran, Sonya Marshall-Gradisnik, Maira Inderyas, and Leighton Barnden. 2025. “Altered Brain Tissue Microstructure and Neurochemical Profiles in Long COVID and Recovered COVID-19 Individuals: A Multimodal MRI Study.” Brain, Behavior, & Immunity – Health 50: 101142. https://doi.org/10.1016/j.bbih.2025.101142.

Thapaliya, Kiran, Sonya Marshall-Gradisnik, Donald Staines, and Leighton Barnden. 2021. “Diffusion Tensor Imaging Reveals Neuronal Microstructural Changes in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” European Journal of Neuroscience 54 (6): 6214–28. https://doi.org/10.1111/ejn.15413.

Tsilioni, Irene, Benjamin Natelson, and Theoharis C. Theoharides. 2022. “Exosome-Associated Mitochondrial DNA from Patients with ME/CFS Stimulates Human Microglia to Release IL-1$\beta$.” European Journal of Neuroscience 56 (10): 5784–94. https://doi.org/10.1111/ejn.15828.

Tsuchida, T. et al. 2022. “Relationship Between Changes in Symptoms and Antibody Titers After a Single Vaccination in Patients with Long COVID.” Journal of Medical Virology 94 (7): 3416–20. https://doi.org/10.1002/jmv.27725.

Vazquez, G., J. P. Abonia, J. Hallgren, J. M. Fahrenholz, A. M. Stevens, C. Akin, J. T. Lyden, et al. 2022. “Hereditary Alpha Tryptasemia Does Not Concentrate in Hypermobility Disorders.” Journal of Allergy and Clinical Immunology 149 (2): 416–426.e7. https://doi.org/10.1016/j.jaci.2021.09.036.

Vergés, Cristina, Ana Giménez-Capitán, Victòria Ribas, Elena Martínez-Pérez, M. José González, and Francesc March de Ribot. 2025. “Clinical and Genetic Profiling of Fibromyalgia-Associated Dry Eye: A Multifactorial Approach.” The Ocular Surface 38: 377–87. https://doi.org/10.1016/j.jtos.2025.10.012.

Verma, Poonam, Alpana Singh, Dominic Ngima Nthenge-Ngumbau, Usha Rajamma, Swagata Sinha, Kanchan Mukhopadhyay, and Kochupurackal P Mohanakumar. 2016. “Attention Deficit-Hyperactivity Disorder Suffers from Mitochondrial Dysfunction.” BBA Clinical 6: 153–58. https://doi.org/10.1016/j.bbacli.2016.10.003.

Vreeland, Allison, Denise Calaprice, Noga Or-Geva, Richard E Frye, Dritan Agalliu, Herbert M Lachman, Christopher Pittenger, et al. 2023. “Postinfectious Inflammation, Autoimmunity, and Obsessive-Compulsive Disorder: Sydenham Chorea, Pediatric Autoimmune Neuropsychiatric Disorder Associated with Streptococcal Infection, and Pediatric Acute-Onset Neuropsychiatric Disorder.” Developmental Neuroscience 45 (6): 361–74. https://doi.org/10.1159/000534261.

Walitt, Brian, Komudi Singh, Samuel R LaMunion, Mark Hallett, Sandra Jacobson, Kong Chen, Yoshihisa Enose-Akahata, et al. 2024. “Deep Phenotyping of Post-Infectious Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” Nature Communications 15 (1): 907. https://doi.org/10.1038/s41467-024-45107-3.

Wallukat, Gerd et al. 2021. “Functional Autoantibodies Against G-protein Coupled Receptors in Patients with Persistent Long-COVID-19 Symptoms.” Journal of Translational Autoimmunity 4: 100100. https://doi.org/10.1016/j.jtauto.2021.100100.

Wang, Ping-yuan, Jin Ma, Young-Chae Kim, et al. 2023. “WASF3 Disrupts Mitochondrial Respiration and May Mediate Exercise Intolerance in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” Proceedings of the National Academy of Sciences 120 (34): e2302738120. https://doi.org/10.1073/pnas.2302738120.

Wang, Yan, Xian Jiang, Jing Li, and Xue Li. 2025. “Predictive Value of Dry Eye Disease Signs and Corneal in Vivo Confocal Microscopy on Serological Activity in Primary Sjögren Disease.” Medicine 104 (14): e42054. https://doi.org/10.1097/MD.0000000000042054.

Watanabe, Makoto, Takuya Goto, Riko Miyoshi, Shohei Nakakubo, Yumiko Hiramatsu, Ryosuke Nakajima, Akihiko Kondo, Kazuki Higashiyama, and Kazuhiro Yamakawa. 2026. “Region-Dependent Differences in Tonic Inhibition Underlie Epileptic Features in Angelman Syndrome Model Mice.” Epilepsia. https://doi.org/10.1002/epi.70307.

Watton, Paul, and Bhupesh K. Prusty. 2026. “Reframing ME/CFS: Toward a Unified Mechanistic Model of Chronic Post-Infectious Diseases.” Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-08319-3.

Weissman-Tsukamoto, Rachel, Kaitlin R Carroll, and Betty Diamond. 2025. “Diffuse Neuropsychiatric Lupus: Clinical Evidence, Immune-Mediated Mechanisms, and Therapeutic Insights.” Seminars in Immunology 78: 101981. https://doi.org/10.1016/j.smim.2025.101981.

Whiteley, Paul, Ben Marlow, Ritika R Kapoor, Natasa Blagojevic-Stokic, and Regina Sala. 2021. “Autoimmune Encephalitis and Autism Spectrum Disorder.” Frontiers in Psychiatry 12: 775017. https://doi.org/10.3389/fpsyt.2021.775017.

Williams, Grant E, Sharon Hausman-Cohen, Maryelaine Sotos, Emily Gutierrez, Carol Bilich, Francis W Mueller, and Shaun Jagshi. 2025. “The Role of GCH1 Deficiency and Tetrahydrobiopterin in Mental Health.” International Journal of Molecular Sciences 26 (16): 8030. https://doi.org/10.3390/ijms26168030.

Williams, Kyle A, Susan E Swedo, Cristan A Farmer, Heidi Grantz, Paul J Grant, Precilla D’Souza, Rebecca Hommer, Liliya Katsovich, Robert A King, and James F Leckman. 2016. “Randomized, Controlled Trial of Intravenous Immunoglobulin for Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal Infections.” Journal of the American Academy of Child and Adolescent Psychiatry 55 (10): 860–867.e2. https://doi.org/10.1016/j.jaac.2016.06.017.

Wilson, F. C., D. J. Zangerle, L. E. Rozen, J. J. Fliess, A. A. Darakjian, K. A. Sacco, C. Hamilton, et al. 2026. “Mast Cell Score Associates with Wide-Spread Mast Cell Symptoms and Comorbidities in Patients with hEDS and HSD.” medRxiv [Preprint]. https://doi.org/10.64898/2026.05.31.26354552.

Wirth, Klaus J., and Carmen Scheibenbogen. 2025. “Imbalance of Excitatory and Inhibitory Neurotransmitter Pathways in ME/CFS and Long COVID.” Preprints.org (preprint, not peer-reviewed). https://www.preprints.org/frontend/manuscript/025f093892ed0dc2aef00d95d0f2fb85/download_pub.

———. 2026. “Imbalance of Excitatory and Inhibitory Neurotransmitter Systems in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.” International Journal of Molecular Sciences 27 (9): 4041. https://doi.org/10.3390/ijms27094041.

Wu, Kai, Zhiliang Wu, Siyuan Feng, Tianci Zhou, Yuping Ning, Kuangshi Li, and Hongxiao Jia. 2026. “Microstructural White Matter Impairments in Chronic Fatigue Syndrome: Evidence of Segmental Injury in the Cingulum Bundle.” Brain Research Bulletin 234: 111671. https://doi.org/10.1016/j.brainresbull.2025.111671.

Wu, Kevin Y., Omar Serhan, Antoine Faucher, and Simon D. Tran. 2024. “Advances in Sjögren’s Syndrome Dry Eye Diagnostics: Biomarkers and Biomolecules Beyond Clinical Symptoms.” Biomolecules 14 (1): 80. https://doi.org/10.3390/biom14010080.

Xu, Wen-Tao, Xian-Bin An, Mei-Jun Chen, Jie Ma, Xue-Qin Wang, Jia-Nan Yang, Qiang Wang, et al. 2025. “A Gene Cluster of Mitochondrial Complexes Contributes to the Cognitive Decline of COVID-19 Infection.” Molecular Neurobiology 62 (6): 6869–83. https://doi.org/10.1007/s12035-024-04471-3.

Yamamoto, Yukichika, Yuki Otsuka, Kazuki Tokumasu, Naruhiko Sunada, Yasuhiro Nakano, Hiroyuki Honda, Yasue Sakurada, Toru Hasegawa, Hideharu Hagiya, and Fumio Otsuka. 2023. “Utility of Serum Ferritin for Predicting Myalgic Encephalomyelitis/Chronic Fatigue Syndrome in Patients with Long COVID.” Journal of Clinical Medicine 12 (14): 4737. https://doi.org/10.3390/jcm12144737.

Yokokura, Masamichi, Kenji Takedera, Ken Kazumata, et al. 2021. “In Vivo Imaging of Dopamine D1 Receptor and Activated Microglia in Attention-Deficit/Hyperactivity Disorder: A Positron Emission Tomography Study.” Molecular Psychiatry 26 (9): 4958–67. https://doi.org/10.1038/s41380-020-0784-1.

Zeineh, Michael M, Jangho Kang, Scott W Atlas, Maneesh M Raman, Allan L Reiss, John L Norris, Ines Valencia, and Jose G Montoya. 2015. “Right Arcuate Fasciculus Abnormality in Chronic Fatigue Syndrome.” Radiology 274 (2): 517–26. https://doi.org/10.1148/radiol.14141079.

Zhang, Z, X Ma, Z Xia, J Chen, Y Liu, Y Chen, J Zhu, et al. 2017. “NLRP3 Inflammasome Activation Mediates Fatigue-Like Behaviors in Mice via Neuroinflammation.” Neuroscience 358: 115–23. https://doi.org/10.1016/j.neuroscience.2017.06.048.

Zhang, ZT, XM Du, XJ Ma, Y Zong, JK Chen, CL Yu, YG Liu, YC Chen, LJ Zhao, and GC Lu. 2016. “Activation of the NLRP3 Inflammasome in Lipopolysaccharide-Induced Mouse Fatigue and Its Relevance to Chronic Fatigue Syndrome.” Journal of Neuroinflammation 13 (1): 71. https://doi.org/10.1186/s12974-016-0539-1.

Zheng, Daisi, Xiaolong Li, Peicheng Wang, Qingmiao Zhu, Zhiyan Huang, et al. 2024. “Exploring the Shared Mechanism of Fatigue Between Systemic Lupus Erythematosus and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Monocytic Dysregulation and Drug Repurposing.” Frontiers in Immunology 15: 1440922. https://doi.org/10.3389/fimmu.2024.1440922.

Zhu, J., D. Sun, G. Zeng, K. Wu, X. Yang, X. Zhu, D. Aierken, et al. 2026. “Histaminylation Alters Collagen Matrix Mechanics and Attenuates Cardiac Fibrosis Post-Myocardial Infarction via Mechanotransduction Signaling Axis.” Signal Transduction and Targeted Therapy 11 (1): 229. https://doi.org/10.1038/s41392-026-02721-5.