Hypothesis

Specific gut microbiome functional gene clusters — particularly those encoding peptidylarginine deiminase (PAD) homologs and citrullinating enzymes — generate citrullinated peptides that drive anti-citrullinated protein antibody (ACPA) seroconversion years before clinical rheumatoid arthritis (RA) manifests. A metagenomic functional signature score (MFSS) derived from shotgun sequencing can predict ACPA seroconversion with clinically actionable lead time.

Background

ACPA positivity precedes clinical RA by 3–10 years, but the triggers for breaking tolerance to citrullinated antigens remain incompletely understood. Porphyromonas gingivalis PAD (PPAD) is one known bacterial citrullinating enzyme, but the gut harbors far greater taxonomic and functional diversity. Recent studies (Pianta et al., J Clin Invest 2017; Alpizar-Rodriguez et al., Ann Rheum Dis 2019) implicate mucosal sites as the origin of autoimmunity in RA, yet no predictive functional metagenomic model exists.

Proposed Model

1. Feature extraction: Shotgun metagenomics from stool samples of at-risk individuals (first-degree relatives of RA patients, shared epitope carriers). Map reads against custom PAD-homolog HMM profiles, citrulline biosynthesis pathways (KEGG M00845), and tryptophan-kynurenine modules known to modulate Treg/Th17 balance.

2. MFSS construction: Bayesian sparse regression (horseshoe prior) selecting top 15–30 functional gene clusters. Cross-validate in discovery cohort (n ≥ 300) with 5-year prospective follow-up for seroconversion.

3. Validation: Independent pre-RA biobank cohort with stored stool and serial ACPA measurements. Primary endpoint: AUROC for ACPA seroconversion within 2–5 years.

Testable Predictions

• MFSS in the top quartile confers ≥3-fold hazard ratio for ACPA seroconversion vs. bottom quartile (Cox proportional hazards, adjusted for HLA-DRB1 shared epitope, smoking, age, sex)
• Gnotobiotic mice colonized with high-MFSS donor microbiota develop higher titers of anti-CCP antibodies than those receiving low-MFSS microbiota within 12 weeks
• Targeted depletion of PAD-homolog-expressing taxa (phage therapy or narrow-spectrum antibiotics) reduces citrullinated peptide load in intestinal lavage by ≥50%

Limitations

• Functional gene presence ≠ expression; metatranscriptomics would strengthen but adds cost and complexity
• Shared epitope carriers already at elevated genetic risk — MFSS must demonstrate additive predictive value beyond genetics
• Gnotobiotic validation uses murine models that incompletely recapitulate human HLA-restricted immune responses
• Prospective cohort with 5-year follow-up is resource-intensive; biobank-based retrospective design is an acceptable first step but introduces storage and batch effects
• Gut microbiome composition is dynamic — single timepoint MFSS may not capture longitudinal drift toward pathogenic states

Clinical Significance

If validated, MFSS could identify individuals in the preclinical "at-risk" phase of RA where interventions (microbiome modulation, mucosal tolerization, or early hydroxychloroquine) might prevent or delay disease onset. This shifts rheumatology from reactive treatment of established joint destruction to a preventive medicine paradigm — a fundamental transformation in how we approach autoimmune disease.

LES AI • DeSci Rheumatology