Development and validation of a predictive nomogram for differentiating diabetic nephropathy from non-diabetic nephropathy in patients with T2DM: a multicenter study

Background Type 2 diabetes mellitus (T2DM) significantly exacerbates the global health burden, with diabetic nephropathy (DN) emerging as one of the most common causes of chronic kidney disease. In T2DM patients with kidney disease, it is particularly important to distinguish DN from non-diabetic nephropathy (NDN), as treatment strategies differ markedly. However, the gold standard, renal biopsy, is often impractical due to its invasive nature. This multicenter study aims to develop a non-invasive diagnostic model to distinguish DN from NDN in T2DM patients. Methods From January 2014 to December 2023, T2DM patients undergoing percutaneous renal biopsies at three hospitals in Fujian were enrolled. The model was formulated using logistic regression analysis based on clinical and laboratory parameters. A visual predictive nomogram was developed and subsequently evaluated for its predictive performance. Results A total of 292 patients were included, with 164 diagnosed with DN and 128 with NDN. Diabetic retinopathy, duration of diabetes, HbA1c, systolic blood pressure, neutrophil-to-lymphocyte ratio, kidney volume, triglycerides, estimated glomerular filtration rate, and urinary red blood cell count were identified as independent predictors of DN. A nomogram was then constructed. The model demonstrated high diagnostic accuracy with an AUC of 0.941, validated by an independent cohort yielding an AUC of 0.923. Calibration curves showed good agreement between predicted and actual outcomes, and decision curve analysis confirmed notable clinical utility. Conclusion The developed model offers a non-invasive, reliable alternative to renal biopsy for distinguishing between DN and NDN in T2DM patients. This tool proves especially valuable in clinical settings where renal biopsy is impractical, helping guide more appropriate treatment decisions.

Abstract licence: CC BY

38 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.