A Personalized Rituximab Retreatment Approach Based on Clinical and B-Cell Biomarkers in ANCA-Associated Vasculitis.

Background: Time to relapse after rituximab for the treatment of antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) is variable, and optimal retreatment strategy has remained unclear. In AAV following rituximab induction, the study objective was to evaluate clinical and B-cell predictors of relapse in order to develop a retreatment algorithm. Methods: A retrospective observational study was conducted in 70 rituximab-treated ANCA-associated vasculitis patients followed up for over 10 years. Complete response (CR) was defined as Birmingham Vasculitis Activity Score v3.0 = 0. Retreatment was given on clinical relapse, defined as new features or worsening of persistent disease (not by biomarker status). Peripheral B-cell subsets were measured using highly sensitive flow cytometry. Predictors were tested using multivariable Cox regression. Results: Median time to retreatment for cycles 1-5 were 84, 73, 67, 60, and 73 weeks. Over 467 patient-years follow-up, 158 relapses occurred in 60 patients; 16 (in 15 patients) were major (renal = 7, neurological = 4, ENT = 3, and respiratory = 2). The major-relapse rate was 3.4/100 patient-years. In multivariable analysis, concomitant immunosuppressant [HR, 0.48 (95% CI, 0.24-0.94)], achieving CR [0.24 (0.12-0.50)], and naïve B-cell repopulation at 6 months [0.43 (0.22-0.84)] were associated with longer time to relapse. Personalized retreatment using these three predictors in this cohort would have avoided an unnecessary fixed retreatment in 24% of patients. Area under the receiver operating characteristic for prediction of time to relapse was greater if guided by naïve B-cell repopulation than if previously evaluated ANCA and/or CD19 Conclusion: Our findings suggest that all patients should be coprescribed oral immunosuppressant. Those with incomplete response or with absent naïve B cells should be retreated at 6 months. Patients with complete response and naïve repopulation should not receive fixed retreatment. This algorithm could reduce unnecessary retreatment and warrant investigation in clinical trials.