As an alternative to measure glomerular filtration rate (GFR) with an exogenous tracer (mGFR) for all candidates to living kidney donation (LKD) Huang et al. developed, from a cohort representative of the US population, an online tool, based on estimated GFR (eGFR), to streamline GFR evaluation.1 A threshold mGFR ≥90 ml/min/1.73 m2 has been recommended as acceptable for candidates to LKD.2, 3 Hence, Huang's calculator may be useful for LKD screening if it predicts an mGFR ≥90 ml/min/1.73 m2 with sufficient accuracy to avoid further testing. However, Huang's model is based on accepted donor candidates and it is not clear whether it is applicable to all donor candidates.

We used Huang's calculator in a French cohort of 2733 candidates to LKD screened between 2007 and 2017 and who had (as required by French regulation) an mGFR with iohexol, inulin, or 51Cr-EDTA. They also had plasma creatinine measured with enzymatic IDMS traceable method.4 To assess the calibration of Huang's predictive tool in our population, we plotted eGFR and mGFR for each probability-group in Figure 1. In the right panel, the probability of having an mGFR ≥90 ml/min/1.73 m2 is between 96 and 100%. In this group, all individuals had an eGFR ≥90 ml/min/1.73 m2 but only 89.7% of females, 88.6% of males, and 89.3% of ≤65 years had an mGFR ≥90 ml/min/1.73 m2. Assuming that individuals with a probability >95% of mGFR ≥90 ml/min/1.73 m2 may not require mGFR, Huang reported that 53% of GFR measurements could have been omitted as unnecessary.1 In our cohort, only 839/2733 (30.6%) candidates are in this situation.

Dot plot of observed mGFR and eGFR according to the probabilities (given by Huang's calculator) of having an mGFR ≥90 ml/min/1.73 m2. In the left panel, the probability of having an mGFR ≥90 ml/min/1.73 m2 according to the calculator is between 0 and 5%. Numbers in red represent the proportion of individuals with an mGFR ≥90 ml/min/1.73 m2 for each probability group (red area): (A) female candidates for donation, (B) male candidates for donation, (C) candidates for LKD younger than 65 years, and (D) candidates for LKD older than 65 years

Several factors explain this difference. Huang included only donors while we included donors and screened nondonors. In our cohort, the percentage of individuals with a probability >95% of having an mGFR ≥90 ml/min/1.73 m2, was higher for donors than for non-donors (32% vs. 24%, p < .001). Second, a population of candidates is older than one of donors (50.8 ± 11.7 years vs. 49.8 ± 11.7 years, p = .012) in our cohort. While 57.2% of donors were younger than 44 years in Huang cohort, 29.1% of candidates were in this age group in our cohort. Third, a population of candidates has a lower eGFR than one of donors (94.5 ± 14.3 ml/min/1.73 m2 vs 95.6 ± 13.9 ml/min/1.73 m2, p = .013) in our cohort. Mean eGFR was 100 ± 18 ml/min/1.73 m2 in Huang study while it was 94.7 ± 13.9 ml/min/1.73 m2 in our cohort. Age and eGFR are variables of the calculator, therefore changing the characteristics of the reference group has an impact on the performance of the calculator.

Among individuals with a 100% probability of having an mGFR ≥90 ml/min/1.73 m2 32/432 (6.8%) have an mGFR <90 ml/min/1.73 m2 suggesting that even within this very restrictive framework, Huang's calculator exhibits suboptimal accuracy.

As a limitation, we did not evaluate cystatin-C, that could increase the performance of the calculator because this measure is not widely used. Second, we conducted this analysis only in France. Calibration of the calculator may vary by country. Third, we aggregated different mGFR techniques and did not account for their variability.

Including all candidates to LKD (not just those who donated), better reflects the clinical use of this calculator and highlights its limitations to ascertain an mGFR ≥90 ml/min/1.73 m2.

ACKNOWLEDGMENTS

The authors thank all the contributors to the French living kidney donor database, Pr Christophe LEGENDRE, Pr Maryvonne HOURMANT, Pr Nassim KAMAR, Pr Lionel ROSTAING, Pr Laurence DUBOURG, Pr Lionel COUZI, Pr Marie COURBEBAISSE, Dr Cyril GARROUSTE.

DISCLOSURE

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.