Thermoregulatory Drug Candidates
Certainty: 0.25. Bosentan (dual ET_A/ET_B antagonist, approved for pulmonary arterial hypertension) reduces vascular tone by blocking endothelin-1 signaling. If ET-1-mediated tonic vasoconstriction contributes to thermoregulatory failure (ch10:Tonic Cutaneous Vasoconstriction Bias Explains Dual Heat+Cold Intolerance, Cambras et al. (Cambras et al. 2023)), bosentan could restore vasomotor flexibility and improve heat tolerance. However, bosentan carries hepatotoxicity risk (monthly liver enzyme monitoring required) and is teratogenic (contraindicated in pregnancy), making risk-benefit for non-life-threatening thermoregulatory symptoms highly unfavorable outside clinical trials. Falsifiable: bosentan vs placebo x 8 weeks in ME/CFS patients with elevated ET-1 (>2.5 pg/mL) improves skin temperature rhythm amplitude by >30% and heat tolerance by Δ2°C, with no benefit in ET-1-normal subgroup. No ME/CFS data; ET-1-thermoregulation link is correlational only (Cambras et al. 2023).
Certainty: 0.25. Arimoclomol (HSP70 co-inducer, approved for Niemann-Pick disease type C in the EU) amplifies HSP70 expression during cellular stress without requiring thermal exposure. If WBH improves mitochondrial function in ME/CFS via HSP70 induction (Hochecker et al. 2025), arimoclomol could provide the same benefit without sauna access or thermal intolerance risk. The drug amplifies endogenous stress responses rather than creating new ones, potentially lowering the risk of triggering PEM compared to thermal therapy. Falsifiable: arimoclomol vs placebo x 12 weeks increases PBMC HSP70 protein >2-fold and improves ISR biomarkers (p-eIF2α, ATF4) by >30%. No ME/CFS data; approved only for Niemann-Pick type C (orphan drug).
Certainty: 0.25. Hypothesis H3 (brainstorm) proposes that mitochondrial ROS overload prevents adequate HSP70 induction in ME/CFS, creating a self-reinforcing loop: ROS → proteotoxic stress → HSP70 sequestration → worse mitochondrial quality → more ROS. MitoQ (mitochondria-targeted ubiquinone) concentrates at the inner mitochondrial membrane and reduces superoxide production at source. By lowering mitochondrial ROS burden, MitoQ could restore HSP70 inducibility and break the ROS-HSP70 feedback loop — independently of any thermal stimulus. Falsifiable: MitoQ vs placebo x 8 weeks reduces PBMC mitochondrial ROS (MitoSOX) by >30% and restores thermal-HSP70 inducibility to >80% of healthy control levels. No ME/CFS MitoQ data; human safety established in Parkinson’s and NAFLD trials.
Certainty: 0.20. Pregnenolone sulfate is an endogenous neurosteroid and TRPM3 agonist. If TRPM3 channelopathy impairs calcium signaling in ME/CFS immune cells, and if TRPM3 agonism functionally inhibits TRPV1-mediated heat hyperresponsiveness (brainstorm Hypothesis H8), pregnanolone sulfate could simultaneously restore TRPM3-dependent calcium flux and reduce heat-triggered MCAS symptoms. The dual mechanism — calcium rescue + TRPV1 suppression — makes it mechanistically distinct from purely thermal TRPM3 activation. Falsifiable: pregnanolone sulfate normalizes TRPM3 calcium influx in ME/CFS PBMCs and reduces heat-triggered CGRP release by 40% within 4 weeks. No ME/CFS data; TRPM3 agonism is established in vitro only; TRPM3-TRPV1 antagonism demonstrated in sensory neurons, not immune cells.
Certainty: 0.25. Clonidine (α2-agonist, 0.1 mg/24h patch) reduces central sympathetic outflow, which could counteract the tonic vasoconstriction bias proposed to underlie dual heat+cold intolerance (ch10:Tonic Cutaneous Vasoconstriction Bias Explains Dual Heat+Cold Intolerance). Reduced sympathetic tone would permit cutaneous vasodilation during heat stress and allow perfusion recruitment for thermogenesis during cold stress — restoring vasomotor dynamic range from the neural side rather than the vascular (ET-1) side. However, clonidine can worsen orthostatic hypotension in POTS patients, making it contraindicated in the dysautonomic ME/CFS majority. Further, Sulheim et al. (2014) RCT of clonidine in adolescent CFS (n=120) found no clinical improvement (Sulheim et al. 2014), and the central NE deficiency documented in ME/CFS (Aregawi 2026) suggests clonidine would further suppress already-low central noradrenergic output (Aregawi et al. 2026). Falsifiable: clonidine patch over 4 weeks increases baseline cutaneous perfusion by 25% and restores vasomotor responses to both heat and cold challenge; orthostatic symptoms do not worsen in patients without pre-existing hypotension. No ME/CFS clonidine thermoregulation data; safety concern in dysautonomia is well-documented; the null RCT in CFS weakens the mechanistic rationale.
LDN (1–4.5 mg nightly) modulates microglial activation via toll-like receptor 4 antagonism and has documented effects on TRPM3 calcium channel function in ME/CFS NK cells (Cabanas et al. 2021). LDN has four mechanisms with non-overlapping dose optima — TLR4/Nrf2 hormetic priming (0.5–1.5 mg), opioid compensatory upregulation (1.5–3.0 mg), TRPM3 restoration (3.0–4.5 mg), and orexin disinhibition — the thermoregulatory benefit may itself show inverted-U dose-dependence consistent with the broader hormetic framework (hormetic dose response). If central thermoregulatory centers (hypothalamus, brainstem) are under chronic microglial-mediated inflammatory tone, LDN could improve thermoregulation indirectly by reducing neuroinflammation, independently of its TRPM3 effects. The dual mechanism — peripheral TRPM3 normalization + central glial modulation — would predict thermoregulatory benefit as a secondary outcome. Falsifiable: LDN over 12 weeks improves TRPM3 calcium influx by 25% and reduces heat intolerance symptoms by 30%; improvement correlates with baseline inflammatory markers. No ME/CFS LDN thermoregulation data; LDN evidence in ME/CFS is from small open-label studies.
The FINAL trial (n=99 women with fibromyalgia, LDN 6 mg vs placebo for 12 weeks) found no significant primary pain benefit (Due Bruun et al. 2024). A responder re-analysis of the six secondary non-pain outcomes (tenderness, fatigue, sleep disturbances, depression, memory problems, stiffness) found no significant difference in 30% responder rates on any outcome (Nielsen, Vaegter, and Due Bruun 2026). The highest risk ratio was memory (RR 1.67, 95% CI 0.82–2.95), which does not reach significance — the “may improve memory problems” signal from the primary paper does not survive responder-category analysis. An independent 12-month RCT (INNOVA) also found no pain benefit over placebo (Rodríguez-Freire et al. 2026), and a meta-analysis found no between-group superiority vs placebo (Ologunowa et al. 2025). The re-analysis group itself concluded LDN efficacy in fibromyalgia is lower than previously reported (Due Bruun et al. 2026).
This does not question LDN’s favorable safety profile — a meta-analysis found no excess serious adverse events vs placebo (RR 0.84) (Bolton et al. 2019) — but the clinical-benefit case for fibromyalgia secondary symptoms is not supported by the responder evidence. The mechanistic rationales (TLR4 antagonism, opioid-growth-factor rebound) are biologically plausible but did not translate into measurable symptom benefit at clinical doses in fibromyalgia. The responder analysis itself carries methodological caveats: n=99 is underpowered for six secondary outcomes, the 30% responder threshold is arbitrary, and multiple testing inflates false-positive risk; the memory CI (RR 1.67, 0.82–2.95) spans null-to-large, so it neither establishes nor excludes a real cognitive effect. Severity applicability: fibromyalgia population, not stratified by severity; ME/CFS applicability unknown — no direct ME/CFS responder data in this analysis.
Consequence: Patients and clinicians should not expect low-dose naltrexone to reliably improve fibromyalgia symptoms such as fatigue, sleep, or memory based on the current trial evidence — the promising “memory improvement” signal from one trial did not replicate as a responder benefit. Falsifiable: a pooled responder meta-analysis of LDN in fibromyalgia showing a significant advantage on any secondary outcome would overturn this null framing.
Certainty: 0.30. The null fibromyalgia responder results may be mechanistically expected rather than contradictory. (+)-naltrexone is a weak TLR4 antagonist; the optimized derivative CIAC101 required ~6200× potency gain to reach nanomolar TLR4 antagonism (Gao et al. 2025). Clinical LDN is racemic (both enantiomers) at 4.5–6 mg. If the TLR4-blocking species needs concentrations orders of magnitude above what LDN achieves at these doses, then the FM null reflects a dose-to-target translation failure — the TLR4 rationale was never actually tested by LDN at clinical dose. This is consistent with the alternative OGFr/enkephalin-rebound mechanism (Zagon and McLaughlin 2018), which may require longer duration or a fatigue-specific subtype not captured by the FM responder categories (Nielsen, Vaegter, and Due Bruun 2026). Note this occupancy question stands in tension with the hormetic-window framing in hormetic dose response, which assumes LDN engages TLR4 at low doses to trigger Nrf2 priming — if the TLR4 target is never engaged at clinical dose, that specific arm of the hormetic model is conditional on a dose-occupancy measurement. Severity applicability: fibromyalgia, not severity-stratified; ME/CFS applicability unknown.
Consequence: The fibromyalgia null result does not disprove the TLR4/microglial rationale for naltrexone-class drugs — it may mean the dose never reached the target, which changes the research direction from “does the mechanism work” to “was it ever tested at an engaging dose.” Falsifiable: a dose-occupancy study showing clinical-dose LDN achieves below 10% TLR4 occupancy (no downstream NF-\(\kappa\)B suppression) in patient PBMCs would support this below-threshold reading; measurable TLR4 antagonism at these doses would falsify it.
Two meta-analyses of LDN in fibromyalgia reach opposite conclusions on pain. A 2025 meta-analysis found no between-group superiority of LDN over placebo (Ologunowa et al. 2025), while a 2024 meta-analysis reported significant pain reduction (MD −0.86) and higher pressure pain threshold (Vatvani et al. 2024). The discrepancy reflects differing trial sets and analytic methods, and the positive analysis carries an erratum. Because the certainty difference between the two (0.56 vs 0.52 discounted) is within measurement uncertainty, the paper presents both without weighting toward either. Fibromyalgia responder data (null) leans toward the null meta-analysis (Nielsen, Vaegter, and Due Bruun 2026), but the pain-specific question remains open. Severity applicability: fibromyalgia, not severity-stratified.
Consequence: A clinician deciding whether to offer LDN for fibromyalgia pain cannot yet rely on a consistent evidence verdict — the meta-analytic literature is genuinely divided, so any expectation of benefit should be modest and framed as uncertain. Falsifiable: a pre-registered pooled responder analysis of LDN in fibromyalgia demonstrating a significant ≥30% pain-responder advantage over placebo would resolve the conflict toward the positive meta-analysis; conversely, a pooled responder analysis finding no ≥30% advantage (or equivalence) would resolve it toward the null meta-analysis.
1 GPCR Autoantibody-Targeted Pharmacological Strategies
All below are research-stage only; no ME/CFS human dosing data exists for any. No clinical GPCR autoantibody test is available — CellTrend ELISA is research-only; these stratifications cannot be implemented in routine care. No stopping criteria, monitoring parameters, or drug-interaction data specific to ME/CFS exist for any proposal below. (Azcue et al. 2026).
Certainty: 0.50. Low-dose propranolol (5mg BID, titrated to max 40mg BID) as competitive antagonist blocking \(\beta_2\)-AAb binding, allowing receptor resensitization. Titration based on HRV response and orthostatic BP. Prediction: \(\beta_2\)-AAb-positive patients show greater HRV improvement on propranolol vs \(\beta_2\)-AAb-negative. Contraindicated in asthma, bradycardia, hypotension. Not a clinical recommendation — research-stage. (Azcue et al. 2026) (Stein et al. 2025).
Experimental M1/M4 PAMs (e.g., VU0467154) enhance endogenous acetylcholine signaling without direct agonism, potentially synergizing with moderate M1/M4-AAb-mediated cognitive protection. Could benefit M1/M4-AAb-low ME/CFS with cognitive impairment. No approved M1/M4 PAMs exist; all are experimental compounds. Prediction: greater cognitive improvement in moderate-AAb vs low-AAb patients on PAM. Safety: cholinergic excess (bradycardia, GI upset) at high doses. (Azcue et al. 2026).
Certainty: 0.45. If \(\beta_2\)-AAbs are pathogenic via Fc\(\gamma\)R engagement (not direct receptor agonism), Fc(ab’)\(""_2\) fragments (IVIG digested, Fc removed) would block Fc\(\gamma\)R-mediated monocyte activation without removing antibodies. Requires experimental compounding. Prediction: Fc\(\gamma\)R blockade reduces IL-6 and TNF\(\alpha\) in \(\beta_2\)-AAb-positive patients vs placebo. High cost. (Hackel et al. 2025) (Azcue et al. 2026).
Certainty: 0.35. Raloxifene 60mg daily (FDA-approved SERM for osteoporosis), ER\(\alpha\) antagonist, may reduce \(\beta_2\)-AR expression on B cells and thus \(\beta_2\)-AAb antigenicity in postmenopausal females. Prediction: \(\beta_2\)-AAb titer reduction at 6 months vs placebo. Contraindicated in premenopausal women; thrombosis risk. Purely mechanistic — no AAb-specific data. (Azcue et al. 2026).
Certainty: 0.30. Dronabinol 2.5mg BID as partial CB1 agonist to amplify muscarinic-AAb-mediated cognitive protection (via CB1-mAChR heterodimer cross-activation) in M1/M4-AAb-low ME/CFS. Prediction: cognitive improvement greater in low-AAb vs high-AAb patients (ceiling effect). Psychoactive side effects limit tolerability. (Azcue et al. 2026).
Certainty: 0.40. Quercetin 500mg phytosome BID upregulates \(\beta_2\)-AR expression via CREB pathway, potentially resensitizing receptors desensitized by chronic \(\beta_2\)-AAb binding. Prediction: HRV improvement in \(\beta_2\)-AAb-positive vs negative patients at 12 weeks. Caution: quinolone antibiotic interaction reduces absorption. (Azcue et al. 2026).
Certainty: 0.45. EPA 2g + DHA 1g daily incorporates into immune cell membranes, altering lipid raft composition and reducing Fc\(\gamma\)R signaling affinity. May dampen Fc\(\gamma\)R-mediated cytokine production triggered by GPCR-IgG1 AAbs. Prediction: IL-6 reduction correlates with omega-3 index >8% at 24 weeks. Well-tolerated; low risk. (Hackel et al. 2025) (Azcue et al. 2026).
Certainty: 0.38. NAC 600mg TID increases glutathione, shifting redox balance toward reduced state. Oxidative stress promotes B cell activation; NAC may suppress aberrant B cell activity and reduce GPCR AAb production. Prediction: GPCR AAb titer reduction correlates with glutathione increase at 16 weeks. Caution: rare paradoxical bronchospasm. (Azcue et al. 2026).
Certainty: 0.42. Vitamin D3 5,000 IU daily to 25-OH-D 50–80 ng/mL induces Tregs and promotes B cell tolerance. Repletion may reduce GPCR AAb production by restoring immune regulation. Prediction: AAb reduction correlates with CD4+CD25+FoxP3+ Treg increase at 6 months. Low risk; vitamin D deficiency common in ME/CFS. (Azcue et al. 2026).
Certainty: 0.20. PDE4 inhibitors increase intracellular cAMP, which modulates autophagy via AMPK and mTOR pathways. In ME/CFS, where autophagy is dysregulated (elevated ATG13, elevated LC3-II), roflumilast (PDE4 inhibitor, approved for COPD, 500 µg/day) could shift autophagy balance from chronic selective overload toward bulk degradation — complementing the WBH-induced autophagy switch proposed at ch06:Heat-Induced Autophagy Switch: Selective-to-Bulk Transition via HSP70 Redistribution. Falsifiable: roflumilast over 4 weeks reduces LC3-II/p62 ratio toward control levels and improves mitochondrial spare capacity by 30%. No ME/CFS PDE4 data; gastrointestinal side effects in 20% at COPD dose.
Certainty: 0.20. 4-Phenylbutyrate (4-PB, chemical chaperone approved for urea cycle disorders, 9 g/day TID) reduces ER stress by stabilizing protein folding, potentially freeing HSP70 from ER chaperone duty for mitochondrial protein import support. If ER stress contributes to functional HSP70 deficiency in ME/CFS, 4-PB could restore HSP70 availability independently of thermal induction. Falsifiable: 4-PB over 4 weeks reduces ER stress markers (GRP78, CHOP) by 30% and increases available HSP70 by 25%. No ME/CFS data; taste disturbance in 20% at therapeutic dose.
Certainty: 0.20. Mirabegron (beta3-adrenergic agonist, 50 mg/day, approved for overactive bladder) activates brown adipose tissue via UCP1-dependent thermogenesis. If ME/CFS cold intolerance reflects failed BAT activation from sympathetic dysregulation (ch10:Could Cold Intolerance Reflect Failed Brown Adipose Tissue Activation via Sympathetic Dysfunction?), mirabegron could bypass the sympathetic deficit and restore cold thermogenesis directly at the adipocyte beta3 receptor. However, active BAT is minimal in adults over 40 — the typical ME/CFS demographic — and tachycardia is a known side effect that could worsen POTS symptoms. Falsifiable: mirabegron over 4 weeks increases BAT ^18F-FDG uptake by >50% in ME/CFS patients under 40 with detectable baseline BAT. No ME/CFS data; BAT activation demonstrated in healthy young adults only. Pre-existing POTS is a relative contraindication.