Predicting Nondiabetic Kidney Disease in Patients With Type 2 Diabetes Mellitus

In patients with type  2 diabetes mellitus, identifying nondiabetic kidney disease (NDKD) is crucial in delaying progression of chronic kidney disease (CKD). The gold standard to detect the presence of NDKD is renal biopsy. However, according to Vamsidhar Veeranki and colleagues at the Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, Uttar Pradesh, India, renal biopsy is associated with a risk of life-threatening complications.

The researchers conducted a study to develop a noninvasive scoring model to predict the presence of NDKD using clinical and laboratory parameters. Results were reported during a poster session at the American Society of Nephrology Kidney Week 2023 in a poster titled Machine Learning Algorithm in Predicting Nondiabetic Kidney Disease in Type 2 Diabetes Mellitus: Development and Validation of a Noninvasive Predictor Scoring Model.

Patients with type 2 diabetes mellitus who underwent biopsy for various indications were included in the study. Participants were divided into two cohorts: a derivational cohort and a validation cohort. Using variables significantly associated with the presence of NDKD on biopsy on univariate analysis, a model was developed using multivariate logistic regression based on machine learning algorithm. The model was then run on the derivation (internal validation) cohort and the validation cohort (temporal validation). Receiver operating characteristic area under the curve (ROCAUC) was used to assess the performance of the model.

The study analysis included 538 patients with type 2 diabetes mellitus. Of the overall cohort, 376 were in the derivational cohort and 162 in the validation cohort. In the final model, diabetes mellitus derivation less than 5 years, absence of coronary artery disease, absence of diabetic retinopathy, presence of oliguria, acute rise in creatinine, and low serum complement-C3 level significantly predicted the presence of NDKD. The model performed robustly with AUC-ROC of 0.869 (95% CI, 0.805-0.933) in the validation cohort.

“The clinical and laboratory parameter-based prediction model robustly predicted the NDKD among type 2 diabetes patients, and a cut-off total score of six or more has a high sensitivity and specificity of 86% and 80%, respectively, in predicting NDKD,” the authors said.

Source: Veeranki V, Prasad N, Meyyappan J. Machine learning algorithm in predicting nondiabetic kidney disease in type 2 diabetes mellitus: development and validation of a noninvasive predictor scoring model. TH-PO013. Abstract of a poster presented at the American Society of Nephrology Kidney Week 2023; November 2, 2023; Philadelphia, Pennsylvania.