Comparison of estimated GFR and measured GFR in prospective living kidney donors

The aim of this study was to examine the usefulness of three GFR-estimating equations (eGFR) compared with measured GFR (mGFR) in potential living kidney donors.

We compared the performance of the MDRD, CKD-Epi and Cockcroft-Gault equations with mGFR measured using (51)Cr-EDTA in 508 consecutive potential living kidney donors. Each equation was assessed for bias, precision and accuracy compared with mGFR, and the sensitivity and specificity for the identification of donors with mGFR<80 mL/min/1.73 m2 was evaluated.

Two hundred and forty-four subjects were male, 398 Caucasian, 60 Afro-Caribbean and 50 from other ethnic groups. Median age and mGFR were 44.1 year and 91.7 mL/min/1.73 m2, respectively. Spearman correlation coefficients between eGFR and mGFR were in the range R s=0.520-0.593. Median bias (eGFR-mGFR) for the MDRD, CKD-Epi and Cockcroft-Gault equations were -1.0 (p=0.98), +8.8 (p<0.0001) and +11.1 (p<0.0001) mL/min/1.73 m2, respectively. Significant differences in bias between Afro-Caribbean and Caucasian subjects were found. The sensitivity (specificity) for the MDRD, CKD-Epi and Cockcroft-Gault equations for identifying subjects with mGFR<80 mL/min/1.73 m2 was 60 (83), 39 (95) and 44% (95%), respectively.

The level of agreement between mGFR and all three eGFR values was poor, with the MDRD equation performing best. We conclude that reliance on creatinine-based eGFR values is unsatisfactory for the evaluation of potential living kidney donors.