Establishment And Verification Of Prognostic Model Of Idiopathic Membranous Nephropathy

Objective:To analyze the prognostic factors of patients with idiopathic membranous nephropathy,to establish and verify the clinical predictive model of idiopathic membranous nephropathy according to the prognostic factors,and to explore the value of prognostic model in the prognosis of patients with idiopathic membranous nephropathy.Methods:In this study,189 patients with idiopathic membranous nephropathy were retrospectively collected from the Department of Nephrology,from2019.09.01to.2020.12.31,and the end of follow-up on 2020.12.The follow-up period was not less than one year.The relationship between initial clinical indicators and remission was analyzed by univariate and multivariate COX regression analysis and Lasso regression analysis to establish a predictive model of idiopathic membranous nephropathy and validate it internally.71 patients from 2019.01.01 to 2019.12.31 were validated externally（time period verification）and evaluated.Results:A total of 189 patients were included in this study,of which 118 were modeled and 71 were externally validated.Univariate and multivariate COX regression analysis and Lasso regression analysis showed that 24-hour urinary protein,urea and blood PLA2R were independent risk factors for idiopathic membranous nephropathy.Finally,thefollowingformula:S（t）=S₀（t）^exp[-0.048839×24-hoururinary protein（g/d）-0.141259×urea（mmol∕L）-0.002598×serum PLA2Rantibody（RU/ML）]was obtained by COX regression.S₀（12）=0.3072583.The discrimination degree and calibration degree were evaluated in the modeling crowd,and the results showed that the AUC value of the evaluation differentiation degree was 0.725and the Brier value of the evaluation accuracy degree was 0.184 at 12 months.The formula is diagnosed by COX model with multiple collinearity and so on,and the VIF value is less than 5,and there is no multiple collinearity.This COX model strengthens the internal verification of Bootstrap,and the Harrell-C statistics of the model is 0.6689077.The external verification of the model shows that the Harrell-C statistic is 0.5751162 and the lp value is 0.4642421.Conclusions:24-hour urinary protein quantification,serum urea and serum PLA2R antibody levels are independent risk factors for whether idiopathic membranous nephropathy proteinuria remits,and a predictive model developed from these three independent risk factors is an important guide for assessing and predicting whether remission of proteinuria is achieved in patients with idiopathic membranous nephropathy.