Performance of the creatinine-based and the cystatin C-based glomerular filtration rate (GFR) estimating equations in a heterogenous sample of patients referred for nuclear GFR testing

Abstract

Cystatin C may be a more accurate marker of the glomerular filtration rate (GFR) than creatinine. We evaluated the performance of the creatinine-based abbreviated modification of diet in renal disease (MDRD), the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations, and 6 cystatin C-based equations in estimating GFR (eGFR) in a heterogeneous sample of patients. Measured GFR (mGFR) was obtained from the plasma clearance of 99mtechnetium Diethylenetriaminepentaacetic acid in 42 adult patients referred for nuclear GFR testing (January to March 2008). We evaluated the bias, precision, and accuracy of the abbreviated MDRD, CKD-EPI, Filler, Grubb, Hoek, Larsson, Le Bricon, and Rule eGFR equations. Participants had a mean mGFR of 70.9 mL/min/1.73 m2 (range: 22-125 mL/min/1.73 m2), a median age of 57 years (interquartile range: 45, 66), were 62% male, and were 38% liver transplant recipients. Correlation coefficients between eGFRs and mGFR ranged from 0.65 to 0.87 (each P<0.001). The cystatin C-based Hoek equation had the best overall performance with a low bias (-1.4 mL/min/1.73 m2), good precision (13.3 mL/min/1.73 m2), and greatest accuracy, with 93% of values within 30% of mGFR. Although the CKD-EPI equation had the lowest bias (-0.6 mL/min/1.73 m2), it had poor precision (20.7 mL/min/1.73 m2) and low accuracy, with only 69% of values within 30% of mGFR. The Hoek equation remained accurate and had the least bias when patients were grouped according to the history of liver transplantation and the mGFR above or below 60 mL/min/1.73 m2. In this heterogeneous sample, the cystatin C-based Hoek equation performed the best overall, regardless of mGFR level or history of liver transplantation.