Pharmacodiagnostic Matrix - Diagnostic Methodology

1 Laragh et al. 1988 — Diagnosis Ex Juvantibus: Individual Response Patterns to Drugs Reveal Hypertension Mechanisms

Full Citation:: Laragh JH, Lamport B, Sealey J, Alderman MH. Diagnosis ex juvantibus. Individual response patterns to drugs reveal hypertension mechanisms and simplify treatment. Hypertension. 1988;12(3):223–226. (Laragh et al. 1988)
DOI:: https://doi.org/10.1161/01.hyp.12.3.223
PMID:: 3049337
Key Findings::
- Heterogeneity of response to antihypertensive therapy is well-recognized: an agent antihypertensive in one patient may increase BP in another or have no effect in a third
- Sequential trials of drugs with different mechanisms (diuretics, calcium channel blockers, α-blockers, β-blockers, ACE inhibitors) classify patients into biologically relevant groups
- Drug response patterns provide a framework for investigating pathophysiology and stratifying clinical trial participants
- Conceptually establishes “diagnosis ex juvantibus” — drug response as diagnostic evidence
Conclusion:: Foundational paper establishing that multi-drug response heterogeneity is not noise but signal — it reveals the underlying mechanism. This is the direct conceptual ancestor of the pharmacodiagnostic matrix proposal.
Limitations:: Conceptual/review paper, not empirical trial; hypertension-specific; no formal constraint-satisfaction or algorithmic framework.

2 Samuel et al. 2023 — N-of-1 Trials vs. Usual Care in Children with Hypertension: A Pilot Randomized Clinical Trial

Full Citation:: Samuel JP, Bell CS, Samuels JA, Rajan C, Walton AK, Green C, Tyson JE. N-of-1 trials vs. usual care in children with hypertension: a pilot randomized clinical trial. American Journal of Hypertension. 2023;36(2):126–132. (Samuel et al. 2023)
DOI:: https://doi.org/10.1093/ajh/hpac117
PMID:: 36227203
Key Findings::
- RCT: n-of-1 trial arm (n=23) vs usual care (n=26) in pediatric hypertension
- 69% probability n-of-1 trials increased BP control at 6 months (Bayesian OR 1.24, 95% CrI 0.51–2.97)
- 93% probability of greater systolic BP reduction in n-of-1 group (mean difference −3.6 mmHg)
- Bayesian framework used for single-patient trial inference
Conclusion:: Direct RCT evidence that n-of-1 methodology improves treatment outcomes. The Bayesian framework used is directly applicable to multi-drug pharmacodiagnostic inference in ME/CFS.
Limitations:: Small pilot trial (n=49); pediatric hypertension only; 6-month follow-up; single center.

3 Samuel et al. 2019 — Treating Hypertension in Children with N-of-1 Trials

Full Citation:: Samuel JP, Tyson JE, Green C, Bell CS, Pedroza C, Molony D, Samuels J. Treating hypertension in children with n-of-1 trials. Pediatrics. 2019;143(4):e20181818. (Samuel et al. 2019)
DOI:: https://doi.org/10.1542/peds.2018-1818
PMID:: 30842257
Key Findings::
- n=32 children with hypertension: compared lisinopril, amlodipine, hydrochlorothiazide via repeated ABPM
- No single medication preferred for >50% of patients; 49% lisinopril, 24% amlodipine, 12% HCTZ, 4/32 uncontrolled on any monotherapy
- Conservative Bayesian analyses with credible intervals for each treatment preference proportion
- Unacceptable side effects in 24% HCTZ, 16% lisinopril, 13% amlodipine
Conclusion:: Demonstrates that treatment response is genuinely heterogeneous — no “first-line for everyone.” Individualized n-of-1 testing reveals the optimal drug per patient. Directly supports the premise that multi-drug response patterns carry diagnostic information.
Limitations:: Single center; pediatric only; three drug comparison; no mechanistic hypotheses tested.

4 Friston 2023 — Computational Psychiatry: From Synapses to Sentience

Full Citation:: Friston K. Computational psychiatry: from synapses to sentience. Molecular Psychiatry. 2023;28(1):256–268. (Friston 2023)
DOI:: https://doi.org/10.1038/s41380-022-01743-z
PMID:: 36056173
Key Findings::
- Brain as generative model that underwrites sentient processing and scientific diagnosis
- Neuronal message passing, belief propagation, and precision encoding as formal Bayesian inference
- Dysconnection → aberrant belief updating → false inference — bridges pathophysiology to phenomenology
- Computational phenotyping and computational nosology as emerging frameworks
Conclusion:: Provides the formal mathematical infrastructure for pharmacodiagnostic reasoning: drug response = new evidence that updates a Bayesian diagnostic belief distribution over competing mechanistic hypotheses.
Limitations:: Review article; primarily psychiatric; theoretical framework rather than empirical validation; computational phenotyping reliability concerns not addressed.

5 Strauss et al. 2021 — Data-Driven Identification of Complex Disease Phenotypes

Full Citation:: Strauss MJ, Niederkrotenthaler T, Thurner S, Kautzky-Willer A, Klimek P. Data-driven identification of complex disease phenotypes. Journal of the Royal Society Interface. 2021;18(180):20201040. (Strauss et al. 2021)
DOI:: https://doi.org/10.1098/rsif.2020.1040
PMID:: 34314651
Key Findings::
- Population-wide (n=9M) medical claims dataset used to construct disease interaction network
- Higher-order diagnosis features capturing interactions between ≥2 diseases better characterize patient health states
- Data-driven detection of metabolically healthy vs unhealthy obesity phenotypes — previously controversial distinction
- Metabolically healthy obesity shows progression toward unhealthy over time
Conclusion:: Multi-feature interaction patterns reveal phenotypic subtypes invisible to single-feature analysis. The pharmacodiagnostic matrix extends this logic: multi-drug response interaction patterns reveal bottleneck subtypes invisible to single-drug analysis.
Limitations:: Claims data (limited clinical detail); cross-sectional network; obesity-specific phenotypes; no medication response data.

6 Scheibenbogen et al. 2018 — Immunoadsorption to Remove β2 Adrenergic Receptor Antibodies in CFS/ME

Full Citation:: Scheibenbogen C, Loebel M, Freitag H, Krueger A, Bauer S, Antelmann M, Doehner W, Scherbakov N, Heidecke H, Reinke P, Volk HD, Grabowski P. Immunoadsorption to remove β2 adrenergic receptor antibodies in Chronic Fatigue Syndrome. PLoS One. 2018;13(3):e0193672. (Scheibenbogen et al. 2018)
DOI:: https://doi.org/10.1371/journal.pone.0193672
PMID:: 29543914
Key Findings::
- n=10 post-infectious ME/CFS with elevated β2AdR autoantibodies; 5-day immunoadsorption
- 7/10 rapid improvement during IA; 3/10 long-lasting improvement (6–12+ months); 2/10 short; 2/10 improvement after initial worsening
- β2 IgG antibodies rapidly decreased during IA; memory B cells decreased, plasma cells increased
- Heterogeneous response trajectories suggest mechanistic subgroups among β2AdR-positive patients
Conclusion:: Exemplifies pharmacodiagnostic principle in ME/CFS: same intervention produces qualitatively different response patterns (sustained, transient, delayed), implying different underlying mechanisms even within an autoantibody-defined subgroup. This heterogeneity is the signal the pharmacodiagnostic matrix aims to systematize.
Limitations:: Small n (10); open-label; single-arm; uncontrolled confounding from concomitant medication; IA procedure itself non-specific.

7 Zhang et al. 2022 — Ruxolitinib Response-Based Stratified Treatment for Pediatric HLH

Full Citation:: Zhang Q, Zhao YZ, Ma HH, Wang D, Cui L, Li WJ, Wei A, Wang CJ, Wang TY, Li ZG, Zhang R. A study of ruxolitinib response-based stratified treatment for pediatric hemophagocytic lymphohistiocytosis. Blood. 2022;139(24):3493–3504. (Zhang et al. 2022)
DOI:: https://doi.org/10.1182/blood.2021014860
PMID:: 35344583
Key Findings::
- n=52 newly diagnosed pediatric HLH: ruxolitinib monotherapy as first-line agent
- ORR 69.2% at day 28; 42.3% sustained complete remission; all responders achieved first response within 3 days
- Response significantly associated with underlying etiology (EBV-HLH most sensitive, ORR 87.5%)
- 57.7% stratified to additional chemotherapy based on early response; median interval to additional treatment: 6 days
- 12-month overall survival 86.4%
Conclusion:: Direct demonstration of response-based stratified treatment: early drug response reveals etiology and guides next therapeutic step. This is the clinical paradigm for a pharmacodiagnostic decision tree — drug A response patterns determine whether to escalate or switch.
Limitations:: Single country (China); pediatric only; HLH-specific (rare disease); ruxolitinib-specific; no formal Bayesian/diagnostic framework applied to response data.

8 Schaaf et al. 2024 — Test-Retest Reliability of Reinforcement Learning Parameters

Full Citation:: Schaaf JV, Weidinger L, Molleman L, van den Bos W. Test-retest reliability of reinforcement learning parameters. Behavior Research Methods. 2024;56(5):4582–4599. (Schaaf et al. 2024)
DOI:: https://doi.org/10.3758/s13428-023-02203-4
PMID:: 37684495
Key Findings::
- n=69 (bandit task) + n=47 (reversal learning task); 5-week test-retest interval
- RL model parameters showed poor test-retest reliability: ICC 0.02–0.52 for bandit, 0.01–0.71 for reversal learning
- Mood (stress, happiness) partially explained within-participant variability
- Simulations confirmed procedures could detect high reliability if present — unreliability is genuine
Conclusion:: Critical caution for computational phenotyping: if RL parameters are this unstable, drug-response-derived diagnostic parameters may also show substantial within-patient variability. Pharmacodiagnostic matrix must incorporate repeat-measurement designs and account for state-dependent confounds (mood, sleep, recent exertion, infection status).
Limitations:: Healthy volunteers; online testing; RL-specific tasks; 5-week interval may not generalize to drug response timecourses.

9 Falaguera-Vera et al. 2020 — Pressure Point Thresholds and ME/CFS Comorbidity as Indicators of Treatment Response in Fibromyalgia

Full Citation:: Falaguera-Vera FJ, Garcia-Escudero M, Bonastre-Férez J, Zacarés M, Oltra E. Pressure point thresholds and ME/CFS comorbidity as indicators of patient’s response to manual physiotherapy in fibromyalgia. International Journal of Environmental Research and Public Health. 2020;17(21):8044. (Falaguera-Vera et al. 2020)
DOI:: https://doi.org/10.3390/ijerph17218044
PMID:: 33142896
Key Findings::
- Baseline pressure point thresholds (PPT) inversely correlate with treatment response to manual physiotherapy in fibromyalgia
- Post-stratification revealed ME/CFS comorbidity as significant effect modifier — FM+ME/CFS patients respond differently than FM-only
- Gene expression profiling collected for differential response analysis
Conclusion:: Direct ME/CFS-relevant evidence: baseline biomarkers + comorbidity status predict differential treatment response. This is a two-axis constraint on a pharmacodiagnostic matrix: PPT × ME/CFS-status narrows which patients benefit from manual therapy.
Limitations:: Modest sample size; single physiotherapy intervention; exploratory post-hoc subgroup analysis; gene expression results not yet reported.

10 Hu et al. 2024 — Approaches to Early Parkinson’s Disease Subtyping

Full Citation:: Hu M, Skjærbæk C, Borghammer P. Approaches to early Parkinson’s disease subtyping. Journal of Parkinson’s Disease. 2024;14(s2):S297–S306. (Hu, Skjærbæk, and Borghammer 2024)
DOI:: https://doi.org/10.3233/JPD-230419
PMID:: 39331104
Key Findings::
- PD subtyping at clinical, genetic, imaging, and molecular levels improves prediction of medication response, clinical endpoints, and progression trajectory
- Subtyping reduces confounding effects of heterogeneity in drug trials
- Multi-dimensional classification needed: integrates genetic, pathological, imaging, multi-omics markers
- Theoretical a priori disease models prerequisite to understanding biological subtypes
Conclusion:: PD subtyping is the most mature parallel to proposed ME/CFS pharmacodiagnostic subtyping. Medication response itself is used as a subtyping dimension — the same dimensions used for classification can be predicted by the classification.
Limitations:: Review article; PD-specific (α-synuclein pathology); most subtyping methods not yet validated for clinical decision-making; no cross-disease comparison.

11 Newberry et al. 2016 — Diagnosis of Gout (AHRQ Comparative Effectiveness Review)

Full Citation:: Newberry SJ, FitzGerald J, Maglione MA, O’Hanlon CE, Han D, Booth M, Motala A, Tariq A, Dudley W, Shanman R, Shekelle PG. Diagnosis of Gout. AHRQ Comparative Effectiveness Reviews. 2016;Report No.: 15(16)-EHC026-EF. (Newberry et al. 2016)
PMID:: 26985540
Key Findings::
- Clinical algorithms (Diagnostic Rule: sensitivity 88%, specificity 75%; CGD: sensitivity 97%, specificity 96%) form diagnostic decision tree
- DECT imaging: sensitivity 85–100%, specificity 83–92%; ultrasound: sensitivity 37–100%, specificity 68–97%
- Recommendation: clinical algorithm → rule in/out → escalate to imaging/synovial fluid analysis for ambiguous cases
- Missed diagnosis resulted in unnecessary surgery, longer hospital stays, delayed treatment
Conclusion:: Explicit diagnostic decision tree methodology: tiered escalation based on test performance characteristics. Paradigm for pharmacodiagnostic decision tree: drug A response → narrows hypothesis space → drug B to distinguish remaining hypotheses → escalate to laboratory testing only for residual ambiguity.
Limitations:: Strength of evidence rated low due to limited validation; gout-specific; most studies in academic rheumatology settings, not primary care.

12 Ravichandran et al. 2024 — Active Learning with Human Heuristics: An Algorithm Robust to Labeling Bias

Full Citation:: Ravichandran S, Sudarsanam N, Ravindran B, Katsikopoulos KV. Active learning with human heuristics: an algorithm robust to labeling bias. Frontiers in Artificial Intelligence. 2024;7:1491932. (Ravichandran et al. 2024)
DOI:: https://doi.org/10.3389/frai.2024.1491932
PMID:: 39628837
Key Findings::
- 4 active learning algorithms × 2 human heuristics (fast-and-frugal tree, tallying) × 3 classifiers × 15 datasets
- Human labeling heuristics significantly degrade active learning performance — sometimes below random
- Proposed inverse information density algorithm robust to labeling bias, achieving 87% improvement
Conclusion:: Critical caution for pharmacodiagnostic implementation: clinician heuristics and biases in labeling “responder” vs “non-responder” degrade diagnostic inference. Any pharmacodiagnostic algorithm must be robust to labeling noise — patients may misattribute improvement to the wrong drug, or clinicians may use heuristics rather than systematic criteria.
Limitations:: Simulation study, not clinical; labeling heuristics are stylized (fast-and-frugal tree, tallying) — real clinical labeling biases may differ; no medical domain datasets specifically tested.

13 Monini et al. 2006 — Silent Reflux: Ex Juvantibus Criteria for Diagnosis and Treatment of Laryngeal Disorders

Full Citation:: Monini S, Di Stadio A, Vestri A, Barbara M. Silent reflux: ex juvantibus criteria for diagnosis and treatment of laryngeal disorders. Acta Otorhinolaryngologica. 2006;126(8):866–871. (Monini et al. 2006)
DOI:: https://doi.org/10.1080/00016480500504242
PMID:: 16846931
Key Findings::
- n=60 patients with suspected laryngopharyngeal reflux; randomized to omeprazole or immunostimulant for 3 months
- Omeprazole-treated patients showed significant improvement of laryngeal features and symptoms
- PPI response used as “ex adjuvantibus” diagnostic criterion for confirming reflux-associated laryngitis
Conclusion:: Clinical implementation of diagnosis ex juvantibus: therapeutic response to PPI confirms the reflux diagnosis. Demonstrates that drug-probe methodology is already used in clinical practice for conditions where objective biomarkers are lacking — directly analogous to ME/CFS.
Limitations:: Small RCT (n=60); single center; short follow-up (3 months); laryngopharyngeal reflux only.

14 Kim et al. 2023 — Characterizing Sjögren-Associated Fatigue: A Distinct Phenotype from ME/CFS

Full Citation:: Kim L, Kedor C, Buttgereit F, Heidecke H, Schaumburg D, Scheibenbogen C. Characterizing Sjögren-associated fatigue: a distinct phenotype from ME/CFS. Journal of Clinical Medicine. 2023;12(15):4994. (Kim et al. 2023)
DOI:: https://doi.org/10.3390/jcm12154994
PMID:: 37568396
Key Findings::
- n=19 primary Sjögren’s syndrome patients assessed by Canadian Consensus Criteria for ME/CFS
- 4/18 fulfilled CCC; PEM atypical — triggered by mental/emotional but not physical exertion
- Handgrip strength fully recovered 1 hour post-exertion (contrast to ME/CFS where recovery is absent/delayed)
- β1-, β2-, M4-receptor autoantibodies elevated and correlated with disease activity (ESSDAI) but not fatigue severity
Conclusion:: Exercise challenge test differentiates pSS fatigue from ME/CFS PEM. Demonstrates that physiological challenge responses carry diagnostic information for distinguishing fatiguing illnesses — a core principle of the pharmacodiagnostic matrix.
Limitations:: Small sample (n=19); single-center; cross-sectional; autoantibody measurement by CellTrend ELISA (limited independent validation).

15 Wirth & Scheibenbogen 2021 — Pathophysiology of Skeletal Muscle Disturbances in ME/CFS

Full Citation:: Wirth KJ, Scheibenbogen C. Pathophysiology of skeletal muscle disturbances in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS). Journal of Translational Medicine. 2021;19(1):162. (Wirth and Scheibenbogen 2021)
DOI:: https://doi.org/10.1186/s12967-021-02833-2
PMID:: 33882940
Key Findings::
- β2AdR dysfunction → impaired Na+/K+-ATPase stimulation → intracellular sodium overload
- Sodium overload → NCX reversal (Na+/Ca2+ exchanger imports calcium instead of exporting it)
- Calcium overload → mitochondrial dysfunction, metabolic disturbance, endothelial dysfunction → PEM vicious circle
- NHE1 upregulation lowers re-induction threshold — repeated PEM at progressively lower exertion levels
- Merges β2AdR autoantibody hypothesis with skeletal muscle bioenergetic disturbance
Conclusion:: Detailed testable mechanistic model: if β2AdR dysfunction is the driver, patients with this mechanism should respond to β2AdR-targeted interventions. This hypothesis generates a specific pharmacodiagnostic prediction: β2-agonist response → β2AdR pathway involvement → narrower differential → tailored treatment.
Limitations:: Hypothesis paper — no direct trial data; animal/ischemia-reperfusion literature extrapolated to ME/CFS; single mechanism (β2AdR); NHE1 role in human ME/CFS muscle unvalidated.

16 Schwartenbeck & Friston 2016 — Computational Phenotyping in Psychiatry: A Worked Example

Full Citation:: Schwartenbeck P, Friston K. Computational phenotyping in psychiatry: a worked example. eNeuro. 2016;3(4):ENEURO.0049-16.2016. (Schwartenbeck and Friston 2016)
DOI:: https://doi.org/10.1523/ENEURO.0049-16.2016
PMID:: 27517087
Key Findings::
- Step-by-step demonstration: build computational model → simulate data → invert model → estimate group effects
- Uses two-step maze task and active inference Markov decision process model
- Cross-validation to assess whether between-subject variables (e.g., diagnosis) can be recovered from model parameters
- Procedures applicable across computational psychiatry domains
Conclusion:: Concrete methodological template for pharmacodiagnostic computational phenotyping: model drug response as a decision process, infer subject-level parameters, test whether parameters discriminate mechanistic subgroups.
Limitations:: Single worked example; simulated data validation; two-step maze task (simple); no patient data; active inference framework not universally adopted.

17 Charlton et al. 2025 — Skeletal Muscle Properties in Long COVID and ME/CFS Differ from Those Induced by Bed Rest

Full Citation:: Charlton BT, Slaghekke A, Appelman B, Eggelbusch M, Huijts JY, Noort W, Hendrickse PW, Bloemers FW, van Amstel P, Posthuma JJ, Goulding RP, van Vugt M, Wüst RCI. Skeletal muscle properties in long COVID and ME/CFS differ from those induced by bed rest. medRxiv. 2025. (Charlton et al. 2025)
DOI:: https://doi.org/10.1101/2025.06.23.25330153
Study Design:: Cross-sectional with comparator cohort; maximal cycling test + VO₂max + NIRS + vastus lateralis biopsy (IHC + EM)
Sample Size:: Long COVID n=24, ME/CFS n=26 (pre-2020 dx), healthy controls n=30, strict 60-day bed-rest cohort n=10
Key Findings::
- Lower VO₂max (p \(<\) 0.0001) and peak power output (p \(<\) 0.0001) in both Long COVID and ME/CFS vs healthy controls
- Lower tissue O₂ uptake (p = 0.001) and vasodilatory capacity (p = 0.011) in both patient groups
- Lower capillarization only in ME/CFS (p \(<\) 0.0005)
- Both groups: decreased capillary contact lengths (p \(<\) 0.0001), decreased capillary tortuosity (p \(<\) 0.0001)
- Substantial capillary basement membrane thickening (p \(<\) 0.0001) — max thickness in patients far exceeds controls
- EM confirmed: basement membrane thickening, microvacuolization, endothelial hypertrophy, degeneration signs
- Bed rest: muscle atrophy + reduced OXPHOS. Patients: no atrophy, but less capillaries, more glycolytic fibers — neither explained VO₂max
Conclusion:: Physical inactivity/deconditioning alone does not explain lower exercise capacity in Long COVID and ME/CFS; microvascular pathology — especially capillary basement membrane thickening and reduced tortuosity — is a primary contributor to impaired O₂ extraction. Full paper corresponding to Slaghekke et al. 2026 AMS abstract.
Certainty Assessment::
- Quality: Medium (preprint only, not yet peer-reviewed; rigorous methodology with gold-standard IHC+EM and bed-rest comparator)
- Sample: Medium (n=80 total across all groups)
- Replication: Partially replicated by Aschman 2023 (capillary BM thickening in PCS), Agergaard 2023 (capillary injury in Long COVID), Joseph 2022 (O₂ extraction impairment in ME/CFS). Core finding (capillary BM thickening) by independent Berlin group.
- Cohort: Amsterdam UMC (VU/VUmc); recruitment years not fully specified (ME/CFS dx pre-2020); PIs: van Vugt, Wüst, Goulding, Charlton
- Score: 0.55
Limitations:: Preprint (not peer-reviewed); modest sample sizes per group; ME/CFS group recruited pre-2020 (no pandemic confound); bed rest comparator is healthy young people (potentially not matched for aging-related muscle changes); no assessment of cumulative PEM versus acute study effects.

18 Aschman et al. 2023 — Post-COVID Exercise Intolerance is Associated with Capillary Alterations and Immune Dysregulations in Skeletal Muscles

Full Citation:: Aschman T, Wyler E, Baum O, Hentschel A, Rust R, Legler F, Preusse C, et al. Post-COVID exercise intolerance is associated with capillary alterations and immune dysregulations in skeletal muscles. Acta Neuropathologica Communications. 2023;11(1):193. (Aschman et al. 2023)
DOI:: https://doi.org/10.1186/s40478-023-01662-2
PMID:: 38066589
PMCID:: PMC10704838
Study Design:: Cross-sectional; vastus lateralis muscle biopsy + serum proteomics + clinical phenotyping
Sample Size:: n=11 post-COVID syndrome (PCS) patients vs 2 independent historical control cohorts
Key Findings::
- Fewer capillaries and thicker capillary basement membranes in PCS patients vs controls
- Increased CD169+ macrophages in muscle tissue
- SARS-CoV-2 RNA not detected in muscle tissue — argues against viral persistence in muscle
- Complement system related proteins elevated in serum, matching transcriptomic signature in muscle
- Patients had handgrip weakness, prolonged post-exertional malaise, elevated fatigue scores
Conclusion:: Immune-mediated structural changes of the skeletal muscle microvasculature explain exercise-dependent fatigue and muscle pain in post-COVID syndrome. Capillary basement membrane thickening is a key finding — not attributable to viral persistence in muscle. Findings reminiscent of ME/CFS, suggesting shared final common pathway.
Certainty Assessment::
- Quality: High (peer-reviewed in Acta Neuropathologica Communications; IHC + EM + proteomics + transcriptomics)
- Sample: Low (n=11 PCS, small for mechanistic study)
- Replication: Partially replicated by Charlton 2025 (capillary BM thickening, more comprehensive n=50 patient biopsies); Agergaard 2023 (capillary injury in Long COVID n=18)
- Cohort: Charité — Universitätsmedizin Berlin, Germany; recruitment ~2021–2022; PIs: Stenzel, Dengler, Heppner, Scheibenbogen
- Score: 0.65
Limitations:: Small n (11 PCS); no ME/CFS biopsy arm — comparison to ME/CFS is by literature; historical controls (not contemporaneous); no longitudinal follow-up; CD169+ macrophage finding not replicated yet.

19 Joseph et al. 2022 — Neurovascular Dysregulation and Acute Exercise Intolerance in ME/CFS: A Randomized Trial of Pyridostigmine

Full Citation:: Joseph P, Pari R, Miller S, Warren A, Stovall MC, Squires J, Chang C-J, Xiao W, Waxman AB, Systrom DM. Neurovascular dysregulation and acute exercise intolerance in myalgic encephalomyelitis/chronic fatigue syndrome: a randomized, placebo-controlled trial of pyridostigmine. Chest. 2022;162(5):1109–1119. (Joseph et al. 2022)
DOI:: https://doi.org/10.1016/j.chest.2022.04.146
Study Design:: Randomized, placebo-controlled, double-blind crossover trial with invasive CPET
Sample Size:: n=45 ME/CFS (Fukuda criteria) + n=18 healthy controls
Key Findings::
- ME/CFS patients had significantly reduced peak systemic O₂ extraction (0.69 vs 0.77, p \(<\) 0.001)
- Cardiac output and pulmonary gas exchange were normal — problem is peripheral extraction, not delivery
- Reduced O₂ delivery as proportion of O₂ demand
- Pyridostigmine (acetylcholinesterase inhibitor, 60mg) increased peak VO₂ by +0.9 mL/kg/min (p = 0.002)
Conclusion:: Impaired peripheral oxygen extraction, not cardiac deconditioning, is the primary limiter of exercise capacity in ME/CFS. Pyridostigmine response suggests neurovascular dysregulation at the capillary level.
Certainty Assessment::
- Quality: High (randomized crossover; Chest journal; invasive CPET gold-standard hemodynamics)
- Sample: Medium (n=45 ME/CFS, well-powered for primary endpoint)
- Replication: Partial — consistent with Franklin 2022 meta-analysis of 2-day CPET; consistent with Squires 2026 invasive CPET + biopsy
- Cohort: Brigham and Women’s Hospital, Boston, MA; recruitment ~2018–2021; PI: Systrom
- Score: 0.65
Limitations:: Fukuda criteria (not CCC/IOM — may include patients without PEM); single center; pyridostigmine effect size modest (+0.9 mL/kg/min); study assesses acute exercise response, not PEM.

20 Agergaard et al. 2023 — Myopathy as a Cause of Long COVID Fatigue

Full Citation:: Agergaard J, Khan BYA, Engell-Sørensen T, Schiøttz-Christensen B, Østergaard L, Hejbøl EK, Schrøder HD, Tankisi H. Myopathy as a cause of Long COVID fatigue: Evidence from quantitative and single fiber EMG and muscle histopathology. Clinical Neurophysiology. 2023;153:149–158. (Agergaard et al. 2023)
DOI:: https://doi.org/10.1016/j.clinph.2023.06.009
PMID:: 37482344
Study Design:: Cross-sectional muscle biopsy + qEMG/sfEMG; biceps brachii and anterior tibialis biopsies
Sample Size:: n=18 Long COVID patients with persistent fatigue
Key Findings::
- Reduced capillary-to-fiber ratio (p=0.009) on muscle biopsy
- Increased capillary basement membrane thickness
- Mitochondrial changes with myopathic EMG findings in 73% of patients
- Terminal nerve and motor endplate damage with acetylcholine receptor loss
- Capillary injury documented as separate from mitochondrial pathology
Conclusion:: Long COVID fatigue has a myopathic basis with evidence of capillary-level injury. Lower capillarization found in Long COVID muscle — partially contradicted by Charlton 2025 who found lower capillarization only in ME/CFS (not Long COVID), though both studies agree on capillary BM thickening.
Certainty Assessment::
- Quality: Medium-High (peer-reviewed; comprehensive EMG + histopathology)
- Sample: Low-Medium (n=18 Long COVID)
- Replication: Partially replicated — capillary BM thickening confirmed by Aschman 2023 and Charlton 2025; capillarization finding partially contradictory to Charlton 2025
- Cohort: Aarhus University Hospital, Denmark; recruitment ~2021–2022; PIs: Agergaard, Tankisi
- Score: 0.60
Limitations:: Small n (18); no healthy control biopsies — comparisons to literature standards; biceps and anterior tibialis rather than vastus lateralis (different fiber types); qEMG/sfEMG normative data from different labs.

21 Franklin & Graham 2022 — Repeated Maximal Exercise Tests of Peak Oxygen Consumption in ME/CFS: Systematic Review and Meta-Analysis

Full Citation:: Franklin JD, Graham M. Repeated maximal exercise tests of peak oxygen consumption in people with myalgic encephalomyelitis/chronic fatigue syndrome: a systematic review and meta-analysis. Fatigue: Biomedicine, Health & Behavior. 2022;10(4):142–160. (Franklin and Graham 2022)
DOI:: https://doi.org/10.1080/21641846.2022.2138339
Study Design:: Systematic review and meta-analysis; 7 databases; 5 included studies
Sample Size:: Pooled n across 5 2-day CPET studies (exact total varies by analysis)
Key Findings::
- Pooled mean decrease in peak work rate at test-retest: −8.55 W greater in ME/CFS vs controls (95% CI −15.38 to −1.72)
- Work rate at anaerobic threshold decreased by −21 W (95% CI −38 to −4)
- 78% probability a future study would show clinically significant AT work rate decline in ME/CFS
- VO₂peak at retest did not show statistically significant between-group difference (moderate heterogeneity)
Conclusion:: ME/CFS patients demonstrate a clinically significant and reproducible test-retest reduction in work rate at anaerobic threshold when compared to healthy controls. Consistent with capillary-level O₂ delivery impairment as an objective and reproducible finding.
Certainty Assessment::
- Quality: High (systematic review; random-effects meta-analysis; pre-registered; independent of any single research group)
- Sample: Medium-High (pooled across multiple independent studies)
- Replication: This is the replication — it synthesizes 5 independent CPET studies showing consistent findings
- Score: 0.70
Limitations:: Only 5 studies met inclusion criteria (limited CPET literature); moderate heterogeneity in VO₂peak findings; studies used primarily Fukuda criteria (not CCC/IOM); CPET may trigger PEM in severe patients, creating selection bias toward milder cases.

22 Squires et al. 2026 — Impaired Systemic Oxygen Extraction and Skeletal Muscle Mitochondrial Dysfunction in ME/CFS and Long COVID

Full Citation:: Squires J, Gustafson S, Li P, Pappano S, LeWine K, Xiao W, Waxman AB, Naviaux RK. Impaired Systemic Oxygen Extraction and Skeletal Muscle Mitochondrial Dysfunction in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Long COVID. American Journal of Respiratory and Critical Care Medicine. 2026. (A64-09 ATS 2026 Abstract). (Squires et al. 2026)
Study Design:: Cross-sectional; invasive CPET + frozen vastus lateralis biopsy for mitochondrial analysis
Sample Size:: n=35 ME/CFS + n=10 LC with poor systemic O₂ extraction (SOE) on iCPET
Key Findings::
- Median CS activity = 66% of reference, mtDNA = 55.5% of reference in patient groups
- Lower Complex II+III activity correlated with worse SOE (ρ = −0.33, p = 0.028)
- CS activity and mtDNA did NOT correlate with SOE (ρ = −0.10 and 0.35, both p>0.1)
- Compared with genetically confirmed OXPHOS disease, ME/CFS + LC had significantly lower CS activity (Cliff’s δ ≈ 0.95, p < 0.001)
- Peak VO₂ = 72.1±10.7% predicted despite cardiac output at 95.2±13.3% predicted
Conclusion:: Impaired systemic O₂ extraction in ME/CFS and Long COVID is associated with functional (Complex II+III activity), not quantitative (CS/mtDNA), mitochondrial defects. Functional rather than biomass mitochondrial deficits underlie impaired oxygen extraction phenotype.
Certainty Assessment::
- Quality: Low-Medium (conference abstract only; peer-reviewed abstract forum but not full publication yet)
- Sample: Medium (n=45 total patient biopsies)
- Replication: Not yet replicated; consistent with Joseph 2022 (O₂ extraction impairment) and Charlton 2025 (capillary-level O₂ delivery problem)
- Cohort: Brigham and Women’s Hospital / Harvard; Baylor Genetics Laboratory; PI: Naviaux (OMF-funded). Same cohort as Joseph 2022 likely — cohort overlap flagged.
- Score: 0.40
Limitations:: Abstract only — limited methodological detail; likely same patient cohort as Joseph 2022 (Systrom group); frozen biopsies only (no EM for ultrastructure); Complex I correlation did not reach significance (ρ = −0.28, p = 0.060), limiting the mechanistic conclusion to Complex II+III.

23 Wirth 2026 — Laxity Comes with Consequences: Connective Tissue Disorders and ME/CFS

Full Citation:: Wirth KJ. Laxity Comes with Consequences: Connective Tissue Disorders and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS). Preprints. 2026. (Wirth 2026)
DOI:: https://doi.org/10.20944/preprints202606.1858.v1
Study Design:: Hypothesis/review paper
Sample Size:: n/a (theoretical)
Key Findings::
- Examines mechanisms driving capillary basement membrane thickening in ME/CFS via connective tissue biology
- Proposes hyaluronan overproduction in connective tissue disorders contributes to capillary BM thickening via CD44/TSG-6 pathway
- Links impaired pericyte recruitment to microvascular pathology
- Integrates Aschman 2023 and Charlton 2025 findings into a connective-tissue framework
Conclusion:: Capillary basement membrane thickening in ME/CFS may be mechanistically linked to connective tissue laxity/hypermobility via excess hyaluronan production and impaired pericyte function. Provides a testable biochemical pathway for the microvascular findings observed by Aschman and Charlton.
Certainty Assessment::
- Quality: Low (preprint; hypothesis paper; single author)
- Sample: n/a
- Replication: None — novel hypothesis
- Score: 0.35
Limitations:: Preprint (not peer-reviewed); single author; all mechanisms extrapolated from general physiology literature; no direct ME/CFS biopsy evidence for TSG-6 or HA pathway involvement.

References

Agergaard, Jane, Benjamin Yamin Ali Khan, Thomas Engell-Sørensen, Berit Schiøttz-Christensen, Lars Østergaard, Eva Klara Hejbøl, Henrik Daa Schrøder, and Hatice Tankisi. 2023. “Myopathy as a Cause of Long COVID Fatigue: Evidence from Quantitative and Single Fiber EMG and Muscle Histopathology.” Clinical Neurophysiology 153: 149–58. https://doi.org/10.1016/j.clinph.2023.06.009.

Aschman, Tom, Emanuel Wyler, Oliver Baum, Andreas Hentschel, Rebekka Rust, Franziska Legler, Corinna Preusse, et al. 2023. “Post-COVID Exercise Intolerance Is Associated with Capillary Alterations and Immune Dysregulations in Skeletal Muscles.” Acta Neuropathologica Communications 11 (1): 193. https://doi.org/10.1186/s40478-023-01662-2.

Charlton, Braeden T., Anouk Slaghekke, Brent Appelman, Morne Eggelbusch, Jelle Y. Huijts, Wendy Noort, Paul W. Hendrickse, et al. 2025. “Skeletal Muscle Properties in Long COVID and ME/CFS Differ from Those Induced by Bed Rest.” medRxiv. https://doi.org/10.1101/2025.06.23.25330153.

Falaguera-Vera, Francisco J., Maria Garcia-Escudero, Javier Bonastre-Férez, Mario Zacarés, and Elisa Oltra. 2020. “Pressure Point Thresholds and ME/CFS Comorbidity as Indicators of Patient’s Response to Manual Physiotherapy in Fibromyalgia.” International Journal of Environmental Research and Public Health 17 (21): 8044. https://doi.org/10.3390/ijerph17218044.

Franklin, John D., and Max Graham. 2022. “Repeated Maximal Exercise Tests of Peak Oxygen Consumption in People with Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: A Systematic Review and Meta-Analysis.” Fatigue: Biomedicine, Health & Behavior 10 (4): 142–60. https://doi.org/10.1080/21641846.2022.2138339.

Friston, Karl. 2023. “Computational Psychiatry: From Synapses to Sentience.” Molecular Psychiatry 28 (1): 256–68. https://doi.org/10.1038/s41380-022-01743-z.

Hu, Michele, Casper Skjærbæk, and Per Borghammer. 2024. “Approaches to Early Parkinson’s Disease Subtyping.” Journal of Parkinson’s Disease 14 (s2): S297–306. https://doi.org/10.3233/JPD-230419.

Joseph, Phillip, Reena Pari, Samuel Miller, Ann Warren, Mary C. Stovall, Johanna Squires, Chen-yen J. Chang, Wenzhen Xiao, Aaron B. Waxman, and David M. Systrom. 2022. “Neurovascular Dysregulation and Acute Exercise Intolerance in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: A Randomized, Placebo-Controlled Trial of Pyridostigmine.” Chest 162 (5): 1109–19. https://doi.org/10.1016/j.chest.2022.04.146.

Kim, Lisa, Claudia Kedor, Frank Buttgereit, Harald Heidecke, Deike Schaumburg, and Carmen Scheibenbogen. 2023. “Characterizing Sjögren-Associated Fatigue: A Distinct Phenotype from ME/CFS.” Journal of Clinical Medicine 12 (15): 4994. https://doi.org/10.3390/jcm12154994.

Laragh, John H., Brent Lamport, Jean Sealey, and Michael H. Alderman. 1988. “Diagnosis Ex Juvantibus. Individual Response Patterns to Drugs Reveal Hypertension Mechanisms and Simplify Treatment.” Hypertension 12 (3): 223–26. https://doi.org/10.1161/01.hyp.12.3.223.

Monini, Simonetta, A. Di Stadio, Angelo Vestri, and Maurizio Barbara. 2006. “Silent Reflux: Ex Juvantibus Criteria for Diagnosis and Treatment of Laryngeal Disorders.” Acta Otorhinolaryngologica 126 (8): 866–71. https://doi.org/10.1080/00016480500504242.

Newberry, Sydne J., John FitzGerald, Margaret A. Maglione, Claire E. O’Hanlon, Dan Han, Marika Booth, Aneesa Motala, et al. 2016. “Diagnosis of Gout.” 15(16)-EHC026-EF. Agency for Healthcare Research; Quality (AHRQ).

Ravichandran, Sharan, Nandan Sudarsanam, Balaraman Ravindran, and Konstantinos V. Katsikopoulos. 2024. “Active Learning with Human Heuristics: An Algorithm Robust to Labeling Bias.” Frontiers in Artificial Intelligence 7: 1491932. https://doi.org/10.3389/frai.2024.1491932.

Samuel, Joyce P., Cynthia S. Bell, Joshua A. Samuels, Charanya Rajan, Ashley K. Walton, Charles Green, and Jon E. Tyson. 2023. “N-of-1 Trials Vs. Usual Care in Children with Hypertension: A Pilot Randomized Clinical Trial.” American Journal of Hypertension 36 (2): 126–32. https://doi.org/10.1093/ajh/hpac117.

Samuel, Joyce P., Jon E. Tyson, Charles Green, Cynthia S. Bell, Claudia Pedroza, Donald Molony, and Joshua Samuels. 2019. “Treating Hypertension in Children with n-of-1 Trials.” Pediatrics 143 (4): e20181818. https://doi.org/10.1542/peds.2018-1818.

Schaaf, Jessica V., Laura Weidinger, Lucas Molleman, and Wouter van den Bos. 2024. “Test-Retest Reliability of Reinforcement Learning Parameters.” Behavior Research Methods 56 (5): 4582–99. https://doi.org/10.3758/s13428-023-02203-4.

Scheibenbogen, Carmen, Madlen Loebel, Helma Freitag, Anne Krueger, Sandra Bauer, Michaela Antelmann, Wolfram Doehner, et al. 2018. “Immunoadsorption to Remove Β2 Adrenergic Receptor Antibodies in Chronic Fatigue Syndrome.” PLoS One 13 (3): e0193672. https://doi.org/10.1371/journal.pone.0193672.

Schwartenbeck, Philipp, and Karl Friston. 2016. “Computational Phenotyping in Psychiatry: A Worked Example.” eNeuro 3 (4): ENEURO.0049–16.2016. https://doi.org/10.1523/ENEURO.0049-16.2016.

Squires, Johanna, Sarah Gustafson, Peter Li, Sarah Pappano, Karen LeWine, Wenzhen Xiao, Aaron B. Waxman, and Robert K. Naviaux. 2026. “Impaired Systemic Oxygen Extraction and Skeletal Muscle Mitochondrial Dysfunction in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Long COVID.” American Journal of Respiratory and Critical Care Medicine.

Strauss, Maximilian J., Thomas Niederkrotenthaler, Stefan Thurner, Alexandra Kautzky-Willer, and Peter Klimek. 2021. “Data-Driven Identification of Complex Disease Phenotypes.” Journal of the Royal Society Interface 18 (180): 20201040. https://doi.org/10.1098/rsif.2020.1040.

Wirth, Klaus J. 2026. “Laxity Comes with Consequences: Connective Tissue Disorders and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS).” Preprints. https://doi.org/10.20944/preprints202606.1858.v1.

Wirth, Klaus J., and Carmen Scheibenbogen. 2021. “Pathophysiology of Skeletal Muscle Disturbances in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS).” Journal of Translational Medicine 19 (1): 162. https://doi.org/10.1186/s12967-021-02833-2.

Zhang, Qing, Yi-Zhuo Zhao, Hong-Hao Ma, Dong Wang, Lei Cui, Wei-Jing Li, Ang Wei, et al. 2022. “A Study of Ruxolitinib Response-Based Stratified Treatment for Pediatric Hemophagocytic Lymphohistiocytosis.” Blood 139 (24): 3493–3504. https://doi.org/10.1182/blood.2021014860.