Biopsy-proven CKD etiology and outcomes: the Chronic Kidney Disease Japan Cohort (CKD-JAC) study
Takayuki Hamano, Takahiro Imaizumi, Takeshi Hasegawa, Naohiko Fujii, Hirotaka Komaba, Masahiko Ando, Masaomi Nangaku, Kosaku Nitta, Hideki Hirakata, Yoshitaka Isaka, Takashi Wada, Shoichi Maruyama, Masafumi Fukagawa
Abstract
BACKGROUND: The Kidney Disease: Improving Global Outcomes guidelines advocate the cause-glomerular filtration rate (GFR)-albuminuria (CGA) classification for predicting outcomes. However, there is a dearth of data supporting the use of the cause of chronic kidney disease. This study aimed to address how to incorporate a prior biopsy-proven diagnosis in outcome prediction. METHODS: We examined the association of biopsy-proven kidney disease diagnoses with kidney failure with replacement therapy (KFRT) and all-cause death before KFRT in patients with various biopsy-proven diagnoses (n = 778, analysis A) and patients with diabetes mellitus labeled with biopsy-proven diabetic nephropathy (DN), other biopsy-proven diseases and no biopsy (n = 1117, analysis B). RESULTS: In analysis A, adding biopsy-proven diagnoses to the GFR-albuminuria (GA) classification improved the prediction of 8-year incidence of KFRT and all-cause death significantly regarding integrated discrimination improvement and net reclassification index. Fine-Gray (FG) models with KFRT as a competing event showed significantly higher subdistribution hazard ratios (SHRs) for all-cause death in nephrosclerosis {4.12 [95% confidence interval (CI) 1.11-15.2)], focal segmental glomerulosclerosis [3.77 (95% CI 1.09-13.1)]} and membranous nephropathy (MN) [2.91 (95% CI 1.02-8.30)] than in immunoglobulin A nephropathy (IgAN), while the Cox model failed to show significant associations. Crescentic glomerulonephritis had the highest risk of all-cause death [SHR 5.90 (95% CI 2.05-17.0)]. MN had a significantly lower risk of KFRT than IgAN [SHR 0.45 (95% CI 0.24-0.84)]. In analysis B, other biopsy-proven diseases had a lower risk of KFRT than biopsy-proven DN in the FG model, with death as a competing event [SHR 0.62 (95% CI 0.39-0.97)]. CONCLUSIONS: The CGA classification is of greater value in predicting outcomes than the GA classification.