Performance of New Race-Free eGFR Equations for Predicting Complications in Chronic Kidney Disease: From the KNOW-CKD Study

Abstract

Introduction: The National Kidney Foundation (NKF) and the American Society of Nephrology (ASN) developed new race-free eGFR equations and recommended using these new equations in 2021. However, clinical implication of these new equations is not determined in Korean adults. Thus, this study aimed to evaluate performances of these new race-free eGFR equations in predicting complications in Korean chronic kidney disease (CKD) patients. Methods: This study analyzed 1,727 participants from the KNOW-CKD cohort. We selected anemia, hyperkalemia, acidosis, hyperphosphatemia, and hyperparathyroidism as five complications of CKD. We determined cross-sectional associations between complications and four eGFR equations. These eGFRs were calculated from 2009 Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine equation (2009 eGFR(Cr)), 2012 CKD-EPI Creatinine-Cystatin C equation (2012 eGFR(CrCys)), 2021 CKD-EPI Creatinine equation (2021 eGFR(Cr)), and 2021 CKD-EPI Creatinine-Cystatin C equation (2021 eGFR(CrCys)). Results: All associations between complications as continuous variables and eGFR by four equations were similar. All associations between complications as dichotomous variable and eGFR values form four equations were similar. For example, C-statistics (95% confidence interval) of the logistic model for anemia and eGFRs were 0.826 (0.806-0.845), 0.827 (0.806-0.846), 0.838 (0.819-0.857), and 0.839 (0.820-0.858) for 2009 eGFR(Cr), 2012 eGFR(CrCys), 2021 eGFR(Cr), and 2021 eGFR(CrCys), respectively. In addition, cross-validated areas under the curve for ROC analysis after predictive modeling for all complications were not significant different according to different eGFR equations. Conclusion: New race-free eGFR equations showed similar performances to existing equations for predicting complications in Korean patients with CKD.