Prodromal Detection and Prevention Research Program
The retrospective prodromal signs described in:prodromal-recognition generate a research programme aimed at prospective validation, early detection, and preventive intervention.
1 Study 1: The Sibling Stress Test β Prospective Metabolic Phenotyping in First-Degree Relatives
Design: Prospective cohort. N=200 asymptomatic siblings of confirmed ME/CFS patients (ages 10β25). Baseline and annual assessments:
- Cardiopulmonary exercise testing (CPET) with 2-day repeat protocol
- 24-hour HRV analysis (Holter)
- PBMC respirometry (Seahorse: spare respiratory capacity)
- Quantitative sensory testing (small fiber function)
- Home polysomnography (sleep architecture)
- Salivary cortisol diurnal curve
- Serum ferritin, BH4 metabolites, methylation panel Primary question: Do siblings show a gradient of subclinical abnormalities correlating with number of prodromal signs (Retrospective Prodromal Syndrome as Subclinical Metabolic Reserve Depletion)? Can a pre-symptomatic metabolic signature predict ME/CFS development within 5 years? Falsification: If siblings show no subclinical metabolic differences from unrelated controls despite sharing genetic background, the genetic-ceiling model of Architecture C is not supported. Power calculation: N=200 siblings (assuming 15% conversion rate over 5 years) provides 80% power to detect an OR of 3.0 for metabolic signature vs. conversion, at \(\alpha = 0.05\).
2 Study 2: Wearable Transition Detection β Machine Learning on Pre-Diagnostic Physiological Streams
Design: Retrospective analysis of consumer wearable data (Garmin, Apple Watch, Oura Ring) from ME/CFS patients who were wearing devices before diagnosis. N=1000+ feasible via patient registries and ME/CFS biobanks. Features extracted: Resting HR trend, HRV trajectory (RMSSD), sleep duration and efficiency, step count variability, recovery time from exercise, respiratory rate during sleep. Primary question: Is the prodromal-to-disease transition abrupt (supporting phase-transition model,:cognitive-cliff) or gradual (supporting linear decline model)? Can machine learning identify the inflection point? Falsification: If no detectable signal precedes diagnosis by >3 months, the prodromal decompensation model has no wearable signature. If the transition is abrupt (within 2 weeks), this supports the hysteretic phase-transition model; if gradual (>6 months decline), this supports linear reserve erosion.
3 Study 3: 2-Day CPET Subclinical PEM β Prodromal Biomarker Validation
Design: Cross-sectional comparison with longitudinal follow-up. Three groups (N=50 each): (a) at-risk individuals (prodromal composite score >8, see below), (b) healthy controls matched for age/sex/activity, (c) established ME/CFS patients. All undergo 2-day CPET protocol. Hypothesis: Prodromal individuals will show a Day 2 VO2max decline of 5β10% β intermediate between controls (0%) and established ME/CFS (>15%). This βsubclinical PEMβ would be an objective, quantitative prodromal biomarker. Longitudinal component: Follow at-risk group for 3 years. Primary outcome: Does Day 2 decline magnitude predict ME/CFS conversion? Expected: >5\(\\times\) conversion rate for those with 5β10% decline vs. those with 0% decline.
4 Study 4: HRV Recovery Kinetics as Population-Scale Screening Tool
Design: Cohort study using existing wearable data. Define βHRV recovery ratioβ = (time to return to resting HRV baseline after standardised exercise) / (expected recovery time for that fitness level, derived from population norms). Ratio >1.5 = prodromal flag. Key advantage: Measurable with consumer wearables at population scale, enabling screening without clinical visits. Validation: In a cohort wearing HRV-capable devices for >12 months, test whether progressive lengthening of HRV recovery time in the 6β12 months preceding ME/CFS diagnosis is detectable. Critical discriminator: recovery ratio should distinguish pre-ME/CFS from simple deconditioning (deconditioned individuals improve recovery time with training; pre-ME/CFS individuals do not)
5 Study 5: Exercise Non-Responder Genomics
Design: Cross-reference existing exercise intervention GWAS data (HERITAGE, DREW, Gene SMART cohorts) with ME/CFS GWAS hits (DecodeME). In large exercise cohorts, approximately 20% are βnon-respondersβ to aerobic training despite adherence. Primary question: Do exercise non-responders and ME/CFS patients share genetic variants in mitochondrial adaptation pathways (PGC-1\(\alpha\) regulators, AMPK signalling, mitophagy genes)? Falsification: If non-responder genetics are entirely unrelated to ME/CFS susceptibility loci, the shared mitochondrial biogenesis failure hypothesis (Adaptation Debt β Training Without Supercompensation as Mitochondrial Biogenesis Failure) is not supported.
6 Proposed Prodromal Composite Score for Risk Stratification
A weighted clinical checklist for identifying at-risk individuals, pending validation:
| Sign | Weight | Rationale |
|---|---|---|
| Training:fitness ratio >3:1 (high volume, poor gains) | 2 | Strongest specific signal for failed adaptation |
| Late-game cognitive fade (duration-dependent) | 1 | Common, moderate specificity |
| Morning fatigue despite adequate sleep duration | 1 | Very common, low specificity alone |
| Fine action tremor (no neurological Dx) | 2 | Unusual in healthy youth, high specificity |
| Cold extremities + orthostatic symptoms | 1 | Common, moderate specificity |
| Post-exertional fog >24h recovery | 2 | Highly specific for energy deficit |
| Holiday/rest crashes (paradoxical) | 2 | Paradoxical pattern, high specificity |
| Family history ME/CFS or POTS | 2 | Genetic loading (Architecture C) |
| Neurodivergence (ADHD/ASD) | 1 | Architecture C risk factor |
| Hypermobility (Beighton β₯5) | 1 | Architecture C risk factor |
| Frequent minor infections (>6/year) | 1 | Immune surveillance deficit |
| Unrefreshing sleep (validated scale) | 1 | Prospectively validated |