Infection as a predictor and potential mediator of poor renal outcomes in rituximab-treated primary membranous nephropathy: a retrospective cohort analysis

Rituximab is effective for primary membranous nephropathy (PMN), yet infectious complications remain a critical safety concern with poorly characterized risk factors and prognostic implications. This retrospective study analyzed 88 biopsy-proven PMN patients receiving rituximab at a tertiary center from January to December 2022. Patients were categorized based on the occurrence of severe infection (necessitating hospitalization, intravenous antibiotics, or causing death) over 24 months. Primary outcomes included cumulative probabilities of complete remission (CR) and clinical remission (CR plus partial remission, PR). Infections occurred in 18 patients (20.5%), comprising 29 episodes (skin or soft tissue, respiratory, perirenal, and sepsis). Infected patients were older and presented with lower baseline serum albumin, higher urea, and reduced renin-angiotensin-aldosterone system blockade usage (all P < 0.05). At 24 months, the infection group demonstrated markedly inferior outcomes: complete remission rate 11.1% versus 38.6% ( P = 0.0537, not significant), overall clinical remission rate 61.1% versus 84.3% (Fisher’s exact P = 0.0467; χ² P = 0.0648, not significant), cumulative first-response rate 72.2% versus 95.7%, and delayed median time to remission (12 versus 3 months, P = 0.0004). Infected patients exhibited persistently worse renal function (lower eGFR) throughout follow-up ( P < 0.05). Multivariate analysis identified age (odds ratio 1.054 per year [95% CI: 1.005–1.105], P = 0.030), serum albumin (odds ratio 0.880 per g/L increase [0.801–0.967], P = 0.008), and treatment-naïve status (odds ratio 0.155 [0.029–0.821], P = 0.028) as independent predictors. Baseline albumin predicted infection (AUC 0.696 [95% CI: 0.534–0.833 by bootstrap, B = 1,000]); the Youden optimal cutoff was approximately 21.0 g/L (sensitivity 0.611, specificity 0.814). In multivariate Cox regression adjusting for confounders, infection status remained independently associated with delayed remission (HR 0.33 [0.17–0.67], P = 0.002). After 1:2 propensity score matching, the 24-month remission rate was 76.5% in the infection group versus 97.1% in the non-infection group (log-rank P = 0.007). Time-dependent Cox analysis treating infection as a time-varying exposure yielded a hazard ratio of 0.20 (95% CI 0.08–0.51, P = 0.001) for clinical remission after infection occurrence. Infections during rituximab therapy for PMN profoundly compromise treatment efficacy and renal outcomes. Baseline hypoalbuminemia and advanced age independently predict infection risk, while albumin level provides a candidate stratification aid warranting prospective validation. These findings support the rationale for risk-adapted monitoring and prophylactic strategies to optimize rituximab safety in PMN.