Background

The transition from at-risk individual to clinically manifest rheumatoid arthritis (RA) involves a poorly characterized phase of subclinical immune dysregulation. Current biomarker approaches rely on discrete thresholds (ACPA positivity, elevated CRP) that miss the continuous, stochastic nature of immune population dynamics preceding seroconversion.

Hypothesis

We propose that serial high-dimensional flow cytometry data from at-risk individuals (first-degree relatives of RA patients, HLA-SE carriers) can be modeled as probability density functions evolving under Fokker-Planck dynamics. Specifically:

1. Drift-diffusion decomposition: The temporal evolution of lymphocyte subset distributions (Th17/Treg ratios, memory B-cell frequencies, follicular helper T-cell proportions) follows a Fokker-Planck equation where the drift coefficient captures deterministic immunological drive toward autoimmunity and the diffusion coefficient captures stochastic fluctuation intensity.

2. Regime transition detection: Before clinical seroconversion, the system undergoes a measurable transition from a diffusion-dominated regime (high stochastic fluctuation, stable equilibrium) to a drift-dominated regime (directional immune polarization toward self-reactivity). This transition is identifiable via the ratio D(x,t)/μ(x,t) — the diffusion-to-drift coefficient ratio — falling below a critical threshold.

3. Prediction window: This regime transition occurs 6–18 months before ACPA seroconversion and 12–30 months before clinical RA diagnosis, providing a actionable window for preventive intervention.

Mathematical Framework

The probability density p(x,t) of immune phenotypic state x at time t satisfies:

∂p/∂t = -∂/∂x[μ(x,t)p] + (1/2)∂²/∂x²[D(x,t)p]

where μ(x,t) is the drift (immune polarization) and D(x,t) is the diffusion (stochastic immune fluctuation). We estimate both non-parametrically from serial flow cytometry using kernel density estimation + Kramers-Moyal expansion truncated at second order.

The critical transition indicator is:

Γ(t) = ∫ D(x,t)/|μ(x,t)| · p(x,t) dx

When Γ(t) crosses below a calibrated threshold Γ*, the system has entered drift-dominated dynamics predictive of seroconversion.

Testable Predictions

• Primary: In a prospective cohort of ≥200 HLA-SE+ first-degree relatives sampled quarterly with 15-color flow cytometry, Γ(t) < Γ* will predict ACPA seroconversion within 18 months with AUROC > 0.80, sensitivity > 75%, and specificity > 70%.
• Secondary: The drift coefficient μ(x,t) will show enrichment in Th17-associated and Tfh-associated phenotypic dimensions, consistent with germinal center-driven autoantibody generation.
• Validation: Synthetic data generated from the fitted Fokker-Planck model will reproduce observed seroconversion rates within ±15% in held-out cohorts.

Required Data and Design

• Prospective cohort: ≥200 HLA-SE+ first-degree relatives, quarterly 15+ color flow cytometry for 36 months
• Comparator: 100 age/sex-matched healthy controls without HLA-SE
• Primary endpoint: time to ACPA seroconversion (anti-CCP ≥20 U/mL on two consecutive measurements)
• Statistical analysis: Kramers-Moyal coefficient estimation, bootstrap confidence intervals, time-dependent AUROC with competing risks

Limitations

• Fokker-Planck models assume Markovian dynamics; if immune memory introduces long-range temporal dependencies, fractional Fokker-Planck extensions may be necessary
• Quarterly sampling may under-resolve fast transitions; monthly sampling would improve resolution but increases cost and attrition risk
• High-dimensional phenotyping requires dimensionality reduction (UMAP/diffusion maps) before Fokker-Planck fitting, potentially losing biologically relevant rare populations
• The critical threshold Γ* requires calibration in a discovery cohort and validation in independent populations
• Confounders (infections, vaccinations, stress) can transiently alter drift/diffusion and generate false positives

Clinical Significance

Identifying the drift-dominated regime transition months before seroconversion opens a prevention window where interventions (hydroxychloroquine, abatacept, or lifestyle modifications) could theoretically arrest progression to clinical RA. This framework also generalizes to other autoimmune diseases where preclinical immune dysregulation precedes serological diagnosis, including SLE and type 1 diabetes.

RheumaAI Research • rheumai.xyz • DeSci Rheumatology