Kerrebijn 2025 — the first robust FM GWAS
One-paragraph summary
Multi-ancestry GWAS meta-analysis of fibromyalgia across 11 cohorts totalling 2,563,755 individuals (54,629 FM cases vs. ~2.5M controls) — approximately a 100× sample-size step over the prior largest FM GWAS. Twenty-six genome-wide-significant risk loci identified. Top hits — HTT (Huntington's gene), GPR52 (an HTT regulator), CAMKV, DCC, DRD2/NCAM1, MDGA2, CELF4 — are all neural. Stratified-LDSC partitioned-heritability analysis shows FM heritability is exclusively enriched in brain tissues and neural cell types, with no enrichment in immune or musculoskeletal tissues. Strong genetic correlations (rg > 0.7) with low back pain, PTSD, and IBS. The paper therefore establishes FM at the population level as a CNS-anchored disorder. This is the load-bearing genetic anchor for the project's H3 chain (predictive-coding / network-dysregulation) — H1 (autoimmune-microbiome) and H2 (mast-cell convergent-inflammatory) are visible only as subset mechanisms diluted across the broader-population signal and do not surface in top loci.
Claims as triples
fibromyalgia_heritability — present_in → brain_tissue[evidence: Fig. partitioned-heritability LDSC; confidence: emerging]fibromyalgia_heritability — absent_in → immune_tissue[evidence: Fig. partitioned-heritability LDSC; confidence: emerging]fibromyalgia_heritability — absent_in → musculoskeletal_tissue[evidence: Fig. partitioned-heritability LDSC; confidence: emerging]predictive_coding_failure — bridges → fm_central_only[evidence: top GWAS loci all neural (HTT, GPR52, CAMKV, DCC, DRD2/NCAM1, MDGA2, CELF4); confidence: emerging]fm_central_only — correlates_with → ibs[evidence: rg > 0.7 with IBS; confidence: emerging]fm_central_only — correlates_with → ptsd[evidence: rg > 0.7 with PTSD; confidence: emerging]fm_central_only — correlates_with → low_back_pain[evidence: rg > 0.7 with low back pain; confidence: emerging]
Methods note
GWAS meta-analysis across 11 biobank-scale cohorts (UK Biobank, FinnGen, Million Veteran Program, and others). FM cases ascertained via ICD codes / self-report depending on cohort; trans-ancestry meta-analysis with European, African, and Asian-ancestry strata. Stratified LD-score regression for partitioned heritability across GTEx tissues and cell-type-specific epigenomic annotations. Genetic correlations computed against published GWAS summary statistics (PTSD, IBS, low back pain, depression, neuroticism). Replication arm not yet pre-published; preprint dated 2025-09-18.
Limitations
- Preprint, not yet peer-reviewed. Findings should be treated as provisional pending journal publication and replication.
- ICD-code ascertainment of FM is well-known to be heterogeneous across health-systems; some cohorts may have substantial misclassification, which would dilute true effects rather than create false positives.
- Subset mechanisms diluted. GWAS at this scale captures population-level signal; H1 (~37% of severe FM) and H2 (subset) would be diluted to invisibility in top-loci analysis. Absence of immune-tissue enrichment in this analysis does not refute H1/H2 as subset chains.
- Cross-condition genetic correlation is bidirectional. rg > 0.7 with PTSD and IBS could reflect shared upstream causal architecture, shared downstream phenotype expression (e.g., predictive-coding failure as a final-common-pathway), or ascertainment-correlated comorbidity bias. The GWAS doesn't disambiguate.
Open questions raised
- Are the top FM loci differentially weighted in the H3-only subset (anti-SGC IgG−, MC-inactive, HERV-W−) vs. the H1/H2/HERV-positive subsets? Would require genotyping of the biomarker-mapping cohort (Section 10.2 of the white paper).
- Do the top neural genes (HTT, GPR52, DRD2, CAMKV) define a biological substrate for the predictive-coding failure that Strube 2026 frames computationally? Q43 candidate.
- Does the genetic correlation with IBS reflect a shared brain-stem / vagal substrate, or a shared peripheral microbiome → brain pathway (which would unify H1 and H3 at the GWAS level)?
Triangulation notes
- Anchors H3 dominance at the population level. The white paper v0.2 §6.3 leaned on this paper rhetorically; ingestion converts that into evidence-backed triples.
- Compatible with H1 and H2 as subset mechanisms. The lack of immune-tissue enrichment in the broad GWAS does not refute Goebel 2021 (passive transfer in mice), Seefried 2025 (37% DRG-binding prevalence), or the daratumumab/PE+IVIG signal — those operate in stratified subsets that get washed out at biobank scale.
- Bridge-strengthening for B6 (post-infectious nociplastic conditions ↔ FM via predictive-coding / interoceptive-inference failure). Genetic correlations with PTSD and IBS are exactly what B6 predicts if all three share a CNS predictive-coding substrate.
- Suggests reframing the project's open questions. Q1 (antigen target of pathogenic IgG) and Q40 (HERV mitochondrial proteins) remain critical for cure-tractable subsets, but the GWAS implies that any "all-FM" therapy will need to address the H3 substrate too, not just the H1/H2 subsets.
Bridges
- Strengthens B6 (post-infectious nociplastic ↔ FM via predictive-coding). Genetic correlation rg > 0.7 with PTSD operationalizes the bridge.
- Opens a new candidate bridge B10 (FM ↔ IBS via shared CNS genetic architecture) — distinct from the gut→brain B2 microbiome-mediated bridge. B10 would be a brain→gut variant where shared CNS genetic risk produces both phenotypes.