BACKGROUND: Recently, the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) developed glomerular filtration rate (GFR)-estimating equations based on standardized serum cystatin C (CKD-EPI(cys)) and standardized serum creatinine plus standardized serum cystatin C (CKD-EPI(cr-cys)). We developed new GFR-estimating equations based on standardized cystatin C for a Japanese population and compared their accuracy with the CKD-EPI equations. STUDY DESIGN: Accuracy of diagnostic test study. SETTING & PARTICIPANTS: 413 (development data set) and 350 individuals (validation data set). INDEX TEST: CKD-EPI(cys); CKD-EPI(cr-cys); modifications to CKD-EPI(cys) and CKD-EPI(cr-cys) using Japanese coefficients; and newly developed Japanese eGFR equations based on standardized serum cystatin C (Eq(cys)), cystatin C with a nonrenal factor reflecting hypothesized extrarenal elimination (Eq(cys+nonrenal)), and creatinine in combination with cystatin C (Eq(cr-cys)). Standardized cystatin C values were determined by a colloidal gold immunoassay traceable to the international certified reference material ERM-DA471/IFCC. REFERENCE TEST: Measured GFR by inulin renal clearance. RESULTS: In a development data set, we calculated Japanese coefficients for CKD-EPI(cys) and CKD-EPI(cr-cys) of 0.977 (95% CI, 0.853-1.002) and 0.908 (95% CI, 0.889-0.928), respectively. In a validation data set, we compared CKD-EPI(cys), Eq(cys), and Eq(cys+nonrenal) with each other. Bias and accuracy were not significantly different among the 3 equations. The precision of CKD-EPI(cys) was significantly better than for Eq(cys) (P = 0.007) and not significantly different from Eq(cys+nonrenal) (P = 0.6). We then compared 0.908 × CKD-EPI(cr-cys), Eq(cr-cys), and Eq(average) (the average value of Eq(cr) [previous Japanese equation based on standardized serum creatinine] and Eq(cys+nonrenal)) with each other in the validation data set. Bias and accuracy were not significantly different among the 3 equations. The precision of 0.908 × CKD-EPI(cr-cys) was significantly better than for Eq(cr-cys) (P = 0.004) and not significantly different from Eq(average) (P = 0.06). LIMITATIONS: Limited number of participants with measured GFR >90 mL/min/1.73 m(2). Extrarenal elimination of cystatin C was not measured. CONCLUSIONS: CKD-EPI(cys) performed well in Japanese individuals, suggesting that equations based on serum cystatin C could be used in patients with different races without modification. Accounting for extrarenal elimination of cystatin C may improve the performance of estimating equations.
BACKGROUND: Recently, the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) developed glomerular filtration rate (GFR)-estimating equations based on standardized serum cystatin C (CKD-EPI(cys)) and standardized serum creatinine plus standardized serum cystatin C (CKD-EPI(cr-cys)). We developed new GFR-estimating equations based on standardized cystatin C for a Japanese population and compared their accuracy with the CKD-EPI equations. STUDY DESIGN: Accuracy of diagnostic test study. SETTING & PARTICIPANTS: 413 (development data set) and 350 individuals (validation data set). INDEX TEST: CKD-EPI(cys); CKD-EPI(cr-cys); modifications to CKD-EPI(cys) and CKD-EPI(cr-cys) using Japanese coefficients; and newly developed Japanese eGFR equations based on standardized serum cystatin C (Eq(cys)), cystatin C with a nonrenal factor reflecting hypothesized extrarenal elimination (Eq(cys+nonrenal)), and creatinine in combination with cystatin C (Eq(cr-cys)). Standardized cystatin C values were determined by a colloidal gold immunoassay traceable to the international certified reference material ERM-DA471/IFCC. REFERENCE TEST: Measured GFR by inulin renal clearance. RESULTS: In a development data set, we calculated Japanese coefficients for CKD-EPI(cys) and CKD-EPI(cr-cys) of 0.977 (95% CI, 0.853-1.002) and 0.908 (95% CI, 0.889-0.928), respectively. In a validation data set, we compared CKD-EPI(cys), Eq(cys), and Eq(cys+nonrenal) with each other. Bias and accuracy were not significantly different among the 3 equations. The precision of CKD-EPI(cys) was significantly better than for Eq(cys) (P = 0.007) and not significantly different from Eq(cys+nonrenal) (P = 0.6). We then compared 0.908 × CKD-EPI(cr-cys), Eq(cr-cys), and Eq(average) (the average value of Eq(cr) [previous Japanese equation based on standardized serum creatinine] and Eq(cys+nonrenal)) with each other in the validation data set. Bias and accuracy were not significantly different among the 3 equations. The precision of 0.908 × CKD-EPI(cr-cys) was significantly better than for Eq(cr-cys) (P = 0.004) and not significantly different from Eq(average) (P = 0.06). LIMITATIONS: Limited number of participants with measured GFR >90 mL/min/1.73 m(2). Extrarenal elimination of cystatin C was not measured. CONCLUSIONS:CKD-EPI(cys) performed well in Japanese individuals, suggesting that equations based on serum cystatin C could be used in patients with different races without modification. Accounting for extrarenal elimination of cystatin C may improve the performance of estimating equations.