Primary Mechanism Map: Causal Loops, Detection, and Targeted Interventions
The integrated model (Chapter Integrated Multi-System Models) and global sensitivity analysis (Section Global Sensitivity Analysis and Drug Target Ranking) identify several distinct causal primary mechanisms in ME/CFS, each sustained by its own positive feedback loop and accessible to a different intervention class. This section synthesizes these mechanisms into a single map: what each loop does, how to detect it, what targets it, and on which of the three recovery timescales (Table Three Recovery Timescales: Demand Reduction, Damage Reduction, Ceiling Restoration) the intervention acts.
Three principles frame the map:
Mechanisms are loops, not deficits. Every entry describes a self-sustaining feedback circuit rather than a static abnormality. The disease attractor (Section Hysteresis and the Intervention Window) emerges from the simultaneous engagement of multiple loops; recovery requires shrinking or breaking enough of them that the attractor disappears.
Evidence tiers vary widely. Some interventions have RCT evidence (LDN, low-dose hydrocortisone, fludrocortisone); some have mechanistic rationale and case-series support (CoQ10, nattokinase, cromolyn); others are research-stage only (BC007, daratumumab, gene therapy). Each row of the table specifies tier explicitly. Mechanistic rationale is not clinical recommendation — the warnings in Chapters Urgent Action Plan for Severe Cases and Medications Targeting Underlying Mechanisms apply.
Patients differ in which loops dominate. Sensitivity analysis predicts subtype-specific bottlenecks (Section Treatment Selection and Optimization): the same parameter (\(\alpha_\text{CI}\)) is the dominant target in metabolic-dominant patients (\(S_T \approx 0.35\)) but nearly irrelevant in immune-dominant patients (\(S_T \approx 0.04\)). The map informs which loops to interrogate first for a given patient phenotype, not which interventions to apply universally.
1 Master Map
| Mechanism | Driver parameter | Sustaining loop (one line) | Targeting interventions (evidence tier) | Timescale |
|---|---|---|---|---|
| Complex I deficit | \(\alpha_\text{CI}\) | ETC failure \(\rightarrow\) ROS \(\rightarrow\) further ETC damage | CoQ10/ubiquinol (M); methylene blue (R); NR/NMN (M); riboflavin (M) | Medium–Slow |
| MCAS | \(K_\text{MC}\) | Mast cell mediator release \(\rightarrow\) tissue inflammation \(\rightarrow\) MC sensitization | H1 + H2 antihistamines (E); cromolyn (E); LTRAs (E); ketotifen (M); xolair (R) | Fast–Medium |
| GPCR autoantibodies | Autoantibody titre | Autoantibody \(\rightarrow\) receptor dysregulation \(\rightarrow\) vascular/autonomic failure \(\rightarrow\) ongoing immune activation | Immunoadsorption (R, mixed RCTs); IVIG (M); efgartigimod, daratumumab, BC007 (R) | Medium |
| Microclots / fibrinaloid amyloid | \(\beta_\text{epoxy}\) | Fibrin amyloid \(\rightarrow\) microvascular blockage \(\rightarrow\) tissue hypoxia \(\rightarrow\) oxidative stress \(\rightarrow\) more clotting | Triple anticoagulation (R); nattokinase (M); lumbrokinase (M) | Medium |
| Autonomic / OI / POTS | \(G_\text{baro}\), \(V_\text{blood}\) | Reduced blood volume \(\rightarrow\) cerebral hypoperfusion \(\rightarrow\) orthostatic stress \(\rightarrow\) sympathetic overdrive | Fludrocortisone (E); midodrine (E); ivabradine (E); pyridostigmine (E); compression + salt (E) | Fast |
| HPA / cortisol dysregulation | \(n_F\) | Low cortisol \(\rightarrow\) unrestrained inflammation \(\rightarrow\) energy drain \(\rightarrow\) further HPA suppression | Low-dose hydrocortisone 5–15 mg (E); DHEA (M); circadian protocol (M) | Medium |
| Immune exhaustion / chronic activation | \(k_\text{exh}\) | Persistent low-grade activation \(\rightarrow\) T cell exhaustion \(\rightarrow\) poor pathogen control \(\rightarrow\) sustained activation | LDN (E); rintatolimod (R); mycophenolate (R); inebilizumab (R) | Medium |
| BBB hyperpermeability | \(P_0\) | Peripheral inflammation \(\rightarrow\) BBB leak \(\rightarrow\) neuroinflammation \(\rightarrow\) autonomic dysreg \(\rightarrow\) more peripheral stress | PEA (M); luteolin (M); low-dose aripiprazole (R); minocycline (R) | Medium |
| Neuroinflammation / microglial priming | \(M_a\) | Microglial activation \(\rightarrow\) cytokine release \(\rightarrow\) behavioral/cognitive symptoms | LDN (E); low-dose aripiprazole (R); intranasal insulin (R); dextromethorphan (R) | Medium |
| BH4 depletion | \([\text{BH4}]\) | Inflammation \(\rightarrow\) BH4 oxidation \(\rightarrow\) catecholamine deficit + iNOS uncoupling \(\rightarrow\) more ROS | Sapropterin (R); folinic acid (M); 5-MTHF (M) | Medium |
| Viral persistence / EBV reactivation | \(V(t)\) | Viral replication \(\rightarrow\) chronic immune drain \(\rightarrow\) energy exhaustion | Valacyclovir, valganciclovir (M, Lerner); EBV-targeted (R) | Medium–Slow |
| Gut dysbiosis / SIBO / motility | \(\mathcal{G}_\text{set}\), \(B_\text{SI}\) | Vagal impairment \(\rightarrow\) SIBO \(\rightarrow\) endotoxin \(\rightarrow\) systemic inflammation \(\rightarrow\) more vagal impairment | Rifaximin (E); prucalopride (E); low-FODMAP (M); butyrate, PHGG (M) | Medium |
| TRP channelopathy | TRPM3, TRPM7 activity | Calcium dysregulation \(\rightarrow\) NK/immune dysfunction + smooth muscle issues | LDN (TRPM3 partial, M); naltrexone (M); MDC002 (R); magnesium (M) | Medium |
| Glymphatic / sleep failure | Glymphatic flux | Sleep disruption \(\rightarrow\) glymphatic failure \(\rightarrow\) protein accumulation \(\rightarrow\) neuroinflammation \(\rightarrow\) more sleep disruption | Trazodone (M, glymphatic-friendly); glycine (M); avoid Z-drugs (M); orexin caution (M) | Medium–Slow |
| hEDS connective tissue coupling | \(\kappa\) | Vascular laxity \(\rightarrow\) venous pooling \(\rightarrow\) sympathetic compensation \(\rightarrow\) energy tax | Compression garments (E); supine work (E); ivabradine, beta-blockers (E) | Fast |
| NAD+ / redox imbalance | \(\gamma\) | NAD+ depletion \(\rightarrow\) SIRT inactivation \(\rightarrow\) biogenesis block \(\rightarrow\) metabolic inflexibility | NR, NMN (M); niacinamide (M); NAD+ IV (R); NAD+-sparing lifestyle (M) | Medium–Slow |
2 Mechanisms with No Established Targeted Therapy
The following primary mechanisms have model-derived rationale but no established targeted therapy — they represent the highest-leverage research gaps:
Cell Danger Response (CDR) persistence. Naviaux’s CDR framework posits that ME/CFS represents a failure to complete the salugenesis (recovery) phase after danger response. No specific Tx; suramin trials in autism showed transient effects but ME/CFS RCT not done.
Itaconate / aconitase block. TCA cycle interruption at step 2; metabolomic signatures consistent (Yamano 2016, Ciregia 2016) but no targeted intervention exists. Theoretical: itaconate antagonists (none developed for ME/CFS).
Sphingolipid / ceramide accumulation. Multiple metabolomic studies report altered sphingolipid profiles. No targeted Tx; speculative interventions include myriocin (preclinical only) and dietary ceramide reduction (no evidence).
Lactate / GPR81 bistability. Section Lactate Kinetics and Metabolic Flexibility predicts a metabolic bistable state below \(\alpha_\text{CI} < 0.65\) where the lactate → GPR81 → FFA suppression loop locks glycolytic dominance. Theoretical Tx: medium-chain triglycerides (CPT-I bypass) and GPR81 antagonists (none clinical-grade).
WASF3 supercomplex disruption. Wang/Hwang 2023 identified WASF3 over-expression disrupting mitochondrial supercomplex assembly. No targeted Tx; potential CRISPR or small-molecule WASF3 modulators are research-stage only.
These five mechanisms are flagged as Research Priority in Chapter Entries added 2026-08-26: Central Motor-Drive Fatigability Cascade (Bedard 2026) and represent first-in-class therapeutic opportunities if targeted molecules can be developed and validated.
3 Why Single-Target Trials Underperform: The Multi-Loop Attractor
Network controllability analysis (Section Network Controllability and Minimum Intervention Sets) predicts that the ME/CFS system requires at least 4–6 independent driver nodes for full structural controllability. This is a property of the network topology — independent of parameter values — and provides a mathematical explanation for why monotherapy trials in ME/CFS have consistently shown small average effect sizes even for treatments that produce dramatic responses in individual patients.
4 Master Map
5 Mechanisms with No Established Targeted Therapy
6 Why Single-Target Trials Underperform: The Multi-Loop Attractor
The disease attractor is sustained by multiple loops simultaneously, with patient-specific variation in which loops dominate. The model predicts that biomarker-stratified combination therapy targeting the patient’s 4–6 most-engaged loops will substantially outperform either monotherapy or unstratified combination therapy. (Certainty: 0.55.)
Specifically:
Single-loop interventions reduce attractor depth incrementally. If only one loop is targeted, the attractor remains stable but the patient may relax to a higher \(B\) within the same basin (within-envelope improvement, Table Three Recovery Timescales: Demand Reduction, Damage Reduction, Ceiling Restoration). Symptoms improve; the disease state persists.
Multi-loop interventions can collapse the attractor. Targeting four or more dominant loops simultaneously can shrink the basin enough that noise-driven escape (stochastic resonance, Section Endogenous Oscillations and Hopf Bifurcation) becomes feasible, or can reverse the saddle-node bifurcation entirely.
Loop dominance is patient-specific. Sensitivity analysis predicts that the relevant 4–6 loops differ between metabolic-dominant, immune-dominant, autonomic-dominant, and severe/locked subtypes (Section Global Sensitivity Analysis and Drug Target Ranking).
Falsifiable predictions:
In a stratified RCT comparing monotherapy vs. patient-tailored 4–6-mechanism combination therapy, the combination arm should show \(> 2 \times\) the effect size of monotherapy at 12 months.
Patients whose biomarker profile identifies \(\geq 4\) engaged loops should respond worse to monotherapy than patients with \(\leq 2\) engaged loops, because monotherapy in the multi-loop case leaves most of the attractor intact.
The minimum cocktail size (Section Network Controllability and Minimum Intervention Sets) of 4–6 should match the empirical observation that successful clinical responders typically use 4–6 concurrent interventions, while non-responders or partial responders typically use 1–3.
This hypothesis builds on the network controllability prediction (already in Network Controllability and Minimum Intervention Sets) and the subtype-specific sensitivity prediction (Global Sensitivity Analysis and Drug Target Ranking), unifying them with the practical observation that experienced ME/CFS clinicians (Mestinon + LDN + fludrocortisone + antihistamine + supplement stack patterns) achieve better results than single-agent trials.
7 Limitations of the Map
Table Primary Mechanism Map: Causal Loops, Detection, and Targeted Interventions synthesizes published mechanistic literature, the integrated model’s sensitivity analysis, and clinical observations from established ME/CFS practitioners. It has not been prospectively validated. The evidence tier assignments reflect current literature as of the document version date but should be re-evaluated as new RCTs report. The timescale assignments derive from the three-recovery-timescales framework (Table Three Recovery Timescales: Demand Reduction, Damage Reduction, Ceiling Restoration) and are model-predicted, not measured. Loop-dominance assessment requires biomarker panels not yet standardized in clinical practice. Use this map as a structured framework for hypothesis-driven treatment planning under specialist supervision, not as a substitute for clinical judgment or as a one-size-fits-all protocol.
Table Primary Mechanism Map: Causal Loops, Detection, and Targeted Interventions synthesizes published mechanistic literature, the integrated model’s sensitivity analysis, and clinical observations from established ME/CFS practitioners. It has not been prospectively validated. The evidence tier assignments reflect current literature as of the document version date but should be re-evaluated as new RCTs report. The timescale assignments derive from the three-recovery-timescales framework (Table Three Recovery Timescales: Demand Reduction, Damage Reduction, Ceiling Restoration) and are model-predicted, not measured. Loop-dominance assessment requires biomarker panels not yet standardized in clinical practice. Use this map as a structured framework for hypothesis-driven treatment planning under specialist supervision, not as a substitute for clinical judgment or as a one-size-fits-all protocol.