Summary: Integrated Cardiovascular Model
Cardiovascular dysfunction in ME/CFS involves multiple interacting abnormalities ((Walitt et al. 2024)):
Reduced blood volume: Hypovolemia compromises cardiac preload and limits cardiac output reserve
Autonomic dysfunction: Parasympathetic withdrawal reduces HRV and impairs baroreflex function; chronotropic incompetence limits exercise heart rate response
Endothelial dysfunction: Impaired vasodilation reduces tissue perfusion
Cardiac limitation: Preload failure and chronotropic incompetence reduce maximal cardiac output
Microcirculatory impairment: Abnormal capillary perfusion and oxygen extraction limit peripheral oxygen delivery
Exercise intolerance: The cumulative effect is reduced VO2peak and early anaerobic threshold, objectively confirmed by the NIH study
Post-exertional deterioration: Unique to ME/CFS, the failure to recover exercise capacity on day 2 CPET reflects pathological response to exertion
Orthostatic intolerance: Blood pressure dysregulation (POTS, NMH, OH) produces symptoms with upright posture
This cardiovascular dysfunction explains much of the disability in ME/CFS: patients cannot sustain physical activity because their cardiovascular system cannot deliver adequate oxygen to meet metabolic demands. The objective documentation of reduced VO2peak and chronotropic incompetence in the NIH deep phenotyping study provides biological validation of patients’ reported exercise intolerance. These cardiovascular abnormalities integrate with metabolic dysfunction (Chapter Energy Metabolism and Mitochondrial Function), autonomic dysfunction (Chapter Neurological and Neurocognitive Dysfunction), and immune dysregulation (Chapter Immune System Dysfunction) to produce the multi-system pathophysiology synthesized in Chapter Integrative Models and Multi-System Pathophysiology.
Treatment approaches targeting cardiovascular dysfunction include volume expansion (fludrocortisone, increased fluid and salt intake), direct-acting autonomic agents (midodrine as alpha-agonist for vasoconstriction), and careful activity management to avoid exceeding the reduced aerobic threshold. The efficacy of midodrine (acting directly on peripheral alpha-receptors) is consistent with any model in which peripheral vasoconstriction aids orthostatic tolerance — including simple hypovolemia, peripheral autonomic failure, or central coordination failure. It is not specific evidence for the selective energy dysfunction hypothesis, as multiple mechanistic frameworks predict the same clinical response.
The recognition that cardiovascular abnormalities are objective and measurable helps counter misconceptions that ME/CFS exercise intolerance reflects psychological factors or simple deconditioning. Charlton et al. (2026) formalized this argument in the British Journal of Sports Medicine, concluding that post-exertional malaise cannot be explained by cardiac deconditioning and calling for a fundamental rethinking of exercise pathophysiology in post-viral illness (Braeden T. Charlton et al. 2026a). Their primary research paper (Braeden T. Charlton et al. 2026b) published concurrently in Nature Communications provides the empirical backbone: a direct comparison of 60-day strict bed rest against Long COVID and ME/CFS muscle biopsies showing that the patient phenotype (glycolytic shift, Type I atrophy, OXPHOS–V̇O₂ₘₐₓ uncoupling) is qualitatively different from deconditioning — not just a more severe version of it.
Measurement note — the resting-normal / provocation-abnormal pattern: The cardiovascular abnormalities documented above illustrate a broader phenomenon formalized in Section The Subjective-Measurable Discrepancy: A Diagnostic Pattern Across ME/CFS Measurement Domains: many ME/CFS pathologies are provocation-dependent, detectable only under orthostatic or exercise stress but absent at rest. van Campen et al. demonstrated that cerebral blood flow decline during tilt testing occurs regardless of resting hemodynamics or deconditioning severity Campen, Rowe, and Visser (2021). Novak et al. found zero correlation between subjective autonomic symptom questionnaires (COMPASS-31, SAS) and objective autonomic testing (QASAT, CASS) across 2,627 patients (Novak et al. 2024). Friedberg et al. showed that lower HRV — an objective autonomic measure — was associated with patient-reported nonimprovement at 6-month follow-up, while self-reported activity patterns (push-crash, limiting, pacing) did NOT discriminate improvers from non-improvers (Friedberg et al. 2022). The implication for cardiovascular assessment: normal resting ECG, blood pressure, and HR do not rule out orthostatic pathology — and a patient whose resting vitals are normal but whose orthostatic symptoms are severe is consistent with the expected clinical pattern of provocation-dependent autonomic dysfunction. Provocation testing (tilt table or active stand test) can reveal cardiovascular abnormalities that resting measurements systematically miss and should be considered when results would change management. Safety note: 2-day CPET is the gold-standard provocation for PEM documentation but carries non-trivial PEM risk and should be severity-gated and patient-discussed; tilt table and active stand test are lower-risk alternatives for orthostatic assessment. Bedbound very-severe patients may not tolerate any form of provocation testing, and the resting-normal/provocation-abnormal framework may not apply in this subgroup where resting abnormalities may be profound.
Physical activity requires simultaneous CNS coordination of motor output (muscle recruitment, movement planning, proprioceptive feedback) and autonomic regulation (heart rate adjustment, blood pressure maintenance, thermoregulation, respiratory drive). In healthy individuals, these coordination tasks operate efficiently within the brain’s energy budget. We hypothesize that in ME/CFS, where CNS energy is the primary bottleneck (Section Selective Energy Dysfunction Hypothesis), motor and autonomic coordination compete for insufficient resources, causing both systems to fail under demand.
The Dual-Coordination Problem. During exercise, the CNS must:
- Motor coordination: Generate movement commands, integrate proprioceptive feedback, maintain balance, adjust force output—all requiring continuous cortical, cerebellar, and spinal processing.
- Autonomic coordination: Increase heart rate, redistribute blood flow, maintain blood pressure during postural changes, regulate respiration, initiate sweating—all requiring brainstem and hypothalamic processing.
- Integration: Coordinate motor and autonomic outputs so that cardiovascular supply matches muscular demand in real time.
In ME/CFS, if total CNS energy available for coordination is reduced, attempting both tasks simultaneously will exceed the available budget sooner than either task alone. This explains the central governor theory observation ((Noakes, St Clair Gibson, and Lambert 2005), (St Clair Gibson and Noakes 2004)): the brain limits motor output to protect itself from energy depletion.
ME/CFS-Specific Predictions. This hypothesis explains several puzzling CPET findings:
Reduced VO2peak beyond deconditioning. The NIH deep phenotyping study documented VO2peak reductions exceeding what deconditioning alone would predict (Walitt et al. 2024). If the brain limits motor output to preserve autonomic coordination capacity, peak exercise performance reflects the CNS energy budget, not peripheral muscle capacity.
Chronotropic incompetence. The failure to achieve age-predicted maximal heart rate (Walitt et al. 2024) may reflect CNS prioritization: under energy constraint, the brain may reduce autonomic drive to the heart in order to preserve motor coordination, or vice versa. The specific pattern of failure depends on which system the CNS prioritizes in a given individual.
Day-2 CPET deterioration. The pathognomonic worsening on repeat CPET the following day reflects CNS energy depletion that has not recovered. The first test depletes CNS coordination reserves; insufficient recovery time means the second test starts from a lower baseline, producing objectively worse performance.
PEM as coordination exhaustion. Post-exertional malaise may represent the downstream consequence of depleting CNS coordination reserves. Once exhausted, the brain cannot adequately coordinate autonomic function (causing orthostatic symptoms, heart rate irregularity) or motor output (causing weakness, poor coordination), producing the multi-system symptom exacerbation characteristic of PEM.
Testable Predictions.
- Cognitive-physical interference: ME/CFS patients should show greater cognitive impairment during physical activity (dual-task paradigm) than healthy controls, reflecting competition for shared CNS resources.
- Autonomic-motor trade-off: During exercise, ME/CFS patients should show an inverse relationship between motor performance and autonomic function quality (e.g., better muscle output correlates with worse HRV, and vice versa).
- Separate-task preservation: Motor tasks without significant autonomic demand (e.g., seated fine motor tasks) and autonomic challenges without motor demand (e.g., passive tilt testing) should each show less impairment than combined motor-autonomic challenges (e.g., exercise).
- Pharmacological bypass: Agents that directly support autonomic function (midodrine, pyridostigmine) should improve exercise tolerance by offloading CNS autonomic coordination, freeing energy for motor output.
Treatment Implications.
- Pre-treatment with autonomic agents: Taking autonomic-supporting medications before planned physical activity could extend exercise tolerance by reducing CNS autonomic coordination demands.
- Activity design: Activities that minimize simultaneous motor-autonomic demand (recumbent exercise, swimming) should be better tolerated than upright weight-bearing exercise.
- Pacing rationale: The coordination overload model provides a mechanistic rationale for pacing: staying below the threshold where motor and autonomic demands simultaneously exceed CNS capacity prevents the cascade of coordination failure that produces PEM.
Limitations. The hypothesis assumes CNS energy is the primary constraint, which remains debated. Peripheral factors (mitochondrial dysfunction, reduced blood volume, deconditioning) independently contribute to exercise intolerance. The dual-task prediction requires careful experimental design to distinguish CNS resource competition from general fatigue. Central governor theory itself remains controversial in exercise physiology.
Certainty: 0.55 (CPET findings well-documented; CNS coordination mechanism plausible; dual-task predictions testable but not yet tested in ME/CFS)
Small fiber neuropathy affects approximately 30% of ME/CFS patients (Azcue et al. 2023) (Azcue et al. 2025), creating a bidirectional communication burden between the peripheral nervous system and central nervous system that may amplify energy constraints.
Afferent Signal Degradation. SFN reduces the quality of autonomic afferent signals reaching the CNS—temperature sensing, baroreceptor feedback, visceral sensation. Degraded sensory input increases CNS processing demands to extract meaningful information. Neural systems must increase firing rates quadratically to achieve linear improvements in signal-to-noise ratio (Laughlin and Sejnowski 2003), creating disproportionate metabolic costs when processing noisy peripheral signals. This is analogous to listening to conversation in a noisy environment: the brain expends more energy processing degraded input to achieve adequate perception.
Efferent Command Amplification. When efferent small fibers are damaged, the CNS must generate stronger, more frequent, or redundant autonomic commands to achieve target physiological responses. Fewer functional nerve fibers mean each must be driven harder, or signals must be repeated, increasing the metabolic cost of autonomic control. During orthostatic stress, the CNS may detect inadequate vasoconstriction (via baroreceptor feedback) despite issuing normal commands, triggering escalating compensatory signals that further drain central energy reserves.
Testable Predictions.
- Intraepidermal nerve fiber density (IENFD) should inversely correlate with brainstem and hypothalamic glucose uptake (FDG-PET) during autonomic challenges such as tilt testing (\(r < -0.5\) expected).
- ME/CFS patients with confirmed SFN should demonstrate worse cognitive fatigue and brain fog than those without SFN, controlling for pain severity and autonomic dysfunction magnitude.
- Treatment of autoimmune SFN with IVIG should reduce CNS metabolic burden measurable by PET or MR spectroscopy, with corresponding improvements in cognitive symptoms.
- Corneal nerve fiber tortuosity (measured non-invasively via corneal confocal microscopy) should correlate with CNS lactate accumulation and cognitive impairment.
- Cognitive load should exacerbate autonomic dysfunction more severely in SFN-positive patients, reflecting competition for limited CNS energy resources.
Treatment Implications. If this hypothesis is correct, treating SFN may reduce CNS metabolic burden and improve cognitive symptoms even without direct CNS interventions. The non-length-dependent SFN pattern documented in ME/CFS (Azcue et al. 2023) suggests autoimmune etiology, potentially responsive to immunomodulation. Case series (low certainty) suggest IVIG improves pain and autonomic symptoms in autoimmune SFN (Liu et al. 2018) (Oaklander and Nolano 2019), though randomized controlled trials in idiopathic SFN have shown mixed results. The distinct autoimmune pattern in ME/CFS-associated SFN may predict better immunotherapy response than idiopathic cases.
Limitations. No studies have directly measured CNS metabolic demand in relation to SFN severity in ME/CFS. SFN and cognitive dysfunction may share common causes (e.g., autoimmunity or inflammation) rather than having a causal relationship. The relative contribution of SFN to overall CNS energy constraints is unknown and may be minor compared to other factors.
Current Evidence. Azcue et al. documented that ME/CFS patients show prolonged heat response latencies indicating C-fiber dysfunction, with 31% meeting POTS criteria (Azcue et al. 2023). A follow-up study using corneal confocal microscopy demonstrated increased small fiber tortuosity in ME/CFS compared to controls (\(F=6.80\), \(p < 0.01\)), with tortuosity serving as the primary discriminator between patients and controls (AUC\(=0.720\)) (Azcue et al. 2025). The non-length-dependent pattern (upper and lower extremities equally affected) distinguishes ME/CFS-associated SFN from metabolic neuropathies like diabetic neuropathy; by analogy with other immune-mediated neuropathies (Oaklander and Nolano 2019), this distribution pattern is consistent with immune-mediated damage, though the specific antigens targeted in ME/CFS have not been identified. The connection between reduced parasympathetic activation and worse cognitive performance (Azcue et al. 2023) provides indirect support for peripheral-CNS interface dysfunction, though directionality remains uncertain.
Certainty: 0.40 (SFN prevalence established; CNS metabolic mechanism speculative)
The temporal, structural, and interventional evidence is consistent with a model in which vascular and microvascular dysfunction precedes and causes mitochondrial damage, rather than representing a downstream consequence of primary mitochondrial failure. This temporal ordering is the hypothesis, not an established finding: the current evidence is cross-sectional, and the causal direction remains unconfirmed.
Evidence for vascular primacy: + Temporal sequence: Early post-COVID syndrome is dominated by microvascular disturbance (microclots, inflammatory capillary changes), with mitochondrial structural damage appearing later and selectively in patients who develop ME/CFS (Scheibenbogen and Wirth 2024) + Mechanistic chain: Capillary hypoperfusion → ischaemia/reperfusion → proton accumulation → NHE1 sodium influx → NCX1 reversal → mitochondrial Ca(2+) overload → mitochondrial structural damage (Section WASF3–DRP1 Convergence: Two Routes to the Same Mitochondrial Failure in Energy Metabolism and Mitochondrial Function) + Structural irreversibility: Collagen IV deposition and basement membrane thickening in skeletal muscle capillaries (Wüst et al. 2024) represent structural (not merely functional) pathology not explained as consequences of mitochondrial failure + Cell-autonomous RBC sensing failure: ME/CFS RBCs fail to sense and respond to tissue hypoxia (Guo et al. 2025) — a defect independent of target tissue mitochondrial state + Interventional evidence: Correcting preload failure pharmacologically (pyridostigmine) improves peak VO2 and cardiac output (Joseph et al. 2022) — consistent with a model in which preload failure contributes to exercise limitation. Note: pyridostigmine acts via cholinergic autonomic mechanisms, not direct vascular effects; its benefit is consistent with multiple models (autonomic, vascular, mixed) and does not differentially confirm vascular primacy over other explanations
Certainty: 0.45. Evidence for vascular primacy is strengthening but remains circumstantial for the causal direction claim. The temporal model from Scheibenbogen and Wirth is the strongest argument (early PCS vascular-only, later ME/CFS mitochondrial), but requires prospective cohort confirmation. Some patients may have primary mitochondrial pathology without preceding vascular disease (e.g., those with genetic mitochondrial variants).
Falsifiable predictions: + Prospective early post-COVID cohort studies should show vascular abnormalities (microclots, endothelial dysfunction) preceding mitochondrial structural changes in muscle biopsies, with mitochondrial changes only in those who progress to ME/CFS + Vascular-targeted interventions (sulodexide, glycocalyx restoration) implemented early in post-COVID course should reduce ME/CFS incidence + Correcting microvascular obstruction (e.g., targeted fibrinolysis) should normalise mitochondrial energetics in ME/CFS patients with demonstrable capillary obstruction
Limitations: No prospective biopsy-series from early to chronic ME/CFS exists; the Wirth/Scheibenbogen temporal model comes from cross-sectional comparison, not longitudinal tracking; some mitochondrial defects (WASF3) may be primary.
Flaskamp et al. (2022) (Flaskamp et al. 2022) demonstrated that post-COVID serum (without ME/CFS) retains the capacity to induce angiogenic tube formation in endothelial cells in vitro, while ME/CFS serum selectively loses this response. This failed angiogenic compensation may represent the biological transition point: post-COVID patients who retain vascular repair capacity recover; those in whom this mechanism fails progress to ME/CFS.
Certainty: 0.30. The in vitro finding is from a single study with small numbers; the clinical implication (failed angiogenesis as the transition event) is extrapolated, not directly tested.
Falsifiable predictions: + Longitudinal sampling of post-COVID patients before and after ME/CFS development should show loss of angiogenic serum activity coinciding with clinical deterioration + Angiogenic capacity of patient serum should correlate inversely with capillary basement membrane thickness on biopsy + Pro-angiogenic interventions (e.g., VEGF-based approaches, hypoxia-inducible factor stabilisers) should attenuate ME/CFS development in at-risk post-COVID patients if given early
CAUTION regarding VEGF-based approaches: The Ribeiro et al. (2026) (Ribeiro et al. 2026) findings complicate the straightforward pro-angiogenic rationale. In post-viral ME/CFS specifically, VEGF elevation from sustained HIF-2\(\alpha\) activation may itself be pathological — driving capillary permeability rather than angiogenesis. Exogenous VEGF or HIF stabilisation (which also increases HIF-2\(\alpha\)) could worsen endothelial barrier dysfunction in patients whose HIF-2\(\alpha\) is already pathologically active (Section Immune Complex–Endothelial Injury as a Central Vascular Mechanism). This means pro-angiogenic strategies must be stratified: patients with elevated VEGF/vWF (HIF-2\(\alpha\)-active phenotype) would be harmed, while those with blunted VEGF (HIF-inert phenotype) might benefit. VEGF or sFlt-1 measurement is recommended before any pro-angiogenic intervention.
Limitations: Single in vitro study; serum angiogenic activity is a composite of many factors; mechanism of selective angiogenic failure in ME/CFS is not established. The HIF-2\(\alpha\) stratification caveat is theoretical — no interventional data exist.
1 Neuroimmune POTS: Autoimmune Evidence and Genetic Predisposition
{{/* ch10: POTS autoimmunity integration */}}
Certainty: 0.60. The evidence that POTS has an autoimmune basis has strengthened substantially and now reaches moderate-high certainty. Multiple independent lines converge: (1) GPCR autoantibodies (α1, β1/β2, M2/M4) found in 89% of POTS patients (ELISA, Gunning 2019 (Gunning et al. 2019)) with functional validation in cell-based assays showing IgG-dependent activation of α1AR, β1AR, and β2AR (Fedorowski et al. 2017); (2) passive transfer: POTS patient IgG induces tachycardia in rabbits (Li et al. 2019), satisfying the key Koch-like criterion for causal autoantibody involvement; (3) serum GPCR autoantibody activity correlates with orthostatic symptom severity in cross-sectional analysis (Lund cohort, Kharraziha 2020 functional bioassay (Kharraziha et al. 2020)); (4) HLA associations: DRB115:01 and DQB106:02 significantly associated with POTS in a Korean cohort (n=93 POTS, 196 controls) (Shin et al. 2019), suggesting an autoimmune genetic predisposition; (5) elevated autoimmune comorbidity prevalence (37% ANA ≥1:80, Hashimoto thyroiditis, SLE, RA, Sjögren’s) (Blitshteyn 2015); (6) immunomodulatory treatment benefit (IVIG, immunoadsorption) in case series.
Mechanistic model. POTS autoimmunity involves multiple autoimmune targets: (a) GPCR autoantibodies directly alter cardiovascular receptor function (α1 → vasoconstriction, β1/β2 → heart rate contractility, M2 → vagal inhibition); (b) autoantibodies may target autonomic ganglia (ganglionic AChR antibodies) impairing signal transmission between pre- and post-ganglionic neurons; (c) brainstem neuroinflammation (Section Brainstem Neuroinflammation at Dorsolateral Inferior Medulla as a Shared CNS Substrate for POTS, ME/CFS, and Long COVID) at NTS/RVLM may be triggered or maintained by circulating autoantibodies accessing medulla via area postrema.
Cross-disease significance. The same GPCR autoantibody profile found in POTS is also found in subsets of ME/CFS and Long COVID patients, supporting the Blitshteyn 2026 thesis that these three conditions converge on shared neuroimmune mechanisms. The HLA association (DRB1*15:01) may represent a common genetic susceptibility across the three conditions, though this has not been tested in ME/CFS or Long COVID.
Limitations. GPCR autoantibody detection is not standardized across labs. The Germain 2025 null (Germain et al. 2025) using REAP/Luminex (in ME/CFS, not POTS) raises the possibility that assay sensitivity issues confound prevalence estimates. The HLA association (Shin 2019) is single-cohort (Korean), not replicated, and may not generalize. No POTS-specific animal model demonstrating disease transfer by GPCR autoantibodies alone has been developed (the rabbit model (Li et al. 2019) shows tachycardia but not the full POTS phenotype).
Replication status. GPCR autoantibodies in POTS replicated by three independent groups (Oklahoma (Li et al. 2014), Lund (Fedorowski et al. 2017), CellTrend/Germany (Gunning et al. 2019)). HLA association (Shin 2019) not yet replicated.
{{/* H5: Ganglionic AChR AAb pan-autonomic ME/CFS (Tier 2, cert 0.25) */}}
Certainty: 0.25. A subset of ME/CFS patients with pan-autonomic involvement (all domains of COMPASS-31 abnormal — orthostatic, GI, thermoregulatory, pupillary, secretomotor) may have ganglionic AChR (α3 subunit) autoantibodies that impair signal transmission from pre- to post-ganglionic autonomic neurons. This would represent an autoimmune autonomic ganglionopathy (AAG) look-alike that responds to standard AAG treatment (IVIG 2 g/kg monthly or immunoadsorption). The ganglionic AChR autoantibody finding in POTS (Blitshteyn, Doherty, and Steinman 2026) (El-Rhermoul et al. 2023) provides the bridge.
Mechanism. Autoantibodies bind the α3 subunit of the nicotinic acetylcholine receptor on autonomic ganglia. This blocks fast synaptic transmission from preganglionic to postganglionic autonomic neurons, producing: (a) impaired sympathetic vasoconstriction → orthostatic pooling → POTS; (b) impaired parasympathetic GI motility → gastroparesis; (c) impaired sudomotor function → thermoregulatory dysfunction; (d) impaired pupillary function → blurred vision. The global pattern — pan-autonomic failure — is the hallmark that distinguishes this subtype from isolated GPCR autoantibody-mediated POTS.
Evidence. Blitshteyn 2026 cites autonomic ganglia as autoimmune targets. AAG is a well-characterized condition with known autoantibody target (ganglionic AChR) and established treatment algorithm (IVIG, immunoadsorption). The autoantibody assays (ganglionic AChR ELISA) are commercially available.
Limitations. Ganglionic AChR autoantibodies have been tested in POTS (some positive, some negative) but not systematically in ME/CFS. AAG is rare (estimated 1:1,000,000) — even if enriched in ME/CFS, the absolute prevalence may be small (under 5% of ME/CFS). The COMPASS-31 pattern predicting ganglionic AChR positivity has not been validated.
Replication status. Not tested in ME/CFS.
Falsifiable prediction: ME/CFS patients with COMPASS-31 >60 will show ≥15% positivity for ganglionic AChR (α3) autoantibodies. Falsified if under 5% positive in this subgroup.
2 Speculative Endothelial Diagnostics and Therapeutics
Watton emphasises that endothelial cells appear normal at rest but fail under stress. Could a standardised provocative endothelial stress test — exposing patient-derived endothelial cells (or in vivo flow-mediated dilation with metabolic challenge) to oxidative (H2O2), immune (TNF-\(\alpha\)), or metabolic (glucose/insulin) challenges — quantify individual adaptive capacity and serve as a clinical diagnostic? Falsifiable: ME/CFS patients will show normal baseline but impaired upregulation of oxidative phosphorylation and delayed barrier recovery after challenge compared to controls, with impairment magnitude correlating with PEM severity. Probability of diagnostic utility: 0.20. (Watton and Prusty 2026)
Certainty: 0.38. Probability of clinically meaningful efficacy: 0.03. If Fn1-IgG complex dysregulation impairs immune-clearance and endothelial resilience, enhancing fibrinolytic activity (tissue plasminogen activator, PAI-1 modulators) could normalise Fn1 handling. However, fibrinolytic therapy carries haemorrhagic risk and is indicated only in acute thrombosis — the risk-benefit ratio for a chronic, non-thrombotic condition is likely unfavourable. The concept is mechanistically interesting but the therapeutic window appears prohibitive for ME/CFS. Falsifiable: plasminogen activator or PAI-1 inhibition in ME/CFS plasma will increase Fn1 content in IgG-bound CICs and improve endothelial barrier function under stress. (Watton and Prusty 2026)
Certainty: 0.50. Probability of clinically meaningful efficacy: 0.06. Moreau’s group identified reduced functional haptoglobin capacity permitting exaggerated haem-mediated oxidative stress during exertion. Heme scavengers — haptoglobin supplementation, hemopexin, or HO-1 inducers (sulforaphane) — could provide complementary haem clearance during PEM when endogenous Hp is insufficient. Hp supplementation is conceptually clean but no pharmaceutical Hp product exists for non-acute use. Sulforaphane (broccoli sprout extract) is a dietary HO-1 inducer with human safety data. Falsifiable: heme scavenger treatment during exertion challenge will reduce plasma free haem, decrease endothelial activation markers (VCAM-1, E-selectin), and preserve barrier integrity. (Watton and Prusty 2026)
Endothelial dysfunction is central to the unified model but flow-mediated dilation is inaccessible outside specialised centres. Could remote endothelial assessment — pulse wave velocity, digital thermal monitoring, or endothelial progenitor cell quantification via dried blood spot — enable primary care screening and early intervention? Falsifiable: remote endothelial assessment will correlate with flow-mediated dilation and identify patients at risk for severe complications earlier than symptom-based referral. Probability of clinical utility: 0.12. (Watton and Prusty 2026)
3 Vasomotor Dysfunction and Thermoregulatory Failure
Certainty: 0.40. ME/CFS patients paradoxically report both heat and cold intolerance — two symptoms that appear to require opposite vascular states. Heat intolerance requires vasodilation failure (cannot dissipate heat), cold intolerance requires thermogenesis failure (cannot conserve or generate heat). Cambras et al. demonstrated that skin temperature circadian rhythm disruption is associated with elevated endothelin-1 (ET-1), a potent vasoconstrictor (Cambras et al. 2023). This suggests a vasomotor bias toward tonic vasoconstriction: ET-1 elevation may lock cutaneous vessels in a chronically constricted state, preventing both heat-induced vasodilation (→ heat intolerance) and cold-induced perfusion for thermogenesis (→ cold intolerance). Normally, cutaneous vessels flip-flop between dilated (heat dissipation) and constricted (heat conservation) states with high dynamic range. In ME/CFS, the vasomotor setpoint may be shifted toward constriction, compressing the dynamic range from both sides — vessels cannot dilate enough for cooling, and cannot recruit enough perfusion for warming. This single mechanism would explain the dual-intolerance paradox without requiring separate heat and cold pathways. Falsifiable: laser Doppler flowmetry will show baseline cutaneous perfusion 40% lower in ME/CFS than controls; both heat challenge (43°C) and cold challenge (15°C) will elicit blunted perfusion responses in ME/CFS; ET-1 receptor antagonism will restore perfusion flexibility.
Cold adaptation requires brown adipose tissue (BAT) activation via sympathetic nervous system norepinephrine release at β3-adrenergic receptors, triggering UCP1-mediated thermogenesis. ME/CFS sympathetic dysregulation — documented adrenergic receptor dysfunction, POTS, and reduced HRV — may prevent BAT activation, causing cold intolerance via failed thermogenesis. This would mirror heat intolerance from failed vasodilation: both reflect sympathetic output that is tonically elevated but dynamically inflexible — high enough to maintain vasoconstriction, but unable to mount the coordinated sympathetic surge needed for BAT activation. Falsifiable: PET-CT with ^18F-FDG after controlled cold exposure (16°C, 2 h) will show absent BAT activation in ME/CFS vs age/BMI-matched controls; β3-agonist (mirabegron) will restore BAT uptake and cold tolerance. This is mechanistically grounded in established BAT physiology but entirely untested in ME/CFS.
Certainty: 0.40. Probability of clinically meaningful efficacy: 0.08. Post-exertional malaise (PEM) involves an inflammatory cascade peaking 6–48 h after exertion: C4a complement elevation (Sorensen et al. 2003), IL-1beta and IL-6 surges, and prostaglandin E2 (PGE2) release from COX-2 upregulation in activated immune cells and endothelium. If COX-2-derived PGE2 is a necessary amplifier of the PEM inflammatory cascade, selective COX-2 inhibition (celecoxib) timed to the peri-exertional window could prevent or attenuate PEM by blocking PGE2 synthesis at its peak.
Mechanism. Exertion in ME/CFS triggers: (1) complement activation → C5a → neutrophil/monocyte COX-2 upregulation; (2) mechanical stress on skeletal muscle → COX-2 induction in myocytes and satellite cells; (3) endothelial shear stress → endothelial COX-2 expression. COX-2 converts arachidonic acid to PGH2, the precursor of PGE2, PGD2, PGI2, and TXA2. PGE2 amplifies the inflammatory signal through EP2/EP4 receptor-mediated cAMP signalling and NF-kappaB activation. If PGE2 is a positive feedback amplifier rather than a downstream bystander in the PEM cascade, blocking its synthesis during the critical 6–24 h post-exertion window could reduce PEM magnitude without requiring full anti-inflammatory immunosuppression.
Timing hypothesis. Standard NSAID dosing (TID, continuous) may not prevent PEM because: (a) continuous COX-2 inhibition allows upstream complement activation to proceed unopposed, and (b) the prostanoid mediator mix after exertion changes over time (PGE2 dominant at 6–12 h, PGI2 at 24–48 h). A peri-exertional protocol — celecoxib 200 mg at exertion onset + 200 mg at 12 h — would target the PGE2-rich window while avoiding continuous suppression. The protocol differs from standard COX-2 inhibitor use in requiring exertion-triggered, not scheduled, dosing.
Safety considerations. Cardiovascular safety is paramount in a population with already-impaired endothelial function: selective COX-2 inhibitors carry cardiovascular risk (increased thrombotic events) by suppressing endothelial PGI2 without platelet COX-1 inhibition. Celecoxib has the most favourable cardiovascular profile among COX-2-selective NSAIDs at 200 mg BID. However, in ME/CFS patients with POTS, hypovolemia, or endothelial dysfunction, even short-course COX-2 inhibition may: (1) raise blood pressure (sodium retention, reduced renal PGE2); (2) impair flow-mediated dilation; (3) interact with antihypertensives (ACE inhibitors, ARBs, beta-blockers common in this population). The risk-benefit ratio is unfavourable unless the PEM-prevention effect size is large (≥50% PEM severity reduction).
Falsifiable predictions. (1) Randomised, placebo-controlled, within-subject crossover trial (n=20, two identical standardized exercise challenges separated by ≥4 weeks): celecoxib 200 mg at exertion start + 200 mg at 12 h will reduce 24 h PEM severity (visual analogue scale) by ≥30% vs placebo. (2) Urinary PGE-M (PGE2 metabolite) at 6 h post-exertion will be suppressed by celecoxib and correlate with PEM severity. (3) C4a and IL-1beta elevations will NOT be suppressed by celecoxib (confirming COX-2 acts downstream of complement/cytokine triggering, not upstream). (4) Patients with the largest C4a spike at 6 h will show the greatest celecoxib response.
Limitations. No ME/CFS-specific COX-2 data exist. The cardiovascular risk in a population with known endothelial dysfunction is a serious concern — a negative safety outcome would outweigh any potential PEM benefit. Celecoxib is a moderate CYP2C9 substrate; ME/CFS patients on other CYP2C9-metabolised drugs (cannabidiol, losartan, fluoxetine) may have altered exposure. The peri-exertional dosing protocol requires patients to predict exertion — feasible for planned activities but not for spontaneous PEM triggers.