Which equation and which biomarker should be used in the kidney failure risk equation? a comparative analysis in the CKD-REIN cohort

Abstract Background The Kidney Failure Risk Equation (KFRE) is a widely used tool to estimate the 2- and 5-year risk of progression to kidney failure in patients with chronic kidney disease (CKD). The choice of estimated glomerular filtration rate (eGFR) equation and biomarker may influence its prognostic performance. We aimed at comparing the predictive performance of the KFRE when used with different eGFR equations (CKD-EPI vs EKFC) and biomarkers (creatinine, cystatin C, or both) in a large French CKD cohort. Methods We analyzed 2 168 patients with CKD stages G3 to G5 from the prospective CKD-REIN cohort. Eight eGFR equations were evaluated in combination with both the 4- and 8-variable KFRE models. Prognostic performance was assessed at 2 and 5 years using time-dependent AUROC, Brier scores, calibration plots, and decision curve analysis. Subgroup analyses were performed by age and sex. Results Discrimination was uniformly high (AUROC 0.89), but calibration patterns and decision curve analyses differed between equations. The creatinine-based CKD-EPI 2009 equation demonstrated robust and consistent performance across models and time horizons. EKFC equations—especially when using creatinine alone—showed slightly improved clinical utility at the 40% decision threshold. Cystatin C-based equations did not improve KFRE predictions compared to creatinine-based ones. Conclusions In this French CKD cohort, both CKD-EPI 2009 and EKFC creatinine-based equations supported accurate risk prediction using the KFRE. These findings support the use of EKFC in European clinical practice and reinforce the importance of aligning eGFR equations with the population context when applying prognostic models.