A LARGE UK-based study has found that novel biomarker-based models can predict kidney failure and all-cause mortality in chronic kidney disease (CKD) patients with discrimination comparable to traditional risk factor models.
The research included 2,884 adults with nondialysis CKD from 16 nephrology centers. Investigators analyzed 21 biomarkers related to kidney damage, fibrosis, inflammation, and cardiovascular disease in urine, plasma, or serum, aiming to capture underlying disease mechanisms.
For kidney failure, a three-biomarker model—soluble TNF receptor 1, soluble CD40, and urinary collagen type 1 α1 chain—achieved strong predictive performance (C-index 0.86) but was slightly outperformed by a model based on established risk factors such as age, sex, ethnicity, eGFR, and urinary albumin-to-creatinine ratio (C-index 0.90).
For all-cause mortality, a different three-biomarker model—high-sensitivity cardiac troponin T, NT-proBNP, and soluble urokinase plasminogen activator receptor—matched the predictive accuracy of traditional models (C-index 0.80 for both).
When evaluating a composite endpoint of kidney failure and mortality, the biomarker-only model marginally outperformed the established risk factor model, and combining both yielded a small but statistically significant improvement in discrimination.
The authors note that while biomarkers provided only limited incremental gains in risk prediction, they offer valuable insights into disease mechanisms and could play a role in personalizing treatment strategies and monitoring therapeutic response.
Reference
Onoja A et al. Biomarkers of Kidney Failure and All-Cause Mortality in CKD. JASN. 2025;DOI: 10.1681/ASN.0000000767.
