Secondary Analysis Opportunities: Leveraging Existing Datasets
1 Rationale for Secondary Analysis
The multi-modal study described in Section Multi-Modal Testing of Selective Energy Dysfunction Hypothesis represents the definitive test of the Selective Energy Dysfunction Hypothesis but requires three years and approximately $2.15M to execute. A complementary near-term strategy leverages existing ME/CFS research datasets that already contain measurements relevant to the hypothesis’s core predictions. Secondary analyses can provide preliminary evidence, identify methodological challenges, and generate pilot data for funding applications within months rather than years, with no new data collection required for several key predictions.
The predicted gradient (controls > neurodivergent-only > ME/CFS-only > neurodivergent + ME/CFS) is the single most direct cellular test of Architecture C. If the gradient is absent — particularly if ADHD-only and ASD-only groups show identical spare respiratory capacity to controls — the constitutional metabolic deficit hypothesis (Neurodivergent Mitochondria: Constitutionally Lower Reserve Capacity) is falsified, and demand-side explanations for neurodivergent ME/CFS risk must be explored instead.
2 Prediction-to-Dataset Coverage
Four of the seven core predictions (brain hypometabolism, exercise demand-response failure, autonomic demand-response failure, and CNS coordination failure) can be tested through secondary analysis of existing datasets. Table PBMC Spare Respiratory Capacity Gradient Across Neurodivergent and ME/CFS Groups maps these predictions to available resources.
| Prediction | NIH 2024 | CPET DB | Autonomic | Brain PET | UK Biobank |
|---|---|---|---|---|---|
| Brain hypometabolism | \(checkmark\) | – | – | \(checkmark checkmark\) | Partial |
| Exercise demand-response | \(checkmark\) | \(checkmark checkmark\) | – | – | – |
| Resting muscle ATP preserved | – | – | – | – | – |
| ATP scaling failure | \(checkmark\) | \(checkmark checkmark\) | – | – | – |
| Autonomic demand-response | \(checkmark\) | – | \(checkmark checkmark\) | – | Partial |
| CNS coordination failure | \(checkmark checkmark\) | – | – | \(checkmark\) | – |
| Hair/nail growth normal | – | – | – | – | – |
\(checkmark checkmark\) = ideal dataset for this prediction;
\(checkmark\) = can test prediction; Partial = incomplete coverage; – = not testable. Three predictions (resting muscle ATP, electrical stimulation versus voluntary force, and hair/nail growth rate) require novel measurement protocols and are addressed exclusively in the multi-modal study.
3 Priority Datasets
3.1 Tier 1: Highest Feasibility
NIH Intramural ME/CFS Study (Walitt et al. 2024) This deep-phenotyping study enrolled 17 ME/CFS patients and 21 healthy controls, collecting comprehensive measurements including two-day cardiopulmonary exercise testing, brain fMRI, cerebrospinal fluid catecholamines and metabolites, autonomic function tests, motor cortex excitability via transcranial magnetic stimulation, and cognitive testing (Walitt et al. 2024). Published findings—reduced CSF norepinephrine and dopamine, reduced temporal-parietal junction activation, motor cortex hyperactivity, and Day 2 CPET functional decline—are broadly consistent with the Selective Energy Dysfunction framework (interpreted here as post-hoc alignment rather than a priori confirmation). A secondary analysis could systematically test the CNS-peripheral selectivity pattern: correlating catecholamine levels with symptom domains (CNS-dependent versus autonomous processes), comparing motor cortex excitability with peripheral force production, and examining demand-response patterns in the 2-day CPET data stratified by catecholamine subgroups. The full dataset may be accessible through NIH data-sharing mechanisms (e.g., dbGaP). Published Two-Day CPET Meta-Analysis Multiple research groups have published studies demonstrating significant Day 2 functional decline in ME/CFS using standardized 2-day CPET protocols (Keller et al. 2024) A systematic meta-analysis of published summary statistics would directly test the demand-response failure prediction: quantifying Day 2 decline magnitude, assessing heterogeneity by symptom domain (cognitive-predominant versus physical-predominant), and characterizing recovery time as a proxy for demand-response failure duration. This analysis requires no data-sharing agreements and can be initiated immediately using published data.
3.2 Tier 2: Medium Feasibility
Van Campen Autonomic Studies A series of studies by van Campen and colleagues characterizes cerebral blood flow via transcranial Doppler during orthostatic challenge in cohorts of 100–400 ME/CFS patients (Campen et al. 2020) Published findings show that 91% of patients with a normal heart rate and blood pressure response exhibit abnormal cardiac output and cerebral blood flow patterns during tilt (Campen et al. 2024), with 82% showing CBF abnormalities even when heart rate and blood pressure remain normal (Campen et al. 2020). This preserved-baseline, impaired-challenge pattern is directly consistent with the selective demand-response failure prediction. Individual patient trajectory data (baseline, tilt, recovery) could provide a granular test of the hypothesis’s autonomic component; accessing the full dataset would require international collaboration. UK Biobank ME/CFS Cohort The UK Biobank contains a substantial subcohort of self-reported ME/CFS cases with standardized measurements including wrist accelerometry, cognitive testing, blood biomarkers, and brain MRI in a subset. The large sample size enables detection of activity-pattern abnormalities, cognitive demand-response patterns, and CNS biomarker correlations, though the absence of clinically validated ME/CFS diagnoses limits interpretability. Access is available to approved researchers through an established application process.
3.3 Tier 3: Exploratory
Brain PET/SPECT Meta-Analysis Published neuroimaging studies report regional brain hypometabolism and neuroinflammation in ME/CFS, including evidence of neuroinflammation elevation of 45–199% above controls in the cingulate cortex, hippocampus, amygdala, thalamus, midbrain, and pons (Nakatomi et al. 2014). A meta-analysis of published regional findings could identify consistently affected brain regions across studies and characterize subtype patterns. Because raw imaging data is generally not publicly available due to regulatory constraints, this approach relies on published summary statistics rather than individual-level data.
4 Relationship to the Multi-Modal Study
Secondary analyses and the multi-modal study serve complementary roles. Secondary analyses can generate preliminary evidence on four of the seven predictions within 6–18 months, identify subtype patterns that should be accounted for in primary study design, and provide pilot data for grant applications at low cost. The multi-modal study remains the definitive experimental test of the hypothesis’s full prediction set, and is the only pathway to testing the three predictions that require novel measurement protocols.