Evaluation of race-free eGFR equations in individuals of different ethnicity

Abstract

Background: Glomerular filtration rate (eGFR) derived from serum creatinine (eGFR_cr), cystatin C (eGFR_cys), or both (eGFR_cr-cys) by race-free equations are recommended staging chronic kidney disease (CKD). The current study aimed to compare these race-free eGFR equations for screening for low-grade CKD in Blacks and non-Blacks and to evaluate their association with mortality.

Methods: Race-free eGFR equations were evaluated in four studies with specific inclusion criteria based on the original research goals: African-PREDICT (341/380 healthy Black/White South Africans), FLEMENGHO (709 White community-dwelling Flemish), NHANES (1760/7931 Black and non-Black adult Americans), and 401 Black African patients hospitalised in Mbuji Mayi, Democratic Republic of Congo. The intraclass correlation coefficient and Bland and Altman statistics were used to assess consistency between eGFR equations and multivariable logistic or Cox regression to evaluate their association with mortality.

Results: Intraindividual discordance between eGFRs was larger in Black than non-Black NHANES and African-PREDICT participants. In NHANES, eGFR_cr-cys was greater than eGFR_cr, but smaller than eGFR_cys, and replacing eGFR_cr-cys by eGFR_cr moved 25% Blacks and 15% non-Blacks to a higher (worse) eGFR KDIGO stage. In African-PREDICT and FLEMENGO, half of the measured creatinine clearance to eGFR ratios fell outside the expected 1.1-1.2 band. In NHANES, multivariable hazard ratios for total and cardiovascular mortality in relation to CKD grade were all lower than unity for grade-1 CKD and greater than unity for grade ≥3 (p < 0.0001) without any racial difference (0.11≤p ≤ 0.98). These NHANES findings were consistent, if CKD stage was replaced by eGFR and in subgroup analyses. Whereas eGFR_cys and eGFR_cr-cys refined models, eGFR_cr did not.

Conclusions: The NHANES mortality outcomes support the use of eGFR_cys and eGFR_cr-cys. However, large intraindividual variability between eGFR estimates may lead to KDIGO eGFR stage misclassification and calls for caution in the opportunistic or systematic screening for CKD in asymptomatic individuals with prevention as objective.

Keywords: Biomedical analytics; chronic kidney disease; estimated glomerular filtration rate; mortality; population science; race.

Figure 1.

Flowchart showing the selection of study participants. The mCl_cr/eGFR ratio was assessed in Black and White South Africans and in White Europeans, enrolled in African-PREDICT and FLEMENGHO, respectively (a). The association between mortality and eGFR derived from race-free equations was studied in Black and non-Black NHANES participants representative of the adult population of the United States and in Black hypertensive patients admitted to the emergency department of two hospitals in Mbuji Mayi, Democratic Republic of Congo (b). mCl_cr: measured creatinine clearance; HT: hypertension; BP: blood pressure; BMI: body mass index.

Figure 2.

Ratio of measured creatinine clearance to eGFR derived by the CKD-EPI 2021 and EKFC 2023 equations. The equations to compute eGFR are given for CKD-EPI 2021 and EKFC 2023 in the Supplementary Tables 1–2. The mCl_cr/eGFR ratios for creatinine-based eGFRs are depicted in (a), for cystatin C-based eGFRs in (b), and for equations including both serum biomarkers in (c). The shaded area in the three panels reflects GFR uncorrected for the proximal tubular creatinine secretion, resulting in a mCl_cr/eGFR ratio ranging from 1.1 to 1.2. mCl_cr is standardised to 1.73 m² body surface area. Plotted values are means ± 95% confidence interval. For ratios falling outside the band, the mean is given alongside the plotted value. For the same eGFR equation, none of the racial differences attained statistical significance (p ≥ 0.10).

Figure 3.

Reclassification of chronic kidney disease grades in NHANES. Numbers in green, orange and red bars are percentages of chronic kidney disease grades (CKD) according to the 2024 KDIGO classification based on eGFR (≥90, 60–89, and <60 mL/min/1.73m²) from serum creatinine (eGFR_cr), serum cystatin C (eGFR_cys) or both (eGFR_cr-cys). Non-Blacks include 2656 Hispanics and 5275 Whites. CKD-EPI denotes Chronic Kidney Disease Epidemiology Collaboration and EKFC the European Kidney Function Consortium. The κ statistic, given with 95% confidence interval, is a measure of the consistency in the distribution of CKD grades between adjacent classifications. κ values of >0.40, >0.60 and >0.80 indicate moderate, substantial and almost perfect agreement.