A nomogram prediction model for treatment failure in primary membranous nephropathy

Abstract

Background Primary membranous nephropathy (PMN) has a heterogeneous natural course. Immunosuppressive therapy is recommended for PMN patients at moderate or high risk of renal function deterioration. Prediction models for the treatment failure of PMN have rarely been reported. Methods This study retrospectively studied patients diagnosed as PMN by renal biopsy at Sichuan Provincial People’s Hospital from January 2017 to December 2020. Information on clinical characteristics, laboratory test results, pathological examination, and treatment was collected. The outcome was treatment failure, defined as the lack of complete or partial remission at the end of 12 months. Simple logistic regression was used to identify candidate predictive variables. Forced-entry stepwise multivariable logistic regression was used to develop the prediction model, and performance was evaluated using C-statistic, calibration plot, and decision curve analysis. Internal validation was performed by bootstrapping. Results In total, 310 patients were recruited for this study. 116 patients achieved the outcome. Forced-entry stepwise multivariable logistic regression indicated that PLA2Rab titer (OR = 1.002, 95% CI: 1.001–1.004, p = 0.003), inflammatory cells infiltration (OR = 2.753, 95% CI: 1.468–5.370, p = 0.002) and C3 deposition on immunofluorescence (OR = 0.217, 95% CI: 0.041–0.964, p = 0.049) were the three independent risk factors for treatment failure of PMN. The final prediction model had a C-statistic (95% CI) of 0.653 (0.590–0.717) and a net benefit of 23%-77%. Conclusions PLA2R antibody, renal interstitial inflammation infiltration, and C3 deposition on immunofluorescence were the three independent risk factors for treatment failure in PMN. Our prediction model might help identify patients at risk of treatment failure; however, the performance awaits improvement.