Mortality risk stratification in antineutrophil cytoplasmic antibody-associated vasculitis: baseline severity, accumulated damage burden and cardio-renal vulnerability in a Chinese two-center cohort.

Objective: Mortality in antineutrophil cytoplasmic antibody-associated vasculitis (AAV) is influenced by active inflammation, organ-threatening severity, irreversible damage and host vulnerability. We evaluated how BVAS-2003, FFS-2009 and VDI relate to all-cause mortality, and compared their endpoint discrimination with a baseline cardio-renal vulnerability model informed by the DANGER framework. Methods: We retrospectively studied 122 patients with AAV from two tertiary rheumatology centers. BVAS-2003 and FFS-2009 were assessed at diagnosis. VDI was updated during follow-up and was interpreted as accumulated damage burden rather than as a time-fixed baseline predictor. Overall survival was estimated using Kaplan-Meier analysis. Endpoint associations were assessed using a parsimonious multivariable logistic model, and discrimination was compared using receiver operating characteristic curves. Results: During a median follow-up of 36 months, 36 patients died. Estimated 1-, 3- and 5-year survival rates were 78, 58 and 39%, respectively. Non-survivors had higher BVAS-2003, higher FFS-2009, higher accumulated VDI, lower hemoglobin, higher serum creatinine and more frequent dialysis. In the endpoint multivariable model, FFS-2009 and VDI retained associations with all-cause mortality, whereas BVAS-2003 did not retain an independent association after adjustment. The DANGER-variable model showed numerically highest discrimination (area under the curve 0.78), followed by VDI (0.74), FFS-2009 (0.73) and BVAS-2003 (0.71). Conclusion: In this severe real-world AAV cohort, mortality stratification was associated with baseline organ-threatening severity, endpoint accumulated damage burden and cardio-renal vulnerability more than with global activity alone. Because VDI was not measured at a uniform landmark, VDI-related findings should be interpreted as endpoint damage-burden associations rather than validated prospective prediction.