Assessing renal graft function in clinical trials: Can tests predicting glomerular filtration rate substitute for a reference method?

Background. In clinical trials, comparison of renal graft function needs a rigorous determination of glomerular filtration rate (GFR). Since reference methods to measure GFR cannot be easily implemented, a number of tests predicting GFR are usually used. However, little is known about their validity in renal transplant patients. We aimed to compare the performances of six GFR tests with inulin clearance in this population. Methods. Five hundred consecutive inulin clearances performed in 294 renal transplant recipients with stable renal function were retrospectively selected. For each of them, we computed six estimates: the 24- hour creatinine clearance, the Cockcroft- Gault, Walser, Jelliffe, Nankivell, and Levey formulas. Their respective performance was assessed by correlation (simple linear regression), accuracy (dispersion of true error), and agreement (Bland and Altman method). Results. Each GFR test closely correlated with inulin clearance (P < 0.0001). Comparisons between pairs of GFR tests did not show any significant difference in accuracy between the Levey, Jelliffe, and Walser formulas. Conversely, each of these formulas demonstrated a significant lower dispersion (P 0.005) than the others. Nevertheless, all GFR tests displayed considerable lack of agreement with limits of agreement over 40 mL/ min/1.73 m(2) apart. The proportion of predicted GFR differing from inulin clearance by +/- 10 mL/ min/ 1.73 m(2), ranged from 34% for the Jelliffe formula to 53% for the Nankivell's one. Conclusion. None of these formulas seems to be able to safely substitute for inulin clearance. In clinical trials, renal graft function should be preferably assessed using a reference method of GFR measurement.