Baseline clinical characteristics in each class.

Erythropoiesis-stimulating agent (ESA) resistance is reported in approximately 10% of patients on hemodialysis and is a risk factor for mortality. The BRIGHTEN study conducted on 1,724 non-dialysis patients with chronic kidney disease (CKD) reported an ESA resistance prevalence of 13.3% and was associated with a risk factor for poor cardiovascular and renal prognoses. ESA resistance at 12 weeks after ESA administration can also indicate renal prognoses; however, a single-point estimate is insufficient because ESA resistance increases with CKD progression. We used the BRIGHTEN data for a sub-analysis and found that the ESA resistance index (ERI) (darbepoetin alpha dose/hemoglobin level) transition pattern might be more appropriate for predicting renal prognoses than an initial single-point ERI value. This BRIGHTEN study sub-analysis included 1,625 patients with complete data, comprising 518 dialysis initiations and 18 renal transplants, with a mean follow-up of 1.79 years. The relationship between disease progression and ESA resistance was analyzed using the joint latent class model of longitudinal and time-to-event data. The Kaplan–Meier method was used to estimate the cumulative incidence for the three classes according to the transition pattern of the ERIs. The patient background characteristics in each class were compared. The ERI transition pattern was classified into 3 classes: ERI unchanged, Class A (1,237 patients); moderately increased, Class B (274 patients); and rapidly increased, Class C (114 patients). Class C showed a significantly worse renal survival curve than the other groups, while Class B showed an intermediate survival curve (P