Comparison of glomerular filtration rate estimating formulas among Japanese adults without kidney disease

Background: Previous studies have proposed different formulas of estimating glomerular filtration rate (eGFR) among clinical patients. The comprehensive comparison of eGFR formulas is not well established in a Japanese population. We compared eGFR values and chronic kidney disease (CKD) classification of nine different eGFR in a Japanese general population sample. Methods: We analyzed 469 Japanese community-dwelling adults (184 men) without any self-reported kidney disease. GFR estimated using the 4- and 6-parameter Modification of Diet in Renal Disease (MDRD) formulas (MDRD4 and MDRD6); the CKD-EPI formulas based on creatinine with (CKD-EPI-2009) and without race coefficient (CKD-EPI-2021), on cystatin C (CKD-EPI-Cys), on both (CKD-EPI-CreCys); the Japanese creatinine-based formula (JPN-Cre), cystatin C-based formula (JPN-Cys), and modified CKD-EPI formula (JPN-CKD-EPI). CKD stages were defined by KDIGO guidelines (eGFR < 60 ml/min/1.73 m2). Results: eGFRJPN-Cre (mean = 71.2; SD = 14.3) were much lower than eGFRCKD-EPI-2021 (mean = 94.2; SD = 12.7), while eGFRJPN-Cys (mean = 102.8; SD = 24.2) was comparable to the MDRD and CKD-EPI formulas. The difference between eGFRCKD-EPI-2021 and eGFRJPN-Cre showed a V-shaped distribution across eGFR levels, indicating complex errors between these formulas. We observed very low agreement in CKD classification between eGFRJPN-Cre and the eGFRCKD-EPI-2021 (kappa = 0.13; 95% confidence interval: 0.06, 0.23). Conclusions: JPN-Cre was substantially different from the CKD-EPI formula without race term (CKD-EPI-2021), which means that it is impossible to recalibrate those with a simple coefficient. Although a comparison with measured GFR should be necessary, choice of the estimation method needs caution in clinical decision-making and academic research.