Applicability of chronic kidney disease epidemiology collaboration equations in a Chinese population.

BACKGROUND: Accurate estimated glomerular filtration rates (eGFR) is an important step in the diagnosis of chronic kidney disease (CKD). The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, based on creatinine alone (eGFRcr), was developed to improve on the Modification of Diet in Renal Disease equation, in particular by addressing the systematic underestimation of high GFR. Whether the CKD-EPI equation, based on cystatin C alone (eGFRcys), or the combined creatinine-cystatin C CKD-EPI equation (eGFRcr-cys C), actually perform better than the CKD-EPI equation based on creatinine (eGFRcr) remains unknown, especially in Asians including Chinese populations, where eGFR equations may overestimate true GFR.

METHODS: A standard dual plasma sampling method (DPSM) of estimating (99m)Tc-diethylene triamine penta-acetic acid clearance was used to determine the reference or measured GFR (mGFR). Linear regression analysis, Bland-Altman analysis, bias, absolute bias and accuracy (P30) were used to compare the performance of the combined creatinine-cystatin C equation (eGFRcr-cys) and equations based on each marker alone (eGFRcr and eGFRcys) in Chinese subjects, including both patients with CKD and healthy individuals.

RESULTS: We enrolled 617 Chinese participants (49.11% female, 47.11 ± 17.25 years old), with a mean mGFR of 73.80 ± 37.55 mL/min/1.73 m(2). The predictive abilities (r), the accuracy (P15, P30, P50), bias and absolute bias of the eGFRcr-cys equation were superior to eGFRcr equation and the eGFRcys equation in overall samples. Bland-Altman analysis also demonstrated a consistent result. When compared in subgroups, the accuracy (P30) of all three equations exceeded 90% at mGFR ≥90 mL/min/1.73m(2); the eGFRcr-cys equation had the highest accuracy (P30: 95.56%). At mGFR 60-89 mL/min/1.73 m(2), the accuracies (P30) of the eGFRcr-cys and eGFRcr equations exceeded the acceptable level (≥70%), and there was no significant difference between them (P = 0.58). At mGFR <60 mL/min/1.73 m(2), the accuracy (P30) of all three equations was below 70%, but the eGFRcr-cys equation had the greatest precision.

CONCLUSIONS: The performances of the eGFRcr-cys and eGFRcr equations were similar to superior to that of the eGFRcys equation at higher GFR levels in an Asian population, especially in normal and mild to moderate kidney disease. Further improvement is needed for these equations at GFR <60 mL/min per 1.73 m(2).