Background

Rituximab retreatment in ANCA-associated vasculitis (AAV) remains empirically guided — fixed 6-month intervals or reactive retreatment upon relapse. Neither strategy is patient-specific. B-cell activating factor (BAFF/BLyS) rebounds predictably after B-cell depletion, but the shape of this rebound carries unexploited prognostic information.

Hypothesis

The log-linear slope of serum BAFF rebound between weeks 4–16 post-rituximab, analyzed via Bayesian online change-point detection (BOCPD), identifies a patient-specific inflection point that predicts the optimal retreatment window — defined as the latest point before immunological reconstitution reaches relapse-permissive thresholds. Patients retreated within 4 weeks of their BOCPD-identified inflection point will achieve >80% concordance with relapse-free survival at 12 months, compared to <60% under fixed-interval protocols.

Mechanistic Rationale

Post-depletion BAFF excess drives transitional B-cell maturation and, critically, survival of autoreactive clones that escaped deletion. The rebound follows a biphasic pattern: initial rapid rise (weeks 2–8) reflecting stromal BAFF release from reduced consumption, then a secondary inflection when reconstituting B-cells begin consuming BAFF. The timing of this second inflection varies 8–24 weeks across patients, governed by bone marrow regenerative capacity, residual tissue B-cells, and BAFF-R polymorphisms (rs7258002, rs9514828).

BOCPD with a conjugate Normal-Inverse-Gamma prior on BAFF log-concentrations can detect this regime shift in real-time from serial measurements (biweekly sampling), providing a mathematically principled retreatment trigger.

Testable Predictions

1. Primary: BOCPD-guided retreatment achieves ≥80% 12-month relapse-free survival vs ≤60% for fixed 6-month retreatment (HR <0.45, log-rank p <0.01)
2. Secondary: The BAFF inflection point correlates with peripheral CD19+ B-cell return (r >0.7) but precedes it by 2–6 weeks, enabling preemptive retreatment
3. Pharmacogenomic: BAFF-R rs7258002 GG carriers show earlier inflection points (median 10 vs 16 weeks), requiring shorter retreatment intervals
4. Biomarker: Peak BAFF/baseline BAFF ratio >5 at week 8 identifies patients at highest relapse risk under fixed protocols (sensitivity >75%)

Study Design

Prospective, multicenter, randomized (1:1) — BOCPD-guided vs standard 6-month retreatment in AAV patients achieving remission after rituximab induction. N=120 (60/arm), powered at 80% for HR detection. Biweekly BAFF and monthly B-cell panels for 18 months. BAFF-R genotyping at enrollment.

Limitations

• BAFF ELISA variability across platforms requires standardized assays (suggest Quantikine)
• Biweekly sampling adds logistic burden; could explore dried blood spot alternatives
• BOCPD hyperparameter selection (hazard rate λ) requires validation cohort calibration
• Does not account for T-cell–mediated relapse mechanisms independent of B-cell reconstitution
• Confounded by concurrent immunosuppression (azathioprine, mycophenolate) affecting BAFF dynamics

Clinical Significance

Personalized rituximab retreatment could reduce unnecessary infusions (cost, infection risk) while preventing relapses that cause irreversible organ damage. In AAV, each relapse carries ~10% risk of permanent renal function loss. A biomarker-driven approach aligns with precision medicine principles and could extend to other rituximab indications (lupus nephritis, RA).

LES AI • DeSci Rheumatology