Comparison of the Predictive Performance of eGFR Formulae for Mortality and Graft Failure in Renal Transplant Recipients

To date, efforts have focused on assessing estimated glomerular filtration rate (eGFR) formulae against measured GFR. However, a more appropriate clinical gold standard is one conveying a defined clinical disadvantage. In renal transplantation, these measures are mortality and graft failure. The Long Term Efficacy and Safety Surveillance database was used to analyze 1344 renal transplant recipients. eGFR was assessed 6 months posttransplantation with the following formulae: Cockcroft-Gault; Walser; Nankivell; abbreviated modification of diet in renal disease (aMDRD); MDRD7; Rule's refitted MDRD; and Mayo Clinic. The outcome measures were mortality and graft failure. Although eGFR was statistically associated with subsequent mortality and graft failure in the Cox model (irrespective of which eGFR formula was used), the clinical utility of eGFR was moderate at best in predicting subsequent mortality and graft failure. No clinically relevant or statistically significant difference was discernable between formulae, with a maximum area under the receiver operating characteristic curve of 0.63 and 0.61 for 3- and 5-year mortality, respectively, and 0.66 and 0.60 for 3- and 5-year graft failure, respectively. Serum creatinine used in isolation displayed similar predictive utility, and no improvement was seen by investigating the change in creatinine or eGFR between 6 and 12 months. In summary, seven eGFR equations showed similar and limited utility in predicting mortality and graft failure after renal transplantation. This has important implications for the management of renal transplant recipients and the use of an eGFR as a surrogate endpoint in clinical trials.