Eckey 2025 Mechanism Re-analysis: Projecting Treatment Rankings onto Mechanisms
This stream assembles the literature that makes or breaks the proposal to re-analyse the Eckey et al. 2025 ME/CFS + long-COVID treatment survey (n=3,925; >150 treatments) as a mechanism-ranking pilot: map each surveyed treatment to its known molecular targets, then ask whether the treatment-benefit ranking, projected onto the hypothesis space, produces a subgroup-specific mechanism ranking that is internally stable and concordant with independent evidence. The driving dataset is already in the corpus ((Eckey et al. 2025)); no direct study of this re-analysis exists. The entries below are the separable methodological links — drug-to-pathway enrichment, data-driven subgroup identification, placebo/expectation network structure, popularity/confounding channels, patient-reported-outcome validity, and drug-to-target mappings for high-ranking treatment classes.
1 Eckey et al. 2025 — Patient-Reported Treatment Outcomes in ME/CFS and Long COVID (dataset feasibility)
Full Citation:: Eckey M, Li P, Morrison B, Bergquist J, Davis RW, Xiao W. Patient-reported treatment outcomes in ME/CFS and long COVID. Proceedings of the National Academy of Sciences. 2025;122(28):e2426874122. (Eckey et al. 2025) DOI:: 10.1073/pnas.2426874122 PMID:: 40627388 Study Design:: Large cross-sectional patient-reported treatment survey (TREATME) Sample Size:: n=3,925 (2,125 ME/CFS + 1,800 long COVID); 440+ questions on 150+ interventions Key Findings::
- Net Assessment Score (NAS) assigned to every treatment with ≥20 responses; oral non-liposomal vitamin C used as the reference comparator in the absence of placebo, with FDR-adjusted p-values.
- Four reproducible symptom/comorbidity clusters: multisystemic, POTS-dominant, cognitive/sleep + pain, and milder; treatment responses differ by cluster (e.g., IgG 73.3% in multisystemic, ADHD stimulants 62.1% in the cognitive cluster but non-significant in the multisystemic cluster).
- Patient capacity (severity) is the strongest predictor of treatment effectiveness, stronger than ME/CFS-vs-long-COVID diagnosis, sex, duration, or age; ME/CFS and long-COVID NASs correlate at R²=0.68.
- *Data availability:* the survey instrument and detailed results are publicly available as Datasets S1–S3 (XLSX/XLSX/PDF) plus SI Appendix Tables S1–S6 — i.e., the raw stratum-level data needed for an independent re-analysis exist.
- Authors explicitly name the confounders the re-analysis must address: self-selection, nonresponse, recall bias, placebo/expectation, and polypharmacy (patients trial multiple treatments at once).Conclusion: The single most direct real-world precedent and the dataset itself. Its public Supplementary Data make the mechanism-ranking re-analysis feasible without new recruitment; its own stated confounders define the null the re-analysis must beat (popularity/expectation explains the ranking as well as mechanism). Limitations: Self-selected, unblinded, retrospective self-report; many treatments have fewer than 100 responses (unreliable NAS); no objective verification of diagnosis or response; reference comparator is a weak placebo proxy. Certainty Assessment: 0.70 raw × 1.00 (ME/CFS cohort) = discounted 0.70 (driving dataset)
2 Napolitano et al. 2016 — Drug-Set Enrichment Analysis for Drug Mode of Action
Full Citation:: Napolitano F, Sirci F, Carrella D, di Bernardo D. Drug-set enrichment analysis: a novel tool to investigate drug mode of action. Bioinformatics. 2016;32(2):235–241. (Napolitano et al. 2016) DOI:: 10.1093/bioinformatics/btv536 PMID:: 26415724 Study Design:: Computational method development with connectivity-map validation Key Findings::
- Groups drugs by shared mechanism of action / target and tests whether a query signature is enriched across the drug set, recovering mode-of-action at the mechanism level.
- Demonstrated on Connectivity Map perturbation data to identify drug classes with shared mechanisms.Conclusion: The core methodological precedent for projecting a treatment list onto target/mechanism classes — the exact operation the Eckey re-analysis needs. It shows mechanism-level aggregation is achievable but depends on reliable drug-to-target assignment and on a reference signature to score against. Limitations: Computational; requires accurate drug-target annotation; validated in cancer cell lines, not patient-reported benefit data; enrichment does not establish causal target mediation. Certainty Assessment: 0.65 raw × 0.30 (in silico) = discounted ~0.20 (supporting method)
3 Napolitano et al. 2019 — gep2pep: Pathway-Based Expression Profiles
Full Citation:: Napolitano F, Carrella D, Gao X, di Bernardo D. gep2pep: a Bioconductor package for the creation and analysis of pathway-based expression profiles. Bioinformatics. 2019;36(6):1944–1945. (Napolitano et al. 2019) DOI:: 10.1093/bioinformatics/btz803 PMID:: 31647521 Study Design:: Software/method note Key Findings::
- Converts drug-perturbation gene-expression signatures into pathway-level profiles.
- Enables mechanism/pathway-resolution aggregation of drug sets, extending DSEA from drug sets to pathways.Conclusion: Supplies the pathway-resolution implementation for the treatment-to-pathway projection. Useful only if the re-analysis adopts a pathway (not merely drug-class) hypothesis space. Limitations: Depends on curated pathway databases and expression data; no patient-reported inputs; software note with limited independent benchmarking. Certainty Assessment: 0.55 raw × 0.30 (in silico) = discounted ~0.17 (supporting method)
4 Lipkovich et al. 2017 — Data-Driven Subgroup Identification in Clinical Trials
Full Citation:: Lipkovich I, Dmitrienko A, D’Agostino RB. Tutorial in biostatistics: data-driven subgroup identification and analysis in clinical trials. Statistics in Medicine. 2017;36(1):136–196. (Lipkovich, Dmitrienko, and D’Agostino 2017) DOI:: 10.1002/sim.7064 PMID:: 27488683 Study Design:: Statistical methodology tutorial (with worked examples) Key Findings::
- Reviews virtual-twins, recursive-partitioning, and model-based recursive-partitioning methods for finding subgroups with differential treatment effect.
- Emphasises multiplicity control and validation (training/test splits, cross-validation) because unconstrained subgroup search overfits.
- A null overall effect can conceal a real, subgroup-specific effect — and vice versa.Conclusion: The canonical statistical machinery for producing subgroup-specific mechanism rankings from outcome data and for guarding against the false subgroup that a large survey will inevitably contain. Directly supports the re-analysis’s split-half replication requirement. Limitations: Methodological, not empirical; the worked examples are not ME/CFS; subgroup stability under data-driven search remains the principal risk the tutorial documents rather than solves. Certainty Assessment: 0.70 raw × 0.75 (general/methodological) = discounted ~0.53 (supporting method)
5 Wang et al. 2017 — Network Analysis of the Genomic Basis of the Placebo Effect
Full Citation:: Wang R-S, Hall KT, Giulianini F, Passow D, Kaptchuk TJ, Loscalzo J. Network analysis of the genomic basis of the placebo effect. JCI Insight. 2017;2(11):e93911. (Wang et al. 2017) DOI:: 10.1172/jci.insight.93911 PMID:: 28570268 Study Design:: Computational network analysis of placebo-response gene sets Key Findings::
- The placebo response maps to a structured network of genes and pathways distinct from the drug target.
- Expectation and context engage measurable, non-random molecular programs rather than unstructured noise.Conclusion: Establishes that expectation can produce a biologically structured signal — so an apparent mechanism ranking can arise from placebo/expectation without the drug’s target. This is the strongest form of the re-analysis’s alternative explanation and must be modelled explicitly. Limitations: Computational; placebo-response genomic data from heterogeneous trials; no direct ME/CFS application. Certainty Assessment: 0.55 raw × 0.30 (in silico) = discounted ~0.17 (competing explanation)
6 Simmering et al. 2014 — Web Search Query Volume as a Measure of Pharmaceutical Utilization
Full Citation:: Simmering JE, Polgreen LA, Polgreen PM. Web search query volume as a measure of pharmaceutical utilization and changes in prescribing patterns. Research in Social and Administrative Pharmacy. 2014;10(6):896–903. (Simmering, Polgreen, and Polgreen 2014) DOI:: 10.1016/j.sapharm.2014.01.003 PMID:: 24603135 Study Design:: Observational infodemiology study Key Findings::
- Internet search-query volume tracks pharmaceutical utilization and prescribing-pattern changes.
- Validates search volume as an operational proxy for treatment popularity/community discussion.Conclusion: Supplies the measurable popularity covariate the re-analysis must include. If the mechanism ranking tracks search popularity as well as target assignment, the pharmacological interpretation fails the proposal’s stated falsification condition. Limitations: General population; specific drug classes; search volume reflects interest, not necessarily patient uptake; not ME/CFS-specific. Certainty Assessment: 0.55 raw × 0.75 (general/methodological) = discounted ~0.41 (confounder measure)
8 Yom-Tov & Gabrilovich 2013 — Postmarket Drug Surveillance from Web Search Queries
Full Citation:: Yom-Tov E, Gabrilovich E. Postmarket drug surveillance without trial costs: discovery of adverse drug reactions through large-scale analysis of web search queries. Journal of Medical Internet Research. 2013;15(6):e124. (Yom-Tov and Gabrilovich 2013) DOI:: 10.2196/jmir.2614 PMID:: 23778053 Study Design:: Large-scale observational infodemiology study Key Findings::
- Web search queries carry pharmacologically interpretable signal, detecting adverse drug reactions that complement formal reporting.
- Web data can reflect genuine drug effects, not only interest.Conclusion: Two-edged precedent: search/discussion volume is both a confounder to control and a partially valid pharmacological channel. The re-analysis should not treat popularity as pure noise, but as a covariate that may itself carry signal. Limitations: General population; ADR detection (harm), not benefit ranking; search volume sensitive to media events; not ME/CFS-specific. Certainty Assessment: 0.50 raw × 0.75 (general/methodological) = discounted ~0.38 (confounder evidence)
9 Rekeland et al. 2022 — Activity Monitoring and Patient-Reported Outcome Measures in ME/CFS
Full Citation:: Rekeland IG, Sørland K, Bruland O, Risa K, Alme K, Dahl O, Tronstad KJ, Mella O, Fluge Ø. Activity monitoring and patient-reported outcome measures in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome patients. PLoS One. 2022;17(9):e0274472. (Rekeland et al. 2022) DOI:: 10.1371/journal.pone.0274472 PMID:: 36121803 Study Design:: Observational measurement-validation study Key Findings::
- Compares device-measured activity with patient-reported outcome measures in ME/CFS patients.
- Quantifies the relationship and divergence between self-report and objective activity.Conclusion: Directly governs the measurement-validity limitation of any patient-reported treatment ranking: self-reported benefit and objectively measured function are related but not interchangeable, so a benefit ranking is a ranking of perceived benefit, not of measured improvement. Limitations: ME/CFS cohort, but focus on activity measurement rather than treatment ranking; observational; single centre. Certainty Assessment: 0.55 raw × 1.00 (ME/CFS cohort) = discounted 0.55 (measurement validity)
10 Haywood et al. 2014 — Systematic Review of PROMs in Paediatric CFS/ME
Full Citation:: Haywood KL, Collin SM, Crawley E. Assessing severity of illness and outcomes of treatment in children with Chronic Fatigue Syndrome/Myalgic Encephalomyelitis (CFS/ME): a systematic review of patient-reported outcome measures (PROMs). Child: Care, Health and Development. 2014;40(6):806–824. (Haywood, Collin, and Crawley 2014) DOI:: 10.1111/cch.12135 PMID:: 24661148 Study Design:: Systematic review of measurement instruments Key Findings::
- Catalogues which PROMs in paediatric CFS/ME have evidence of reliability, validity, and responsiveness to change.
- PROM selection materially affects measured treatment response.Conclusion: Documents that the instrument determines the measured ranking. A re-analysis of self-reported benefit inherits the psychometric limits of the survey instrument; cross-survey concordance must account for instrument differences. Limitations: Paediatric population; instrument-methodological, not a treatment study; older review. Certainty Assessment: 0.55 raw × 0.80 (comorbid/population-specific) = discounted ~0.44 (measurement validity)
11 Dorczok et al. 2026 — Use and Perceived Helpfulness of Interventions in ME/CFS and Depression
Full Citation:: Dorczok MC, Mossaheb N, Mittmann G, Thomas MF, Bartova L, Schrank B, Steiner-Hofbauer V. Use and Perceived Helpfulness of Different Intervention Strategies in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Depression. Journal of Clinical Medicine. 2026;15(2):849. (Dorczok et al. 2026) DOI:: 10.3390/jcm15020849 PMID:: 41598786 Study Design:: Cross-sectional patient survey with comparison group Key Findings::
- Independent patient survey ranking self-management and intervention strategies by perceived helpfulness in ME/CFS, with a depression comparison group.
- Provides a second patient-reported benefit-ranking dataset distinct in instrument and population from Eckey 2025.Conclusion: Enables an external concordance check: if the Eckey mechanism ranking replicates onto a second survey’s ranking, that strengthens the result beyond split-half internal consistency. If the two rankings diverge, the mechanism projection is instrument- or cohort-dependent. Limitations: Small relative to Eckey; self-report; mixed ME/CFS/depression framing; does not rank molecular mechanisms. Certainty Assessment: 0.45 raw × 1.00 (ME/CFS cohort) = discounted 0.45 (external concordance)
12 Löhn & Wirth 2024 — TRPM3 and the Therapeutic Effect of Low-Dose Naltrexone in ME/CFS
Full Citation:: Löhn M, Wirth KJ. Potential pathophysiological role of the ion channel TRPM3 in myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) and the therapeutic effect of low-dose naltrexone. Journal of Translational Medicine. 2024;22(1):630. (Löhn and Wirth 2024) DOI:: 10.1186/s12967-024-05412-3 PMID:: 38970055 Study Design:: Mechanistic hypothesis review Key Findings::
- Proposes TRPM3 channel dysregulation as a contributor to ME/CFS pathophysiology.
- Attributes part of low-dose naltrexone's therapeutic effect to TRPM3 modulation — an explicit drug-to-molecular-target mapping inside ME/CFS.Conclusion: Supplies exactly the treatment-to-mechanism assignment the re-analysis performs for the LDN node, and illustrates its uncertainty: LDN also acts on TLR4/glial pathways, so a single drug maps to multiple mechanisms and the projection must distribute LDN’s benefit across targets (or justify a primary assignment). Limitations: Hypothesis/review, not a trial; TRPM3 role in ME/CFS not independently confirmed; LDN efficacy itself rests on mixed trial evidence. Certainty Assessment: 0.45 raw × 1.00 (ME/CFS cohort) = discounted 0.45 (drug-to-target mapping)
13 Mar & Raj 2020 — POTS: Mechanisms and New Therapies
Full Citation:: Mar PL, Raj SR. Postural Orthostatic Tachycardia Syndrome: Mechanisms and New Therapies. Annual Review of Medicine. 2020;71:235–248. (Mar and Raj 2020) DOI:: 10.1146/annurev-med-041818-011630 PMID:: 31412221 Study Design:: Narrative review of pathophysiology and therapy Key Findings::
- Reviews mechanism-based POTS therapies — beta-blockers, ivabradine, midodrine, volume expansion — each linked to its autonomic target.
- Distinguishes therapies by mechanism (heart-rate control vs vasoconstriction vs volume) rather than by symptom alone.Conclusion: Provides the drug-to-target mapping for the autonomic-modulator class that ranks highly in the Eckey survey. Grounds treating beta-blocker and ivabradine responses as probes of distinct autonomic mechanisms — the mechanistic discrimination the ranking is meant to support. Limitations: Narrative review; POTS (not all ME/CFS); evidence base for several drugs remains weak; mapping is pharmacological inference. Certainty Assessment: 0.60 raw × 0.80 (comorbid/population-specific) = discounted 0.48 (drug-to-target mapping)
14 Hasan et al. 2020 — Treatment of POTS With Medication: A Systematic Review
Full Citation:: Hasan B, Almasri J, Marwa B, Klaas KM, Fischer PR. Treatment of Postural Orthostatic Tachycardia Syndrome With Medication: A Systematic Review. Journal of Child Neurology. 2020;35(14):1004–1016. (Hasan et al. 2020) DOI:: 10.1177/0883073820948679 PMID:: 32838632 Study Design:: Systematic review of pharmacologic POTS treatment Key Findings::
- Grades the evidence for each POTS drug class.
- The evidence base is dominated by small, largely uncontrolled studies.Conclusion: Supplies the external-evidence comparator for the autonomic-modulator mechanism rank and, equally, documents how weak that external evidence is — a limit on how strongly the Eckey projection can be validated against independent data. Limitations: Paediatric population; weak primary evidence; no ME/CFS-specific ranking; no mechanism-level synthesis. Certainty Assessment: 0.55 raw × 0.80 (comorbid/population-specific) = discounted 0.44 (external evidence)