A study on risk factors for the progression from T2DM to end-stage renal disease based on Mendelian randomization and logistic regression analysis.

Using two-sample Mendelian randomization (MR) based on GWAS data from the IEU OpenGWAS project and a retrospective clinical cohort, this study investigated risk factors for progression from type 2 diabetes mellitus (T2DM) to end-stage renal disease (ESRD) and developed a predictive model. The T2DM GWAS dataset (ebi-a-GCST010118; 2020) included 433,540 individuals (77,418 cases and 356,122 controls). Multivariable MR, with inverse variance weighted as the primary method, was used to evaluate the causal effects of metabolic and hematologic traits on ESRD, while MR-Egger and Cochran's Q test assessed pleiotropy and heterogeneity. MR-PRESSO was used for outlier removal. In parallel, 875 patients with T2DM were analyzed using univariable and multivariable logistic regression; 140 patients (16%) progressed to ESRD. MR showed that elevated body mass index (BMI) was a causal risk factor, whereas higher hematocrit was protective. In the clinical cohort, BMI, diastolic blood pressure, and creatinine were identified as independent risk factors, while albumin and hematocrit were protective. Sensitivity analyses showed no significant horizontal pleiotropy, and all instrumental variables were sufficiently strong (F-statistics >10). A logistic regression-based nomogram achieved an AUC of 0.88 (95% CI: 0.85-0.91) with good calibration and outperformed XGBoost, Random Forest, and support vector machine models. Elevated BMI and lower hematocrit increase ESRD risk in T2DM, and the nomogram may support early risk stratification and precision intervention.