Mitochondrial Hypotheses
1 WASF3-Mediated Supercomplex Disruption
Certainty: 0.35. WASF3 supercomplex disruption documented in cell models — Wang 2023 ((Wang et al. 2023)) showed that WASF3 overexpression in skeletal muscle cells disassembles mitochondrial respiratory supercomplexes (SC I/III/IV), reducing ETC efficiency by ~40%. WASF3 appears to act as a supercomplex assembly inhibitor — its upregulation prevents individual ETC complexes from organizing into functional supercomplexes, which are essential for efficient electron channelling (direct transfer of electrons from Complex I to Complex III within the supercomplex, minimizing electron leak and ROS production). HIF-1α-driven WASF3 upregulation is mechanistic inference — HIF-1α binds to hypoxia response elements in the WASF3 promoter under conditions of cellular hypoxia or pseudohypoxia (ROS-mediated prolyl hydroxylase inhibition). The HIF-1α→WASF3→SC disruption pathway provides a mechanistic link between oxidative stress (HIF-1α stabilization) and mitochondrial dysfunction (SC disassembly). No interventions have been tested in ME/CFS for WASF3 specifically.
1.1 Cascade: HIF-1α → WASF3 → supercomplex disruption → ATP deficit
Cascade:
- HIF-1α stabilization (chronic ROS → PHD inhibition, or genetic HIF1A variants → constitutive activity)
- WASF3 transcriptional upregulation → WASF3 protein at mitochondrial outer membrane
- Respiratory supercomplex (SC I/III/IV) assembly failure → reduced electron channelling efficiency
- Increased electron leak at Complex I-III interface → ROS (further stabilizes HIF-1α — positive feedback)
- Reduced OXPHOS ATP yield + compensatory glycolysis (PDK1, LDH-A, GLUT1 upregulation)
- Lactate at rest, reduced maximal ATP output → fatigue, PEM, the metabolic signature of ME/CFS
1.2 Step D1: HIF-1α stabilization → WASF3 upregulation (transcriptional trigger)
Mechanism: HIF-1α is the master transcriptional regulator of the hypoxic response. Under normoxia, HIF-1α is constitutively produced but immediately degraded (prolyl hydroxylase hydroxylates HIF-1α → pVHL ubiquitin ligase recognition → proteasomal degradation). Under hypoxia or pseudohypoxia (ROS accumulation, succinate accumulation, fumarate accumulation — all inhibit prolyl hydroxylase), HIF-1α escapes degradation → translocates to nucleus → dimerizes with HIF-1β → binds hypoxia response elements (HREs) → transcriptionally activates target genes including WASF3, PDK1, LDH-A, GLUT1, and VEGF.
Probes for HIF-1α→WASF3 pathway:
- Dimethyl fumarate (DMF, 120–240 mg BID): Activates Nrf2 → Nrf2 binds antioxidant response elements (AREs) → upregulates antioxidant enzymes (SOD, catalase, glutathione peroxidase) → reduces ROS → reduces ROS-mediated prolyl hydroxylase inhibition → restores HIF-1α degradation → reduces WASF3 expression. Additionally, DMF directly inhibits HIF-1α nuclear translocation (Nrf2 competes with HIF-1α for CBP/p300 coactivators). If DMF improves exercise tolerance and reduces lactate → HIF-1α-driven WASF3 upregulation is present and rate-limiting. If DMF causes severe flushing → mast cell hyper-reactivity (Pattern 3, Side Effects as Diagnostic Probes) — pre-treat with aspirin (blocks PGD2-mediated flushing without affecting Nrf2 activation). If flushing is aspirin-preventable → confirms PGD2-mediated MCAS subtype.
- Vitamin C (500–2000 mg, timed with peak prolyl hydroxylase activity — circadian-optimized, Metabolic “Safe Mode” Hypothesis): Prolyl hydroxylase requires vitamin C as a cofactor. If timed vitamin C improves symptoms → prolyl hydroxylase was cofactor-limited, and HIF-1α stabilization was vitamin-C-reversible. Distinguishes cofactor deficiency (simple, nutritional) from ROS-mediated PHD inhibition (requires antioxidant intervention). If vitamin C + NAC works where vitamin C alone does not → ROS is the dominant PHD inhibitor; the vitamin C provides cofactor, NAC scavenges the ROS that was inhibiting PHD.
HIF-1α stabilization is not the dominant mechanism of WASF3 upregulation; WASF3 may be regulated by genetic variants, epigenetic methylation, or non-HIF-1α transcriptional regulators (NF-κB, STAT3)
- Certainty
- Low to Medium
- Level of action
- Partial root cause
1.3 Step D2: Respiratory supercomplex assembly failure (structural mitochondrial defect)
Mechanism: SC I/III/IV forms in a specific stoichiometry (I:III₂:IV) that channels electrons directly from Complex I to Complex III within the lipid bilayer — no need for ubiquinone to diffuse electrons between complexes. This channelling increases electron transfer efficiency and reduces electron leak (ROS production). WASF3 disrupts this assembly → Complex I and III must rely on ubiquinone-mediated electron transfer (slower, leakier) → electrons escape at the I-III interface → superoxide production → oxidative stress. The result: reduced ATP yield per NADH (~2.5 ATP → ~1.5 ATP) and increased ROS per unit of oxygen consumed.
Probes for supercomplex disruption:
- CoQ10 (ubiquinone, 100–300 mg) + NADH (5–20 mg): CoQ10 shuttles electrons from Complex I/II to Complex III. In SC disruption, CoQ10 + NADH flooding provides supraphysiological substrate to compensate for reduced channelling efficiency — more electrons reach Complex III because there is more electron carrier in the membrane. If CoQ10 + NADH dramatically improves symptoms (≥50% improvement) → SC disruption is partial — the remaining supercomplexes can handle increased electron flux. If CoQ10 + NADH produces only modest benefit (10–20%) → SC disruption is severe — substrate flooding cannot overcome the assembly defect; mitochondria need turnover, not cofactor supplementation. If CoQ10 + NADH produces WORSENING → electron leak at Complex I-III interface increases when substrate flux is increased → more ROS production → oxidative damage. The direction of CoQ10 response IS the SC disruption severity marker.
- L-carnitine (500–1000 mg): Enhances fatty acid oxidation → generates FADH₂ (enters at Complex II, electron-transferring flavoprotein → ubiquinone → Complex III). If L-carnitine improves symptoms where CoQ10 does NOT → Complex I is the specific bottleneck, and bypassing it via Complex II (FADH₂ from FAO) restores electron flux. The SC disruption is Complex I-specific — Complex II-containing supercomplexes are intact.
- MitoQ (mitoquinone, 5–10 mg): Ubiquinone conjugated to triphenylphosphonium cation → accumulates 100–1000× in mitochondrial matrix (driven by membrane potential). More potent than CoQ10. If MitoQ works where CoQ10 does NOT → the ubiquinone pool in the mitochondrial membrane is inaccessible to oral CoQ10 (poor bioavailability) but MitoQ reaches the matrix. MitoQ response confirms ubiquinone deficiency at the mitochondrial level.
1.4 Step D3: Reduced OXPHOS ATP yield → energy deficit
Mechanism: Even with adequate substrate delivery and ETC function, if SC disruption reduces electron channelling efficiency, the ATP yield per NADH is reduced. The result: more oxygen consumed per ATP produced (reduced P/O ratio), more substrate consumed per ATP, and more ROS produced. This is the biochemical signature of mitochondrial inefficiency — not failure, but reduced efficiency.
Probes for compensating for reduced ATP yield:
- Creatine monohydrate (5 g/day): Phosphocreatine buffer — provides rapid ATP regeneration from phosphocreatine + ADP → creatine + ATP via creatine kinase. Bypasses ETC entirely for short-term ATP buffering. If creatine improves exercise tolerance → ATP buffering is insufficient for the reduced ATP production rate. If creatine produces NO improvement → the ATP deficit is too large for buffering to compensate, OR the bottleneck is at ATP utilization (creatine already saturating the creatine kinase system).
- D-ribose (5–15 g/day): Pentose sugar essential for ATP synthesis (ribose → PRPP → purine synthesis → ATP). NOT an energy source — it provides the ribose backbone for ATP. If D-ribose improves symptoms → ATP synthesis capacity is limited by ribose availability (the de novo purine synthesis pathway is rate-limiting). If D-ribose has no effect → the bottleneck is in the ETC or mitochondrial membrane, not in nucleotide synthesis.
- NMN/NR (NAD⁺ precursors): If SC disruption also impairs Complex I NADH oxidation → NADH accumulates, NAD⁺ is depleted, and NAD⁺ supplementation may restore the NAD⁺/NADH ratio. If NMN/NR + CoQ10 produces benefit where either alone does not → dual mitochondrial lesion: SC disruption (CoQ10-dependent) AND NAD⁺ deficiency (NMN/NR-dependent). The NAD⁺ deficiency is compounding the SC disruption.
Consequence: The WASF3 cascade is the only mitochondrial hypothesis in the chapter that links a specific genetic/protein-level defect (WASF3 overexpression) to a specific structural mitochondrial defect (SC disassembly) to measurable metabolic consequences (reduced ATP yield, increased ROS). The diagnostic sequence: (1) CoQ10 ± response magnitude → severity of SC disruption; (2) DMF response → HIF-1α is the transcriptional driver; (3) vitamin C/NAC response → ROS-mediated PHD inhibition vs. cofactor deficiency; (4) L-carnitine response → Complex I-specific vs. global SC disruption; (5) NMN/NR response → compounding NAD⁺ deficiency. The DecodeME GWAS results (if WASF3 locus is implicated) would validate this cascade at the genetic level — if WASF3 variants are associated with ME/CFS, this cascade transitions from 0.35 to 0.50+ certainty. Origin: mechanistic-pathway-tracing.
2 PDH Inhibition / Pyruvate Dehydrogenase
Certainty: 0.40. PDK-driven PDH inhibition is mechanistically well-grounded in cytokine biology — PDK1 and PDK4 are transcriptionally upregulated by PPARα and HIF-1α in response to inflammatory cytokines, fasting, and hypoxia. PDH is the gatekeeper enzyme: it converts pyruvate (from glycolysis) to acetyl-CoA (which enters the TCA cycle). When PDH is phosphorylated by PDK, pyruvate cannot enter the TCA cycle → pyruvate accumulates → converted to lactate by lactate dehydrogenase → lactate exits the cell → systemic lactate elevation at low workloads. The consequence: the muscle is forced into anaerobic metabolism at exertion levels far below the anaerobic threshold of a healthy person. ~83% of glucose-derived ATP depends on PDH-mediated pyruvate entry to TCA — with PDH blocked, glucose becomes an inefficient fuel. This is the biochemical explanation for the PEM phenomenon: exertion produces lactate at workloads that should be aerobic, and ATP yield per glucose molecule drops from ~36 to ~2 (glycolysis only). The 2-day CPET finding (reduced VO₂peak and increased lactate on Day 2) is consistent with PDH inhibition that worsens with repeated exertion.
2.1 Cascade: Cytokines → PDK → PDH block → lactate + energy deficit
Cascade:
- Inflammatory cytokines (TNF-α, IL-1β, IFN-γ) → PDK1/PDK4 upregulation (HIF-1α, PPARα)
- PDK phosphorylates PDH E1α at Ser293 → PDH inactivation → pyruvate blocked from TCA
- Glycolytic-ETC uncoupling → three arms:
- lactate accumulation → muscle fatigue, burning, PEM at low workloads
- reduced acetyl-CoA → reduced TCA flux → reduced ATP yield → energy deficit
- compensatory fatty acid oxidation → lipid dependence → if β-oxidation impaired → dual metabolic blockade
2.2 Step E1: PDK upregulation (inflammatory trigger)
Probe: LDN (0.5–4.5 mg, anti-inflammatory). LDN reduces TLR4-mediated cytokine production → reduces PDK transcriptional drive.
If LDN reduces exertional lactate (measurable with handheld lactate meter at standardized submaximal workload): Inflammatory PDK activation is present and rate-limiting — reducing inflammation restores PDH function. This is strong evidence for cytokine-driven metabolic blockade. The lactate measurement is objective — it does not rely on subjective symptom report.
If LDN reduces symptoms but lactate is unchanged: LDN’s benefit is not through PDH restoration — it is through anti-neuroinflammatory, TRPM3, or endorphin mechanisms independent of muscle metabolism. The lactate dissociation distinguishes metabolic from neurological LDN benefit.
If LDN does NOT reduce lactate: PDK may be induced by non-TLR4 inflammatory pathways (TNF-α from non-TLR4 sources, direct HIF-1α stabilization from ROS/hypoxia independent of inflammation), or PDH is inhibited by mechanisms beyond PDK (PDH phosphatase inhibition, thiamine deficiency, or genetic PDH variants).
2.3 Step E2: PDH phosphorylation — enzymatic block (the enzyme level)
Probes for restoring PDH activity:
- Dichloroacetate (DCA, 12.5–25 mg/kg/day, investigational in ME/CFS): PDK inhibitor — directly blocks the kinase that phosphorylates and inactivates PDH. DCA restores PDH activity regardless of upstream cytokine drive. If DCA dramatically reduces lactate and improves exercise tolerance → PDH phosphorylation IS the rate-limiting metabolic step. DCA is the highest-specificity probe for PDH in this chapter. CAUTION: DCA causes reversible peripheral neuropathy at cumulative doses >30 g — limit trial duration; monitor nerve conduction. DCA is not approved for ME/CFS — requires off-label prescription or clinical trial access.
- Thiamine (B1, 100–900 mg/day, IV or oral): Essential cofactor for PDH (thiamine pyrophosphate is the prosthetic group of PDH E1). If thiamine works → PDH cofactor deficiency. Simple, reversible, nutritional — the most accessible PDH probe. Distinguish: thiamine deficiency vs. PDK-driven phosphorylation. If LDN does NOT reduce lactate but thiamine DOES → PDH inhibition is from cofactor deficiency, not PDK-mediated phosphorylation. If thiamine works and LDN works → dual mechanism: inflammatory PDK activation + thiamine deficiency (common in SIBO, gut dysbiosis).
- Alpha-lipoic acid (300–600 mg BID): Antioxidant that also serves as a PDH cofactor and reduces PDK expression. If ALA works → oxidative stress drives PDK expression, and antioxidant + cofactor support restores PDH.
- Magnesium (400–800 mg, IV or oral): Mg²⁺ is a PDH phosphatase cofactor — PDH phosphatase requires Mg²⁺ to dephosphorylate and activate PDH. If IV magnesium improves exercise tolerance → Mg²⁺ deficiency prevented PDH dephosphorylation. The PDH enzyme is phosphorylated (PDK active) AND cannot be dephosphorylated (PDH phosphatase Mg²⁺-deficient).
- Niacin/NR (NAD⁺ precursors): NAD⁺ is a substrate for PDH (converts NAD⁺ → NADH). If PDH activity is restored (by thiamine, DCA, or Mg²⁺) but symptoms do NOT improve → NAD⁺ may be rate-limiting for PDH throughput. Add nicotinamide riboside (NR) to provide NAD⁺ substrate. If NR + thiamine works where thiamine alone does not → dual cofactor deficiency (thiamine + NAD⁺).
Distinguishing DCA vs. thiamine response: DCA = PDK inhibition (restores PDH regardless of cofactor status). Thiamine = cofactor repletion (restores PDH only if cofactor was deficient). DCA+ but thiamine null → PDK is the dominant inhibitor, cofactor status is adequate. Thiamine+ but DCA null (impossible — DCA should always work if PDH phosphorylation is the block; a null DCA response suggests the PDH complex is structurally damaged or absent, not just phosphorylated). Both null → PDH is not the metabolic bottleneck — the ETC downstream of acetyl-CoA, the TCA cycle itself, or substrate delivery is rate-limiting.
2.4 Step E3: Lactate accumulation and anaerobic shift (metabolic consequence)
Mechanism: PDH block → pyruvate → lactate. Lactate is measurable in real time with a handheld meter (fingerstick). Serial lactate measurements at standardized submaximal workloads (e.g., walking at 1 mph for 5 minutes) before and after interventions provide objective pharmacodynamic data — does the drug reduce lactate at a fixed workload?
Probes for lactate reduction:
- Pyridostigmine (30–60 mg): Improves tissue perfusion → increases oxygen delivery → reduces anaerobic reliance. If pyridostigmine reduces lactate at a fixed workload → the anaerobic shift is perfusion-driven (oxygen not reaching mitochondria), not enzyme-driven (PDH block). The lactate reduction from pyridostigmine quantifies the delivery contribution.
- Midodrine (2.5–10 mg): Increases perfusion pressure. If midodrine reduces lactate → perfusion pressure was the bottleneck. If midodrine does NOT reduce lactate → perfusion pressure is adequate; the lactate is from mitochondrial-level metabolic block.
Both delivery (perfusion) and enzyme (PDH) bottlenecks coexist — combined 70% reduction indicates both are rate-limiting; residual 30% suggests additional bottlenecks: ETC complex deficiency, NAD⁺ deficit, or genetic metabolic limitations
- Certainty
- Low to Medium
- Level of action
- Partial root cause
- Lactate response latency: Pyridostigmine reduces lactate within hours (hemodynamic effect). Thiamine reduces lactate within days (cofactor repletion). DCA reduces lactate within hours (direct PDK inhibition). The latency distinguishes the mechanism: hemodynamic → enzyme cofactor → enzyme activation.
Consequence: The PDH cascade is the most objectively measurable in the chapter — lactate is quantifiable with a $50 handheld meter. The diagnostic sequence: (1) measure resting and exertional lactate; (2) LDN trial → lactate reduction = inflammatory PDK; (3) thiamine trial → lactate reduction = cofactor deficiency; (4) pyridostigmine trial → lactate reduction = perfusion deficit; (5) DCA trial → definitive test of PDH phosphorylation as rate-limiting. The combination of lactate measurement + drug probes converts PDH tracing from subjective symptom report to pharmacometabolic assay. A patient who produces lactate above 4 mmol/L at 50 watts on a stationary bike and drops to 2 mmol/L on thiamine has had their PDH cofactor status measured without a muscle biopsy. Origin: mechanistic-pathway-tracing.
3 Acquired Ischemic Mitochondrial Myopathy (AIMM) / Na⁺-Ca²⁺ Cascade
Certainty: 0.40. NHE1→NCX1→mitochondrial Ca²⁺ overload cascade documented in ischemia-reperfusion injury (cardiac and skeletal muscle). AIMM = acquired ischemic mitochondrial myopathy — the concept proposed by Scheibenbogen and Wirth that repeated microvascular ischemia-reperfusion episodes in skeletal muscle cause cumulative mitochondrial structural damage, producing the PEM phenotype. MDC002/Mitodicure is a dual NHE1 + NCLX modulator in preclinical development for ME/CFS.
#clinical-caution()
3.1 Step F0: Capillary basement membrane thickening — structural diffusion barrier upstream of hypoperfusion
Mechanism: Before any functional microvascular defect (endothelial dysfunction, microclots, RBC stiffness), there is a structural anatomical barrier: capillary basement membrane (BM) thickening. Charlton, Slaghekke et al. (2025) (Charlton et al. 2025) documented near-complete separation of ME/CFS patients from healthy controls by percentage BM coverage in vastus lateralis biopsies (max HC 62.7% vs min patient 63.2%), independently replicated in Berlin (Aschman et al. 2023) and Aarhus (Agergaard et al. 2023). This is a structural lesion: thickened BM increases the O₂ diffusion distance from capillary lumen to myocyte, creating a delivery-complete but diffusion-incomplete state — capillaries are perfused, RBCs are oxygenated, but O₂ physically cannot cross the thickened BM fast enough to sustain aerobic ATP production. Unlike functional perfusion deficits (Steps F1a–d), this is not a flow problem — it is a physical diffusion barrier.
Additional ultrastructural features amplify the diffusion impairment: (a) decreased capillary tortuosity reduces total endothelial surface area for gas exchange; (b) decreased capillary contact length reduces the fraction of the myocyte surface in direct apposition to a capillary; (c) endothelial hypertrophy narrows capillary lumens from the inside, potentially creating heterogeneous perfusion where some capillaries are functionally excluded even though total inflow is normal (Endothelial Hypertrophy and Heterogeneous Perfusion — Functional Capillary Dropout, cert 0.40). These three features compound the BM thickness effect: less surface area × narrower contact × thicker barrier = multiply impaired O₂ diffusion.
The bed rest comparator is critical: healthy volunteers after strict 60-day bed rest showed muscle atrophy + reduced OXPHOS (the expected detraining phenotype), while patients showed no atrophy but BM thickening + glycolytic fiber shift — a structurally distinct phenotype that excludes deconditioning as the sole explanation (Charlton et al. 2025).
Functional signature: Impaired peripheral O₂ extraction despite normal cardiac output and pulmonary gas exchange — the gold-standard finding from invasive CPET (Joseph, Pari, et al. 2022) (O₂ extraction 0.69 vs 0.77 in controls, \(p < 0.001\)). The mitochondrial defect in ME/CFS is functional (not quantitative) — Complex II+III activity correlates with O₂ extraction (\(ρ = -0.33\), \(p = 0.028\)) while mitochondrial biomass (CS activity, mtDNA) does not (Squires et al. 2026) — consistent with mitochondria that are O₂-starved, not intrinsically broken.
Cascade (structural → functional):
Capillary BM thickening (Charlton 2025, Aschman 2023, Agergaard 2023)
→ O₂ diffusion distance ↑ (physical barrier independent of flow rate)
→ + decreased tortuosity → reduced exchange surface area
→ + decreased contact length → reduced myocyte-capillary apposition
→ + endothelial hypertrophy → heterogeneous perfusion → functional capillary dropout
→ Impaired peripheral O₂ extraction (Joseph 2022, Squires 2026)
→ Chronic tissue-level O₂ insufficiency
→ Functional mitochondrial impairment (O₂-starved, not broken)
→ Anaerobic metabolism → lactate accumulation → proton accumulation
→ NHE1 activation → intracellular Na⁺ overload (→ Step F2)
→ NCX1 reverse mode → mitochondrial Ca²⁺ overload (→ Step F3)
→ Mitochondrial structural damage → mitophagy impairment (→ Step F4)
→ ROS → further endothelial damage → more BM thickening → (loop closes)Probes for the structural BM barrier (Step F0 vs functional perfusion deficits Step F1):
- Pyridostigmine (30–60 mg): Improves cardiac preload and cholinergic vasodilation (Joseph, Pari, et al. 2022). If pyridostigmine improves VO₂peak → the deficit is partly perfusion-mediated (preload or autonomic). If pyridostigmine does NOT improve VO₂peak → the deficit is structural — BM thickening is the rate-limiting barrier, and increasing flow through a thickened BM doesn’t accelerate diffusion through it. The +0.9 mL/kg/min improvement in Joseph 2022 suggests ~10% of the O₂ extraction deficit is perfusion-mediated; the remaining ~90% may be structural.
- Pentoxifylline (400 mg TID): Improves RBC deformability — stiff RBCs can now transit narrower capillaries. If pentoxifylline improves NIRS-measured O₂ extraction → endothelial hypertrophy is creating luminal occlusion that flexible RBCs can now pass. If pentoxifylline does NOT improve O₂ extraction → the bottleneck is not luminal occlusion but BM diffusion distance — making RBCs more flexible doesn’t help if the BM itself is the barrier. Pentoxifylline thus discriminates Step F0 (BM thickness) from Step F0c (endothelial hypertrophy).
Primary lesion is the BM itself — structural barrier that no currently available drug directly targets; interventions would need BM thinning, collagen degradation reversal, or downstream mitochondrial protection
- Certainty
- Low to Medium
- Level of action
- Partial root cause
Discriminating probe between structural (F0) vs perfusion (F1) vs mitochondrial (F2–F4):
- Combined pyridostigmine + pentoxifylline + midodrine + compression: Maximizes perfusion through all functional pathways (preload, RBC deformability, perfusion pressure, venous return). If this combination normalizes VO₂peak → the deficit was functional (Step F1), not structural (Step F0). If this combination produces minimal or no improvement → the deficit is structural BM thickening — no amount of improved blood flow can accelerate O₂ diffusion through a pathologically thickened capillary wall.
- No clean discriminator exists between structural BM thickening and mitochondrial primary dysfunction: both predict failure of perfusion-enhancing interventions. The discriminating probe is EM-measured BM thickness itself — if BM thickness correlates with O₂ extraction impairment (r > 0.7) in a cohort where both are measured simultaneously, structural BM thickening is the rate-limiting step. If mitochondrial content out-predicts BM thickness → mitochondrial dysfunction is primary. This test has not been performed.
Certainty: 0.55. The BM thickening finding is replicated in 3 countries. The cascade from BM thickening → impaired O₂ diffusion is structurally plausible. The link to downstream mitochondrial functional impairment is supported by Squires 2026 (functional, not quantitative, mitochondrial defect). The full cascade from BM thickening through the ionic cascade (Steps F1–F4) to PEM is inferential — no study has traced all steps in the same patients. (Evidence source: limb skeletal muscle EM + invasive CPET — Inference target: O₂ cascade through AIMM. Link is indirect — the structural lesion and functional consequence are each independently demonstrated, but the causal chain connecting them through the Na⁺-Ca²⁺ cascade is theoretical.)
(Origin: brainstorm 1.1 — Phase 5d cascade trace.)
Falsifiable predictions: + EM-measured %BM coverage correlates with invasive CPET-measured O₂ extraction (arterial-venous O₂ difference at VO₂max) with r > 0.7 in a single cohort + BM thickness is a stronger predictor of O₂ extraction than capillary density, mitochondrial content (CS activity), or RBC deformability in multivariate regression in the same cohort + Combined pyridostigmine + pentoxifylline + midodrine + compression (maximal perfusion intervention) produces < 20% improvement in VO₂peak in patients with BM coverage > 65% — the improvement plateaus because the structural barrier cannot be overcome by improving flow + Falsified if: BM thickness does not independently predict O₂ extraction in a cohort where both are measured simultaneously — then BM thickening is not rate-limiting for O₂ delivery. Falsified if: maximal perfusion intervention normalizes VO₂peak — then the deficit is entirely functional, not structural.
Consequence: If the primary barrier to oxygen delivery in ME/CFS muscle is a structurally thickened capillary wall — not low blood flow, not stiff blood cells, not mitochondrial failure — then drugs that improve flow (pyridostigmine, midodrine) can only partially help, and the most important treatment target is the capillary wall itself. This is currently untreatable with approved drugs. The finding shifts the research agenda from “improving mitochondrial function” to “restoring capillary structure” — a fundamentally different therapeutic direction.
3.2 Cascade: Hypoperfusion → Na⁺ overload → mitochondrial Ca²⁺ → mPTP → PEM
Cascade:
- Capillary hypoperfusion (endothelial dysfunction, microclots, RBC deformability loss)
- Ischemia during exertion → anaerobic metabolism → intracellular acidosis
- NHE1 activation → intracellular Na⁺ overload → NCX1 reverse mode (imports Ca²⁺)
- Mitochondrial Ca²⁺ overload → mPTP opening → loss of ΔΨm
- Impaired ATP production, cytochrome c release → apoptosis/necrosis
- Structural mitochondrial damage → progressive loss of oxidative capacity → PEM, Day 2 CPET decline
3.3 Step F1: Capillary hypoperfusion (vascular trigger — upstream of ionic cascade)
Mechanism: The initial event is impaired oxygen delivery to working muscle. This can be from: (a) endothelial dysfunction — GPCR AAb (β2-AR, M2/M4) impair endothelium-dependent vasodilation → exercise hyperemia fails → muscle receives less blood during exertion; (b) microclots — amyloid fibrin(ogen) microclots (Pretorius 2021) obstruct capillaries → reduced functional capillary density; (c) RBC deformability loss — stiff erythrocytes cannot pass through capillaries smaller than their diameter → impaired microvascular flow; (d) autonomic failure — impaired sympathetic vasoconstriction in non-exercising muscle beds + impaired vasodilation in exercising muscle → blood flow misdistribution.
Probes for the vascular trigger:
- Pyridostigmine (30–60 mg): Enhances cardiac preload and output → increases total blood flow → more oxygen delivery to all tissues (Joseph, Arevalo, et al. 2022). If pyridostigmine improves VO₂peak at CPET → the delivery deficit is preload/cardiac output-mediated. If pyridostigmine does NOT improve VO₂peak → the delivery deficit is microvascular (capillary-level obstruction, not macro-level flow).
- Midodrine (2.5–10 mg): Increases perfusion pressure. If midodrine improves exercise tolerance → perfusion pressure was insufficient to overcome capillary resistance. If midodrine does NOT improve → perfusion pressure is not the bottleneck.
- Compression garments (abdominal + leg): Reduce venous pooling → maintain preload without drugs. If compression improves exercise tolerance → venous pooling contributes to delivery deficit.
- Pyridostigmine + midodrine + compression: If the combination improves VO₂peak where any single intervention does not → multi-level delivery deficit: preload, perfusion pressure, and venous pooling are all contributing.
Bottleneck is NOT oxygen delivery — it is mitochondrial oxygen utilization; oxygen is reaching mitochondria but they cannot use it; proceed to mitochondrial-level probes
- Certainty
- Low to Medium
- Level of action
- Partial root cause
3.4 Step F2: Intracellular Na⁺ overload (ionic consequence of ischemia)
Mechanism: Ischemic muscle shifts to anaerobic glycolysis → lactic acid production → intracellular pH drops. The Na⁺/H⁺ exchanger (NHE1) is activated by low pH — it extrudes one H⁺ in exchange for one Na⁺ → normalizes pH at the cost of Na⁺ accumulation inside the cell. This is protective in acute ischemia (prevents acid-induced protein denaturation) but pathological in repeated ischemia-reperfusion (chronic Na⁺ overload).
Probes for the Na⁺ overload step:
- MDC002/Mitodicure (investigational, NHE1 inhibitor + NCLX activator): Blocks NHE1 (prevents Na⁺ entry) and activates the mitochondrial Na⁺/Ca²⁺/Li⁺ exchanger (NCLX, extrudes mitochondrial Ca²⁺). If MDC002 improves PEM recovery and CPET Day 2 performance in clinical trials → the Na⁺→Ca²⁺ cascade is causally driving mitochondrial damage. Currently investigational — no clinical trial results in ME/CFS as of writing.
- Amiloride (NHE inhibitor, 5–10 mg): Weak NHE1 inhibitor at clinical doses (more potent at NHE2/NHE3 in kidney). If amiloride improves exercise tolerance → NHE1 is rate-limiting and even weak inhibition produces benefit. This is a low-certainty, high-risk probe (electrolyte disturbances) — only informative if the result is dramatic.
- Ranolazine (late Na⁺ current inhibitor, 500–1000 mg BID): Blocks the late inward Na⁺ current (late INa) in cardiac and skeletal muscle, reducing intracellular Na⁺. Approved for angina. If ranolazine improves PEM → the Na⁺ overload is from late INa channels (not just NHE1) — broader Na⁺ dysregulation. If ranolazine null → NHE1 is the dominant Na⁺ entry pathway, not late INa.
3.5 Step F3: NCX1 reverse mode → mitochondrial Ca²⁺ overload (the tipping point)
Mechanism: NCX1 normally operates in forward mode (3 Na⁺ in, 1 Ca²⁺ out — uses the Na⁺ gradient to extrude Ca²⁺). When intracellular Na⁺ is very high (from NHE1 activity), the Na⁺ gradient collapses. NCX1 reverses: it imports Ca²⁺ into the cell while extruding Na⁺. The result is cytosolic Ca²⁺ overload. Mitochondria buffer cytosolic Ca²⁺ by taking it up through the mitochondrial calcium uniporter (MCU) — but chronic Ca²⁺ uptake → mitochondrial Ca²⁺ overload → mPTP opening → mitochondrial depolarization and structural damage.
Probes for the NCX1→mitochondrial Ca²⁺ step:
- Magnesium (IV or high-dose oral, 400–800 mg): Mg²⁺ is a physiological NMDA receptor antagonist and NCX modulator — high Mg²⁺ competes with Ca²⁺ at multiple sites. If IV magnesium improves PEM recovery → the Ca²⁺ overload is Mg²⁺-suppressible.
- Dantrolene (ryanodine receptor antagonist, 25–50 mg): Blocks Ca²⁺ release from sarcoplasmic reticulum through RyR1 channels in skeletal muscle. If dantrolene improves PEM → the Ca²⁺ overload is from SR Ca²⁺ release (RyR1 leak) + NCX1 reverse mode — dual-source Ca²⁺ pathology. If dantrolene null → NCX1 is the sole Ca²⁺ source.
Ca²⁺ overload has multiple entry pathways — intervention at NHE1 level (MDC002) and Mg²⁺-sensitive level both reduce Ca²⁺ overload, consistent with full NHE1→NCX1→mitochondrial Ca²⁺ cascade as causal
- Certainty
- Low to Medium
- Level of action
- Partial root cause
3.6 Step F4: Mitochondrial structural damage → mitophagy impairment
Mechanism: Chronically damaged mitochondria with open mPTP are targets for mitophagy (selective autophagy of mitochondria). PINK1 accumulates on depolarized mitochondria → recruits Parkin → ubiquitination → autophagosome engulfment → lysosomal degradation. In ME/CFS, mTORC1 is hypothesized to be overactive (driven by chronic low-grade inflammation and insulin signaling) → mTORC1 inhibits autophagy initiation (ULK1 phosphorylation) → damaged mitochondria accumulate because they cannot be cleared.
Probes for the mitophagy step:
- Rapamycin (1–3 mg/week, mTORC1 inhibitor): Inhibits mTORC1 → disinhibits autophagy → allows clearance of damaged mitochondria. If rapamycin improves exercise tolerance over weeks → damaged mitochondria were accumulating and their clearance was mTORC1-suppressed. Onset latency: improvement at 2–4 weeks → mitophagy requires time to clear the backlog of damaged mitochondria. Dose ceiling: mTORC1 inhibition is beneficial; at higher cumulative exposure, mTORC2 inhibition causes immunosuppression and insulin resistance — the therapeutic window is defined by mTORC1:mTORC2 selectivity (Sarbassov et al. 2006) (Lamming et al. 2012).
- Urolithin A (500–1000 mg/day, mitophagy activator): Activates mitophagy through PINK1/Parkin-independent pathway. If urolithin A works where rapamycin does NOT → the mitophagy defect is not mTORC1-mediated — it is at the PINK1/Parkin recognition step, or the lysosomal degradation capacity is preserved but the signal to initiate mitophagy is absent. Urolithin A bypasses the initiation signal.
- NAC + CoQ10 (antioxidant support): If NAC + CoQ10 improves exercise tolerance but rapamycin does NOT → the mitochondrial damage is from ROS-mediated functional impairment (electron leak), not structural damage requiring clearance. The mitochondria can be rescued with antioxidants — they do not need to be replaced.
- Mitochondrial supplement worsening (CoQ10, carnitine worsen symptoms): If CoQ10 worsens → Complex III is blocked — electrons back up, producing superoxide (Pattern 5, Side Effects as Diagnostic Probes). If carnitine worsens → β-oxidation overload produces lipid peroxides in glutathione-depleted cells (Pattern 5). Worsening patterns localize WHICH ETC complex or FAO step is the bottleneck.
Consequence: The AIMM cascade traces the full ischemia→ion→mitochondrial damage pathway in four distinguishable steps: vascular trigger (pyridostigmine, midodrine, compression), NHE1-mediated Na⁺ overload (MDC002, amiloride, ranolazine), NCX1-mediated Ca²⁺ overload (magnesium, dantrolene), and mitophagy failure (rapamycin, urolithin A, NAC+CoQ10). Each step has a specific probe; the pattern of which probe works localizes how far down the cascade the lesion has progressed. A patient who responds to pyridostigmine (delivery) but not rapamycin (mitophagy) has early-stage AIMM — the vascular trigger is rate-limiting and mitochondrial damage is reversible. A patient who responds to rapamycin but not pyridostigmine has late-stage AIMM — mitochondrial damage has accumulated beyond what delivery improvement can compensate for, and mitophagy restoration is needed to clear the backlog. The PEM phenotype in AIMM: if pyridostigmine reduces PEM duration by 50% but rapamycin reduces it by 80% → both delivery AND clearance are rate-limiting, with clearance being the dominant bottleneck. The AIMM cascade is the mechanistic bridge between the vascular hypothesis and the mitochondrial hypothesis — it explains how impaired blood flow becomes impaired mitochondrial function through a specific ionic cascade. Origin: mechanistic-pathway-tracing.
4 NAD⁺ Depletion / Metabolic Trap
Certainty: 0.40. Tryptophan→NAD⁺ pathway is established biochemistry — tryptophan is converted through the kynurenine pathway to quinolinic acid, then to nicotinic acid mononucleotide (NaMN) → NAD⁺. IDO-driven tryptophan diversion (documented in ME/CFS) simultaneously depletes serotonin (tryptophan→5-HT pathway) AND NAD⁺ (tryptophan→kynurenine→NAD⁺ pathway — the kynurenine diversion prevents both). Additionally, NAD⁺ is consumed by: (a) PARP (DNA repair enzyme) — activated by oxidative DNA damage → depletes NAD⁺; (b) CD38 (NAD⁺ glycohydrolase on immune cells) — upregulated by inflammatory cytokines → degrades NAD⁺; (c) sirtuins (NAD⁺-dependent deacetylases) — consume NAD⁺ for epigenetic regulation. The combination of reduced synthesis (tryptophan diversion) and increased consumption (PARP, CD38, sirtuins) → severe NAD⁺ depletion → NAD⁺/NADH ratio collapse → impaired mitochondrial Complex I (NADH→NAD⁺), impaired sirtuin activity (SIRT1, SIRT3), impaired PARP-mediated DNA repair, and impaired glycolysis (GAPDH requires NAD⁺). NMN/NR supplementation has not been studied in ME/CFS controlled trials.
4.1 Cascade: IDO → kynurenine diversion → NAD⁺ synthesis defeat + consumption surge
Cascade:
- IDO/TDO activation (IFN-γ, TNF-α, IL-1β) → tryptophan → kynurenine diversion
- Reduced quinolinic acid → reduced NAD⁺ synthesis (de novo pathway)
- PARP activation (oxidative DNA damage) → increased NAD⁺ consumption
- CD38 upregulation → increased NAD⁺ degradation
- Chronic sirtuin activation → continuous NAD⁺ consumption → NAD⁺/NADH ratio decline
- Sirtuin inhibition (paradoxical), PARP overactivation (parthanatos)
- Impaired Complex I, impaired glycolysis (GAPDH requires NAD⁺) → multi-level metabolic failure
- fatigue, neuroinflammation, impaired DNA repair → accelerated aging, epigenetic dysregulation
4.2 Step G1: IDO/TDO activation (synthesis failure)
Probes: Same as kynurenine cascade Step H1. LDN reduces IDO induction; 1-MT inhibits IDO directly; anti-inflammatory combination suppresses multiple IDO-inducing pathways. The distinction in THIS cascade: NAD⁺ synthesis failure vs. serotonin depletion. Both result from the same IDO activation but have different drug targets.
If LDN restores NAD⁺ levels (measurable in serum/whole blood): TLR4-mediated IDO induction is the dominant mechanism of NAD⁺ depletion — LDN reduces IDO → restores tryptophan availability for NAD⁺ synthesis. This should be correlated with kynurenine/tryptophan ratio.
If LDN does NOT restore NAD⁺ but NMN/NR DOES: NAD⁺ depletion is from tryptophan diversion (LDN reduces IDO but NAD⁺ synthesis capacity from tryptophan is permanently impaired due to enzyme deficiencies in the kynurenine→NAD⁺ pathway — 3-hydroxyanthranilate 3-4-dioxygenase or quinolinate phosphoribosyltransferase deficiency). NMN/NR bypasses the de novo pathway entirely by providing the salvage pathway substrate directly.
4.3 Step G2: NAD⁺ synthesis vs. consumption — which is the dominant mechanism?
Probes for synthesis:
- NMN (nicotinamide mononucleotide, 250–1000 mg/day) or NR (nicotinamide riboside, 300–1000 mg/day): NAD⁺ precursors that enter the salvage pathway (NR → NMN → NAD⁺), bypassing the tryptophan de novo pathway entirely. If NMN/NR improves fatigue, cognition, and exercise tolerance → NAD⁺ deficiency from tryptophan diversion is rate-limiting. The salvage pathway is functional (NR → NMN → NAD⁺ via NMNAT enzymes) — the bottleneck is substrate availability, not enzyme function.
- Niacin (nicotinic acid, 500–2000 mg/day, flush form): NAD⁺ precursor through the Preiss-Handler pathway (different from salvage). Causes prostaglandin-mediated flushing (avoid in MCAS unless pre-treated with aspirin). If niacin works where NMN/NR does NOT → the salvage pathway enzyme (NMNAT) is deficient, but the Preiss-Handler pathway is intact. Niacin flushing: if severe → mast cell hyper-reactivity (Pattern 3, Side Effects as Diagnostic Probes). If absent → prostaglandin system is hyporesponsive (consistent with MCAS desensitization).
- Nicotinamide (non-flush niacinamide, 500–1500 mg): NAD⁺ precursor without flushing — inhibits sirtuins and PARP at high doses. If nicotinamide improves symptoms → NAD⁺ deficiency AND sirtuin/PARP overactivation are both rate-limiting (nicotinamide provides NAD⁺ precursor AND inhibits NAD⁺ consumers). If NMN works but nicotinamide does NOT → the NAD⁺ deficiency is the bottleneck, and sirtuin/PARP inhibition is unnecessary or harmful.
Probes for consumption:
NAD⁺ is NOT rate-limiting, or consumption is so high that precursor supplementation cannot keep up — distinguish between PARP overactivation, CD38 overactivation, sirtuin overactivation, or downstream ETC failure per sub-probes below
- Certainty
- Low to Medium
- Level of action
- Partial root cause
- PARP overactivation: If NMN/NR produces no benefit but the patient has evidence of oxidative DNA damage (elevated 8-OHdG in urine) → PARP is consuming NAD⁺ faster than supplementation can provide. PARP inhibitors are not clinically available for ME/CFS (olaparib, niraparib are oncology drugs with significant toxicity).
- CD38 overactivation: If NMN/NR produces no benefit but serum CD38 is elevated (measurable) → CD38 is degrading NAD⁺. CD38 inhibitors (apigenin, quercetin, luteolin — flavonoids) reduce CD38 activity. If apigenin + NMN/NR works where NMN/NR alone does NOT → CD38-mediated degradation is the dominant NAD⁺ sink.
- Sirtuin overactivation: If NMN/NR provides initial benefit (first 2 weeks) but then stops (tachyphylaxis) → sirtuins are consuming the supplemented NAD⁺ as quickly as it is provided. The NAD⁺ deficiency is sirtuin-consumption-driven. Resveratrol (sirtuin activator) would worsen symptoms; nicotinamide (sirtuin inhibitor) may help by reducing consumption.
- Downstream ETC failure: If NMN/NR produces no benefit AND CD38 is normal AND PARP markers are low → NAD⁺ is not the bottleneck. The metabolic block is downstream of NAD⁺ — ETC Complex I is dysfunctional despite adequate NAD⁺ substrate.
4.4 Step G3: Sirtuin, PARP, and CD38 — therapeutic targets or diagnostic probes?
- Resveratrol (150–500 mg, sirtuin activator): SIRT1 and SIRT3 require NAD⁺. If resveratrol improves mitochondrial function → sirtuin activation promotes mitochondrial biogenesis (SIRT1→PGC-1α) and ETC efficiency (SIRT3 deacetylates Complex I and II). If resveratrol worsens fatigue → sirtuins are already consuming NAD⁺ at maximal rate, and further activation depletes NAD⁺ below critical threshold.
- Apigenin (50–100 mg, CD38 inhibitor, dietary flavonoid): Reduces CD38-mediated NAD⁺ degradation. If apigenin + NMN/NR increases NAD⁺ levels more than NMN/NR alone → CD38 was degrading the supplemented precursor before it could raise NAD⁺.
- NAD⁺/NADH ratio measurement: This is measurable in clinical labs (LC-MS/MS from whole blood). The ratio should guide supplementation: if the ratio is low → supplementation is indicated; if the ratio is normal despite symptoms → NAD⁺ is not the bottleneck, and supplementation is addressing the wrong node. Measure pre- and post-NMN/NR supplementation — if NAD⁺ rises but symptoms do not improve → NAD⁺ is not rate-limiting; the metabolic defect is downstream of NAD⁺ availability.
Consequence: The NAD⁺ cascade is the most biochemically measurable energy cascade in the chapter — NAD⁺/NADH ratio, kynurenine/tryptophan ratio, CD38 levels, and 8-OHdG (oxidative DNA damage) are all clinical-lab measurable. The diagnostic sequence: (1) measure baseline NAD⁺ and kynurenine/tryptophan ratio; (2) NMN/NR trial → NAD⁺ rise = synthesis pathway functional; NAD⁺ rise without symptomatic improvement = consumption rate equals supplementation rate or NAD⁺ is not rate-limiting; (3) If NMN/NR null → check CD38 and PARP markers → treat the dominant consumer (CD38 inhibitor, antioxidant for PARP); (4) If NMN/NR + CD38 inhibitor produces NAD⁺ rise but no symptomatic benefit → NAD⁺ is not the metabolic bottleneck — the block is in ETC Complex I or downstream ATP synthesis. This cascade benefits most from serum biomarker correlation — the biochemical measurements validate the pharmocodiagnostic logic. Origin: mechanistic-pathway-tracing.
5 Metabolic Trap: IDO-Kynurenine Pathway
Certainty: 0.45. IDO-driven kynurenine pathway dysregulation documented in ME/CFS (elevated kynurenine/tryptophan ratio, elevated quinolinic acid in CSF, reduced serotonin in CSF). The cascade has three pharmacologically distinct arms: serotonin depletion, quinolinic acid-mediated NMDA excitotoxicity, and kynurenic acid-mediated NMDA antagonism. Each arm produces different symptoms and responds to different drugs. The diagnostic task is to determine which arm(s) are dominant in a given patient.
5.1 Cascade: IDO → kynurenine → serotonin depletion + QUIN/KYNA imbalance
Cascade:
- Immune activation (IFN-γ, TNF-α) → IDO upregulation in microglia, macrophages, dendritic cells
- Tryptophan → kynurenine (bypassing serotonin and NAD⁺ synthesis)
- Serotonin depletion: ~95% tryptophan diverted → reduced 5-HT → depression, anxiety, insomnia, pain
- Quinolinic acid: KMO in microglia → 3-HK → 3-HAA → QUIN → NMDA excitotoxicity → cognitive dysfunction
- Kynurenic acid: KAT in astrocytes → KYNA → NMDA antagonism → cognitive slowing, reduced plasticity
- QUIN/KYNA balance determines excitotoxicity vs. suppression phenotype
5.3 Step H2a: Serotonin depletion (the 5-HT arm)
Mechanism: With ~95% of tryptophan diverted to kynurenine, serotonin synthesis in raphe neurons is starved of precursor. Tryptophan hydroxylase 2 (TPH2, the brain-specific isoform) converts tryptophan to 5-HTP. If tryptophan levels in CSF are low (documented in ME/CFS), serotonin synthesis drops proportionally. Consequences: reduced serotonin → (a) impaired descending pain inhibition in the periaqueductal gray and rostroventral medulla → widespread pain amplification; (b) reduced 5-HT1A and 5-HT2A signaling in prefrontal cortex and limbic system → depression, anxiety; (c) reduced 5-HT-mediated regulation of sleep-wake cycle (raphe projections to suprachiasmatic nucleus and thalamus → insomnia, fragmented sleep).
Probes for serotonin depletion:
- 5-HTP (50–200 mg): Direct serotonin precursor — bypasses the tryptophan bottleneck by providing the product of TPH directly. If 5-HTP improves mood, sleep, and pain → serotonin depletion from tryptophan diversion is present and rate-limiting. The positive response confirms that TPH2 is functional (can convert 5-HTP to serotonin) but starved of substrate (low tryptophan). Onset latency: improvement within days → serotonin depletion was acute and 5-HTP rapidly replenishes the depleted pool.
- Tryptophan supplementation (500–2000 mg): Competing with IDO for substrate — tryptophan can be converted to serotonin (via TPH) or kynurenine (via IDO). If tryptophan improves symptoms → IDO is NOT saturated — increased substrate overcomes the diversion. If tryptophan worsens symptoms → IDO is saturated and additional tryptophan is preferentially converted to kynurenine → more quinolinic acid production → worsened excitotoxicity. The tryptophan response direction is a diagnostic probe: improvement → IDO is the bottleneck but not saturated; worsening → IDO is saturated and kynurenine pathway is harmful. CAUTION: tryptophan supplementation in a patient with active kynurenine pathway may increase neurotoxic metabolite production.
- Tryptophan vs. 5-HTP differential: If 5-HTP works but tryptophan does NOT work (or worsens) → TPH2 is functional (converts 5-HTP → serotonin) but tryptophan is preferentially shunted to kynurenine (IDO is saturated). This pattern specifically confirms IDO saturation.
- SSRIs (fluoxetine, sertraline, escitalopram): Block serotonin reuptake → increase synaptic 5-HT. If SSRIs improve mood and pain → serotonin reuptake was excessive relative to the reduced serotonin pool, and blocking reuptake increases synaptic 5-HT sufficiently. If SSRIs produce no improvement → the serotonin pool is so depleted that blocking reuptake adds nothing (there is nothing being released to reuptake). If SSRIs worsen symptoms → serotonin syndrome (rare) or the serotonin increase triggers compensatory downregulation of postsynaptic 5-HT receptors. WARNING: SSRIs may worsen POTS hemodynamically (Mar et al. 2014) — the serotonergic system modulates sympathetic outflow, and increasing 5-HT can exacerbate tachycardia in POTS patients. Start at minimal dose; monitor HR and orthostatic tolerance.
- Buspirone (5-HT1A partial agonist, 5–15 mg TID): Directly stimulates 5-HT1A receptors (anxiolytic, antidepressant) and 5-HT1A autoreceptors on raphe neurons (reducing serotonin release — paradoxically anxiolytic). If buspirone works → 5-HT1A receptor signaling is intact but insufficiently activated by endogenous serotonin. More specific than SSRIs — targets the receptor, not the reuptake transporter.
5.4 Step H2b: Quinolinic acid → NMDA excitotoxicity (the glutamate arm)
Mechanism: QUIN is produced in microglia (KMO is predominantly microglial) and acts as an NMDA receptor agonist at the glycine co-agonist site. Unlike glutamate (which is rapidly cleared by astrocytes), QUIN has a long half-life in the extracellular space. Chronic QUIN elevation → sustained low-grade NMDA receptor activation → (a) excitotoxicity — excessive Ca²⁺ influx through NMDA receptors → mitochondrial Ca²⁺ overload, oxidative stress, and synaptic dysfunction; (b) reduced long-term potentiation (LTP) — the NMDA receptor is desensitized from chronic low-level activation, and the dynamic range for LTP (learning, memory) is reduced → cognitive dysfunction; (c) selective vulnerability of GABAergic interneurons (which express high levels of GluN2B-containing NMDA receptors that QUIN preferentially activates) → reduced inhibitory tone → network hyperexcitability → anxiety, hyperarousal, sensory sensitivity, tinnitus.
Probes for NMDA excitotoxicity:
- Memantine (5–20 mg): Uncompetitive NMDA receptor antagonist — blocks the channel only when it is open (use-dependent). Preferentially blocks pathological tonic NMDA activation while sparing physiological phasic activation. If memantine improves cognition, reduces anxiety, and reduces sensory hypersensitivity → NMDA excitotoxicity is present and rate-limiting. Onset latency: improvement within days → functional NMDA overactivation (QUIN is present and active). Improvement at 2–4 weeks → the NMDA system is adapting to reduced activation (receptor trafficking, CREB signaling). Dose titration: if sedation/brain fog appears at low dose (5 mg) → basal glutamate tone is already low (Pattern 4, Side Effects as Diagnostic Probes) — reducing NMDA further tips cognition over a cliff. This patient has kynurenic acid dominance, not quinolinic acid excess. If memantine can be titrated to 20 mg without sedation → QUIN-driven excitotoxicity is dominant and NMDA blockade is well-tolerated.
- Low-dose amitriptyline (5–10 mg at bedtime): NMDA antagonist (weak, at low doses) + NE/5-HT reuptake inhibition + H1 antagonism (sleep). If amitriptyline improves sleep AND cognition → dual mechanism: NMDA antagonism reduces nocturnal excitotoxicity + H1 blockade consolidates sleep. The cognitive improvement may be secondary to improved glymphatic clearance during sleep.
- Ketamine (0.1–0.5 mg/kg IV, investigational): NMDA antagonist (channel blocker, different site from memantine) + rapid antidepressant effect (mTORC1-dependent synaptogenesis). If ketamine produces rapid improvement (within hours) → NMDA antagonism is the mechanism (fast), and the synapse is capable of rapid plasticity. If improvement lasts days to weeks after a single infusion → ketamine has triggered synaptogenesis (new spine formation in PFC), not just blocked NMDA. If ketamine produces transient worsening (dissociation, anxiety) but no subsequent improvement → the NMDA system is too fragile for direct blockade. CAUTION: ketamine increases HR and BP — may worsen hyperadrenergic POTS.
- Riluzole (50 mg BID, glutamate release inhibitor): Reduces presynaptic glutamate release by blocking voltage-gated Na⁺ channels. If riluzole works → the glutamatergic excess is from excessive release, not just from QUIN agonism. If riluzole does NOT work but memantine works → the glutamatergic problem is specifically NMDA receptor-level (QUIN agonism), not excessive glutamate release.
- NAC (600–1200 mg BID): Cysteine donor for glutathione synthesis → reduces oxidative stress from NMDA-mediated Ca²⁺ influx. If NAC + memantine is synergistic (greater than additive, Diurnal Response Window as Circadian Pharmacodiagnostic Probe) → NMDA excitotoxicity drives ROS production, and blocking NMDA (memantine) + scavenging ROS (NAC) closes the excitotoxicity→ROS→mitochondrial damage→ATP deficit→impaired Ca²⁺ clearance→more NMDA activation loop.
5.5 Step H2c: Kynurenic acid vs. quinolinic acid imbalance (the astrocyte arm)
Mechanism: KYNA is produced in astrocytes (KAT is predominantly astrocytic) and acts as an endogenous NMDA antagonist at the glycine site — competing with QUIN at the same binding site. KYNA is broadly neuroprotective (reduces excitotoxicity) but also suppresses glutamatergic transmission → cognitive slowing. The QUIN/KYNA ratio determines the net glutamatergic effect: high QUIN/low KYNA → excitotoxicity; low QUIN/high KYNA → cognitive suppression; both high → competing effects (excitotoxicity from QUIN at NMDA receptors that are partially blocked by KYNA → unpredictable net effect). In ME/CFS, the QUIN/KYNA ratio may be shifted in either direction in different patients.
Probes for KYNA/QUIN imbalance:
- No direct KYNA modulator exists clinically. KYNA is a tryptophan metabolite — reducing kynurenine production (IDO inhibition, LDN) reduces both QUIN and KYNA proportionally. KMO inhibitors (which block QUIN synthesis and shunt kynurenine to KYNA) exist in preclinical development but are not available.
- Memantine vs. LDN head-to-head: If memantine works better than LDN → QUIN excitotoxicity dominates over microglial neuroinflammation. NMDA blockade is more specific than anti-inflammatory suppression. If LDN works better than memantine → microglial neuroinflammation dominates (LDN reduces both IDO induction AND cytokine-mediated symptoms). Memantine may still provide benefit but it is downstream of the primary pathology.
- Memantine + LDN combination: If the combination is synergistic → QUIN excitotoxicity AND neuroinflammation are independent co-drivers (LDN reduces QUIN production by reducing IDO, memantine blocks existing QUIN at the receptor). If the combination is additive only → LDN reduces QUIN production, which reduces NMDA activation → memantine has less to block → the drugs are operating in series on the same pathway.
- Pregnenolone (50–200 mg, neurosteroid): Pregnenolone is a negative allosteric modulator of NMDA receptors (different site from QUIN/KYNA) and a negative allosteric modulator of GABA-A receptors. If pregnenolone improves cognition and reduces anxiety → the NMDA system is hyperactive (consistent with QUIN dominance). If pregnenolone causes sedation → the NMDA negative modulatory effect dominates, consistent with QUIN-driven excitotoxicity being suppressed. If pregnenolone causes agitation → GABA-A negative modulation is tipping excitation-inhibition balance in a system with low GABAergic reserve.
KYNA-dominant kynurenine pathway — basal glutamate tone is already low from KYNA antagonism, and further NMDA blockade reduces glutamate below the threshold for normal cognition; LDN’s benefit is from reducing neuroinflammation, not from restoring glutamatergic balance — patient should avoid all NMDA antagonists
- Certainty
- Low to Medium
- Level of action
- Partial root cause
Consequence: The kynurenine cascade is uniquely diagnostic because it has three arms producing diametrically opposite effects: serotonin depletion (treat with 5-HTP), NMDA excitotoxicity (treat with memantine), and NMDA suppression (avoid NMDA antagonists, treat upstream with LDN). The differential diagnosis: 5-HTP response localizes the lesion to the serotonin arm. Memantine response at standard doses without sedation localizes the lesion to the QUIN arm. Memantine sedation at low dose localizes the lesion to the KYNA arm (basal glutamate tone low). The combination of 5-HTP + memantine responses maps whether both arms are active simultaneously. LDN response tells you IDO is driving the pathway. The tryptophan challenge test (improvement vs. worsening after tryptophan supplementation) tells you whether IDO is saturated — a rapid, low-cost probe that no one performs but that directly informs the entire kynurenine diagnostic logic. This is the most biochemically accessible cascade in the chapter: serum kynurenine/tryptophan ratio and QUIN/KYNA ratio are measurable in clinical labs, unlike ion channel function or autoantibody bioassays. The cascade tracing should be correlated with serum metabolite measurements whenever possible — the pharmocodiagnostic logic and the biochemical measurements validate each other. Origin: mechanistic-pathway-tracing.
No direct ME/CFS trial demonstrates that tryptophan or 5-HTP supplementation changes any outcome; that negative fact is firm. But whether a supplement could act is not a single question — it splits by target compartment, and the two compounds differ. Tryptophan is the weaker candidate on mechanistic grounds: in normal physiology roughly 95% of dietary tryptophan enters kynurenine catabolism rather than serotonin synthesis (tryptophan normal), and chronic inflammation raises that fraction by inducing IDO/TDO (elevated kynurenine/tryptophan ratio, reduced serotonin; Section medication reference) — so oral tryptophan largely feeds the diverting pathway it is meant to bypass, and can worsen brain fog via quinolinic acid in IDO-saturated patients. 5-HTP is not an IDO substrate (Section Theanine Glutamate Analog + GABAergic, ch29), so it escapes that specific diversion; its limitation is different and location-specific. Oral 5-HTP is predominantly decarboxylated peripherally to serotonin, which does not cross the blood–brain barrier — so for a central serotonin deficit (reduced CSF serotonin is documented at the opening of this section, certainty 0.45) oral 5-HTP is a route-of-delivery failure.
That central failure does not, however, extinguish the peripheral hypothesis. The chapter’s own integrative model treats peripheral serotonin — synthesized by gut enterochromaffin cells and signaling through vagal afferents — as a separate, candidate multi-system lesion (Peripheral Serotonin Depletion as Multi-System Convergence Point, cert 0.50; Peripheral Serotonin Depletion: Convergent Evidence, Convergent Caution). Under that model, 5-HTP acting peripherally is on-target rather than a miss, and 5-HTP is established to raise serotonin levels in clinical settings (Section Theanine Glutamate Analog + GABAergic, ch29); what is untested is whether this improves ME/CFS outcomes. The evidence therefore supports a differentiated conclusion: no supplement is demonstrated to benefit ME/CFS; tryptophan’s rationale is undermined by diversion and it carries a worsening risk; 5-HTP’s central rationale is undermined by the blood–brain barrier, but its peripheral (gut–vagal) effect is a plausible and unvalidated hypothesis — a genuine open question, not a disproof. Absence of RCT evidence means benefit is not established, not that none is possible; per-patient flux control (Section LOX-Mediated Collagen Stabilization, ch29) and the location of the serotonin deficit remain unresolved, and wholesale suppression of the kynurenine pathway could even worsen the NAD+ deficit the mitochondrial model describes (Section Tier 2: Amplifier Mechanisms, ch22).
The honest value of the serotonin arm is therefore diagnostic hypothesis-testing rather than treatment, and even that is a candidate probe: a controlled, clinician-supervised challenge that distinguishes tryptophan-worsening (consistent with IDO saturation) from tolerably-null-tryptophan-with-5-HTP-improvement (consistent with a tryptophan-hydroxylase bottleneck) may help localize which mechanism dominates in an individual, but the readout is unblinded, unvalidated, and confounded by placebo and by the differing bioavailability of the two compounds, so a positive response cannot establish that serotonin depletion is the cause of the illness.
Safety: 5-HTP must not be combined with serotonergic agents — SSRIs, SNRIs, MAOIs, TCAs, mirtazapine, tramadol, triptans, ondansetron, linezolid, or other drugs that raise serotonin — because of serotonin-syndrome risk; this is a real, potentially life-threatening interaction requiring medical review. Tryptophan supplementation that causes worsening must be stopped. No dosing, safety, or efficacy data exist in ME/CFS; nothing here is a treatment recommendation or a basis for self-treatment.
5.6 Cascade: Functional Iron Deficiency → Fe-S Cluster Failure → ETC Complex Impairment → ATP Depletion
Cascade:
- Chronic inflammation → IL-6/STAT3 signalling (established in ME/CFS, cert 0.50)
- Hepcidin paradox: hepcidin is LOW (~50% reduction, (Kavyani et al. 2024)) — NOT high as classic inflammatory anemia predicts
- Despite low hepcidin, functional iron deficiency persists: high ferritin, low serum iron, low TSAT ((Świątczak et al. 2022), (Baklund et al. 2021))
- Four candidate hepcidin-independent trapping mechanisms (hepcidin independent ferroportin blockade, cert 0.40):
- ferroportin trafficking defect via IRP/IRE dysregulation,
- ceruloplasmin/ferroxidase deficiency,
- NCOA4 ferritinophagy blockade,
- LCN2/NGAL futile iron scavenging cycle
- Trapped macrophage iron → serum iron restriction → five downstream arms:
- Mitochondrial Fe-S cluster deficiency → Complex I (8 Fe-S clusters), Complex II (3), Complex III (1 Rieske) → impaired ETC → reduced ATP yield → fatigue/PEM
- Aconitase (Fe-S enzyme) inactivation → Krebs cycle bottleneck → reduced NADH → further ATP deficit + citrate accumulation
- Cytochrome c oxidase (heme a/a₃, Complex IV) impairment → terminal electron transport failure → O₂ not reduced to H₂O → reactive oxygen species leak
- Tyrosine hydroxylase (iron cofactor) → reduced dopamine synthesis → low CSF dopamine (documented, (Walitt et al. 2024))
- TPO (thyroid peroxidase, heme-iron) → reduced T4 synthesis → low T3 syndrome (Hepcidin-Inflammation Axis as Endocrine-Immune Bridge, cert 0.40)
- Net convergence: multi-system ATP depletion + dopamine deficiency + thyroid dysfunction → fatigue, PEM, cognitive impairment, orthostatic intolerance
- Temporal trajectory: acute hypoferremia (Phase 0) → stress erythropoiesis + monocyte iron loading (Phase I, (Hanson et al. 2024)) → established functional iron deficiency + low hepcidin (Phase II–III, temporal phase shift iron, cert 0.35)
Cross-disease divergence: In Long COVID, the iron phenotype differs — monocyte iron loading + lymphocyte iron starvation (single-cell, (Hanson et al. 2024)), RBC O₂-binding defect ((Kronstein-Wiedemann et al. 2024)), and ferritin variability (35% hyperferritinaemic, 24% iron deficient, 9% anaemic, (Sonnweber et al. 2022)). The therapeutic bifurcation is at the Fe-S cluster level: ME/CFS needs iron redistribution (intracellular iron is trapped), LC needs iron phenotyping (some patients need repletion, some need chelation — iron redox polarity diagnostic bifurcation, cert 0.55).
Drug interception nodes (from mechanism → drug, tracing the cascade forward):
- Node 1 — Serum iron / TSAT (pre-ETC): Oral iron works only in genuine deficiency (ferritin <30 µg/L or TSAT <16%). In functional deficiency, oral iron fails (ferroportin blocked — see hepcidin independent ferroportin blockade) and may worsen ferroptosis risk. IV iron (iron sucrose, ferric carboxymaltose) may bypass the enteral ferroportin block in some patients by delivering iron directly to transferrin — but this is untested in ME/CFS. If IV iron works: the bottleneck is at the enteral iron absorption step (ferroportin blockade at the enterocyte, not the macrophage). If IV iron fails: the bottleneck is downstream — macrophage iron trapping, Fe-S cluster assembly failure, or ceruloplasmin deficiency preventing iron loading onto transferrin even when iron is IV-delivered. Certainty of inference: 0.30 (IV iron bypass mechanism is mechanistically sound but zero ME/CFS data).
- Node 2 — Fe-S cluster assembly (ETC level): CoQ10 (ubiquinone/ubiquinol) bypasses Complex I and II Fe-S deficiencies by accepting electrons directly from Complex II and ETF dehydrogenase. L-carnitine shuttles fatty acids into mitochondria, bypassing the glycolytic-ETC uncoupling. Both are already in the ME/CFS armamentarium. If CoQ10 works: supports Fe-S cluster failure as rate-limiting — CoQ10’s electron bypass compensates for Fe-S-electrodeficient ETC complexes. If CoQ10 fails: the bottleneck may be at Complex IV (heme-dependent, not Fe-S-cluster-dependent) or downstream at ATP synthase. Certainty: 0.35 (CoQ10 benefit is consistent with Fe-S cluster failure but is not specific — CoQ10 also helps PDH inhibition, NOS uncoupling, and generic mitochondrial support).
- Node 3 — Macrophage iron trapping (pharmacological chelation): Deferiprone (oral, membrane-permeable) chelates intracellular labile iron in macrophages, bypassing the entire hepcidin-ferroportin axis. If deferiprone works: confirms that trapped macrophage iron is the rate-limiting resource — the iron is present but inaccessible, and direct chelation-based removal improves iron availability by reducing the futile LCN2 cycle and freeing NCOA4-ferritinophagy capacity. If deferiprone fails: the bottleneck is NOT macrophage iron trapping — it may be at Fe-S cluster assembly (ISA/ISC machinery defective, iron delivered but cannot be inserted), ceruloplasmin deficiency (iron exported but not loaded onto transferrin), or a non-iron mechanism (NAD depletion, PDH inhibition). Certainty: 0.25 (zero deferiprone data in ME/CFS; black-box agranulocytosis warning; deferiprone functional iron deficiency).
- Node 4 — Ferroptosis (oxidative stress endpoint): NAC (N-acetylcysteine, glutathione precursor), α-lipoic acid (mitochondrial antioxidant), and vitamin E (α-tocopherol, chain-breaking lipophilic antioxidant) each attenuate lipid peroxidation by different mechanisms. If anti-ferroptosis agents work: supports ferroptosis as a contributing mechanism — reducing the labile iron pool’s catalytic capacity (via glutathione) or scavenging lipid peroxyl radicals (via vitamin E) improves outcomes. If they fail: ferroptosis is not a major contributor, or the ferroptosis is iron-independent (e.g., driven by glutathione depletion alone, not catalytic iron). Certainty: 0.30 (anti-ferroptosis agents have multiple mechanisms; benefit is not specific to the iron-ferroptosis link).
Discriminating probe: Deferiprone vs IV iron is the cleanest — deferiprone removes trapped iron (confirms trapping), IV iron bypasses the enteral block (confirms ferroportin blockade at enterocyte). If deferiprone works and IV iron does not → macrophage trapping, not enterocyte blockade. If IV iron works and deferiprone does not → enteral ferroportin blockade, not macrophage trapping. If neither works → bottleneck is downstream (Fe-S assembly, ceruloplasmin, or non-iron). Quality gate: no clean discriminator exists — both drugs have uncertain specificity. Deferiprone chelates zinc as well as iron; its benefit could be zinc-mediated. IV iron’s failure could reflect ceruloplasmin deficiency preventing iron loading onto transferrin — not ferroportin blockade at all.
Certainty: 0.35. The cascade is mechanistically coherent at every step but each node has an alternative explanation. The hepcidin paradox is the strongest empirical anchor — Kavyani2023 is the only measurement and needs independent replication. The cascade below the serum iron level (Fe-S clusters, aconitase, cytochrome c oxidase) is physiological reasoning applied to ME/CFS — none of these nodes have been directly measured in ME/CFS tissue.