The predictive value of Berden's classification versus renal risk score for renal survival of Chinese patients with myeloperoxidase-anti-neutrophil cytoplasmic antibody-associated glomerulonephritis.

OBJECTIVES: This study aims to compare the prognostic values of two histopathological classification, Berden's classification versus renal risk score (RRS) by Brix et al. for predicting renal survival in Chinese patients with myeloperoxidase-anti-neutrophil cytoplasmic antibody (ANCA)-associated glomerulonephritis (MPO-AAGN). METHODS: The medical records of 225 patients with MPO-AAGN diagnosed in our centre between February 2004 and December 2020 were retrospectively analysed. The predictive model of Berden's classification or RRS was established by Cox regression, respectively. The above two models were compared on aspects of discrimination, calibration, and decision curve analysis for predicting the 0.5-, 1-, 3-, and 5-year renal survival. RESULTS: After a median follow up of 38.99 months, 32.44% of patients developed end-stage renal disease (ESRD). In the Kaplan-Meier analysis, there were significant differences in renal survival among groups according to Berden's classification or RRS (both log-rank p<0.001). According to time-dependent receiver operating characteristic (ROC) curve analysis, the model based on RRS showed better discrimination ability than the model based on Berden's classification for predicting 0.5-, 1-, and 3-year renal survival. For calibration analysis, the model based on RRS showed worse calibration than the model based on Berden's classification for predicting 1- and 3-year renal survival. According to the decision curve analysis, the clinical decisions based on RRS could achieve more clinical benefits than those based on Berden's classification in predicting 0.5-, 1-, and 3-year renal survival. CONCLUSIONS: The model based on RRS has better predictive value for renal survival than Berden's classification in aspect of discrimination and clinical decision from 0.5- to 3-year renal survival.