Performance of GFR equations in Japanese subjects

Japanese GFR equations based on serum creatinine (Scr) (Eq(cr)), serum cystatin C (Scys) (Eq(cys)) and average value of Eq(cr) and Eq(cys) (Eq(average)), and coefficient-modified CKD-EPI equations based on Scr (CKD-EPIcr), Scys (CKD-EPIcys) and Scys in combination with Scr (CKD-EPIcr-cys) are now available for Japanese subjects. Performance of these equations has not been well evaluated in subjects stratified by GFR. Therefore, the bias, precision and accuracy of the GFR equations were compared in Japanese subjects stratified by measured GFR. Three hundred fifty Japanese subjects were included for validation. These subjects were stratified by measured GFR (0-29, 30-59, 60-89, 90-119 ml/min/1.73 m(2) and total). Japanese equations (Eq(cr), Eq(cys) and Eq(average)) were compared with coefficient-modified CKD-EPI equations (0.813 x CKD-EPIcr, CKD-EPIcys and 0.908 x CKD-EPIcr-cys), respectively. GFR was measured by inulin renal clearance. Standardized Scr was measured by enzymatic method. Standardized Scys was measured by colloidal gold immunoassay. Bias and accuracy were not significantly different between Japanese GFR equations and coefficient-modified CKD-EPI equations in all mGFR ranges. The precision of Eq(cr) was significantly better in GFR 0-29 ml/min/1.73 m(2) and significantly worse in GFR 60-89 and GFR 90-119 ml/min/1.73 m(2) compared with 0.813 x CKD-EPIcr. The precision of Eq(cys) was significantly better in GFR 30-59 and GFR 60-89 ml/min/1.73 m(2) compared with CKD-EPIcys. The precision of Eq(average) was significantly better in GFR 30-59 ml/min/1.73 m(2) and significantly worse in GFR 90-119 ml/min/1.73 m(2) compared with 0.908 x CKD-EPIcr-cys. Japanese GFR equations performed well in subjects with GFR under 60 ml/min/1.73 m(2) compared with the coefficient-modified CKD-EPI equations.