Abstract: FR-PO1176
Risk Prediction of Incident CKD Using Plasma Proteomics in Individuals Without Diabetes
Session Information
- CKD: Screening, Diagnosis, Serum and Urine Biomarkers, and Scoring Indices
November 07, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: CKD (Non-Dialysis)
- 2301 CKD (Non-Dialysis): Epidemiology, Risk Factors, and Prevention
Authors
- Abbasi, Momen, University of Illinois Chicago, Chicago, Illinois, United States
- Missikpode, Celestin, University of Illinois Chicago, Chicago, Illinois, United States
- Verma, Ashish, Boston Medical Center, Boston, Massachusetts, United States
- Zhao, Runqi, Boston Medical Center, Boston, Massachusetts, United States
- Schmidt, Insa Marie, Boston Medical Center, Boston, Massachusetts, United States
- Lash, James P., University of Illinois Chicago, Chicago, Illinois, United States
- Waikar, Sushrut S., Boston Medical Center, Boston, Massachusetts, United States
- Srivastava, Anand, University of Illinois Chicago, Chicago, Illinois, United States
Background
Existing risk prediction models identify individuals at risk of incident CKD using clinical data, but data are limited on whether proteomics improve prediction of CKD in patients without diabetes.
Methods
We analyzed 2,911 baseline plasma proteins (Olink) in 39,692 UK Biobank participants without CKD and no diabetes. Multivariable-adjusted proportional hazards models tested associations of plasma proteins with incident CKD after adjusting for variables from the CKD Prognosis Consortium (CKD-PC) incident CKD risk equation. LASSO Cox regression identified a parsimonious model of plasma proteins associated with incident CKD to construct a CKD proteomic risk score in a training set. We compared the CKD-PC model to one that further included the CKD proteomic risk score in a test set.
Results
Mean age was 56±8 years, mean eGFR was 96±13 mL/min/1.73m2, and median [IQR] albuminuria was 5.7[3.6,9.2] mg/g. Over a median follow-up of 13.6 years, 1,537 participants developed CKD. After multivariable adjustment and multiple testing correction, 30 plasma proteins were significantly associated with incident CKD (Figure 1). In the training set (n=27,784 with 1104 events), 49 proteins associated with incident CKD were constructed into a CKD proteomic risk score. In the test set (n=11,908 with 433 events), each unit increase in the risk score was associated with a 1.58-fold (95% CI 1.41–1.78) higher risk of incident CKD after multivariable adjustment. Adding the CKD proteomic risk score to the CKD-PC model increased the c-index from 0.82 to 0.83 in the test set (Δ=0.01, 95% CI -0.02 – 0.04).
Conclusion
In individuals without diabetes, plasma proteins modestly improve incident CKD risk prediction beyond a validated model of clinical risk factors.