Abstract: FR-PO0333
Parsimonious Cardiorenal Prognostic Algorithm: Web-Based Risk Calculator for Predicting Cardiovascular and Kidney Outcomes in Type 2 Diabetes and CKD
Session Information
- Diabetic Kidney Disease: Progression, Predictive Tools, Therapeutics, and Outcomes
November 07, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Diabetic Kidney Disease
- 702 Diabetic Kidney Disease: Clinical
Authors
- Md Dom, Zaipul, Joslin Diabetes Center, Boston, Massachusetts, United States
- Tye, Sok Cin, Joslin Diabetes Center, Boston, Massachusetts, United States
- Satake, Eiichiro, Joslin Diabetes Center, Boston, Massachusetts, United States
- Tangiisuran, Balamurugan, Joslin Diabetes Center, Boston, Massachusetts, United States
- Lash, James P., The University of Chicago, Chicago, Illinois, United States
- Greene, Tom, University of Utah Health, Salt Lake City, Utah, United States
- Galecki, Andrzej, University of Michigan, Ann Arbor, Michigan, United States
- Doria, Alessandro, Joslin Diabetes Center, Boston, Massachusetts, United States
- Krolewski, Andrzej S., Joslin Diabetes Center, Boston, Massachusetts, United States
Background
Chronic kidney disease (CKD) progression is linked to kidney and cardiovascular disease (CVD) complications. Existing tools predict these outcomes separately and lack the ability to provide integrated, individualized predictions. This study aimed to develop a multi-marker prognostic tool to predict cardiorenal events in individuals with CKD.
Methods
We utilized the custom-designed Joslin Kidney Panel (JKP) to quantify 21 plasma proteins (pg/ml) in 1,170 participants with CKD stages 1-3 at baseline from the Chronic Renal Insufficiency Cohort study. The cohort was randomly split into development (70%) and validation (30%) sets. The cardiorenal outcome included a composite of kidney failure/50% eGFR decline (KF), heart failure (HF), and other CVD events. Univariate logistic regression was used to assess individual predictors. Elastic net logistic regression with cross-validation was employed to develop a cardiorenal prognostic model. Model calibration was performed by comparing predicted probabilities with observed outcomes. Model discrimination was evaluated using the area under the ROC curves (AUC).
Results
Kidney and CVD outcomes overlapped substantially, with a total of 598 (51%) cardiorenal events over median (IQR) 4.2 (2.2-6.7) years. Individual protein associations were comparable across outcomes with KIM-1 showing odds ratios of 2.22 for cardiorenal, 2.64 for KF, 2.39 for HF, and 2.13 for other CVD events. Fifteen predictors (10 clinical variables and 5 JKP proteins) were selected by the elastic net algorithm to build the cardiorenal predictive model. The model achieved AUCs of 0.781 (95%CI, 0.749-0.812) in development and 0.786 (95%CI, 0.739-0.833) in validation sets, with good calibration (Hosmer-Lemeshow p>0.05). A web-based risk calculator was developed for individual cardiorenal risk estimation.
Conclusion
We have developed a new cardiorenal prognostic tool for simultaneously predicting kidney and CVD outcomes with excellent accuracy, allowing better clinical decisions in high-risk individuals with diabetes and CKD. External validation is needed before clinical implementation. (Data provided by NIDDK-CR, a program of the National Institute of Diabetes and Digestive, and Kidney Diseases).
Funding
- NIDDK Support