Florence Delestre, Pierre Charles, Maxime Samson, Antoine Néel, Stanislas Faguer, Alexandre Karras, François Lifermann, Pascal Godmer, Catherine Hanrotel-Saliou, Nicolas Martin-Silva, Grégory Pugnet, François Maurier, Thomas Le Gallou, Thomas Quéméneur, Nadine Meaux-Ruault, Jean-François Viallard, Eric Hachulla, Xavier Puéchal, Loïc Guillevin, Raphaël Porcher, Benjamin Terrier, French Vasculitis Study Group (FVSG)
The relapse prediction model demonstrated reproducible performance in this external validation cohort and may contribute to clinically relevant risk stratification, especially in PR3-ANCA patients. The infection model showed stronger predictive performance and may support individualized discussions regarding RTX extension.
OBJECTIVES: Long-term follow-up of the MAINRITSAN trials confirmed the efficacy of rituximab (RTX) in preventing relapses in ANCA-associated vasculitis (AAV). Extending RTX maintenance beyond 18 months may benefit high-risk patients. The aim of this study was to validate relapse and infection prediction models to guide individualized RTX extension decisions.
METHODS: Data from 217 patients in the MAINRITSAN1 and MAINRITSAN2 trials treated with RTX for 18 months with follow-up through December 2020 were analyzed. We externally validated relapse and infection prediction models from McClure et al. Primary endpoints were relapse-free survival and serious infection-free survival.
RESULTS: Relapse occurred in 76 patients, with a 72-month relapse-free survival rate of 55% (95% CI 48-64). Serious infections occurred in 33 patients, with a cumulative 72-month incidence of 20%. The relapse model (factors: male sex, age > 60 years, ANCA positivity, relapsing disease, ENT involvement, prednisone dose) predicted major relapse (coefficient 1.34, p = 0.04) and identified high-risk patients with shorter major relapse-free survival (HR 1.91, 95% CI 0.97-3.76, p = 0.059). The relapse score was associated with major relapse risk predominantly within the PR3-ANCA subgroup (coefficient 2.76, p = 0.01). The infection model (factors: male sex, structural lung disease, diabetes, occurrence of infections during maintenance therapy, and gammaglobulin level) identified high-risk patients with shorter serious infection-free survival (HR 2.93, 95% CI 1.28-6.72, p = 0.011).
CONCLUSION: The relapse prediction model demonstrated reproducible performance in this external validation cohort and may contribute to clinically relevant risk stratification, especially in PR3-ANCA patients. The infection model showed stronger predictive performance and may support individualized discussions regarding RTX extension.