Analysis of influencing factors for primary membranous nephropathy complicated with atherosclerotic cardiovascular disease and establishment of a nomogram prediction model

Objective To investigate the risk factors for atherosclerotic cardiovascular disease（ASCVD） in primary membranous nephropathy（PMN） patients by machine deep learning, and to construct a nomogram. Methods This was a retrospective study including 620 patients with membranous nephropathy diagnosed by renal puncture in the First Affiliated Hospital of Xinjiang Medical University from September 1, 2017 to May 31, 2023. Based on inclusion and exclusion criteria. Eligible 424 PMN patients were finally enrolled, involving 219 ASCVD patients. They were assigned into the training （n=297） and validation groups at a 7∶3 ratio. The Mann-Whitney U test and independent-sample t test were used to screen influencing factors for ASCVD in PMN patients. LASSO regression was applied to optimize the screening variables. A nomogram was constructed and validated for its performance through visualized statistical methods. Results Compared with patients in the non-ASCVD group, those in the ASCVD group were significantly older 56（51, 63） years vs 35（28, 44） years, and had significantly higher proportions of males 154（70.3%） vs 117（57.1%）, smokers71（32.4%） vs 43（21.0%）, and alcohol consumption54（24.7%） vs 32（15.6%）. They also had significantly higher systolic blood pressure 130（120, 142） mmHg（1 mmHg=0.133 kPa） vs 122.0（113.5, 135.5） mmHg, higher diastolic blood pressure 80 （75, 90） mmHg vs 80（70, 85.5） mmHg, higher prevalence of hypertension142（64.8%） vs 67 （32.7%）, and higher proportion of manual workers60（27.4%） vs 28（13.7%） Additionally, lymphocyte count （LYM）, platelet count （PLT）, blood sodium concentration （Na）, blood urea nitrogen （BUN）, blood creatinine （Cr）, estimated glomerular filtration rate （eGFR）, 24-hour quantitative urine protein, glutamic transaminase, urinary occult blood, number of urinary hyaline and pathological tubular patterns, C-reactive protein （CRP）, complement C4, plasma fibrinogen, D-dimer, and centrocyte-lymphocyte ratio were all significantly different between the two groups （P<0.05）. Age, hypertension, BUN and eGFR were selected as risk factors for predicting ASCVD of PMN patients through LASSO regression. The area under the ROC curve （AUC） of the nomogram to predict ASCVD was 0.924 （95% CI： 0.895-0.952） in the training group, and 0.932 （95% CI： 0.892-0.972） in the validation group. The calibration curve Hosmer-Leme showed the fit was good （P=0.827 for the training group； P=0.389 for the validation group）. DCA showed that the use of nomogram prediction model was more beneficial in predicting ASCVD in PMN when the threshold probability of patients was 0.01 to 1. Conclusion The nomogram prediction model containing four predictor variables （age, hypertension, BUN, eGFR） developed in this study can be used to predict the risk of ASCVD in patients with PMN.