eGFR slope as a surrogate endpoint for clinical study in early stage of chronic kidney disease: from The Japan Chronic Kidney Disease Database

Abstract

Background: In clinical trials targeting early chronic kidney disease (CKD), eGFR slope has been proposed as a surrogate endpoint for predicting end-stage kidney disease (ESKD). However, it is unclear whether the eGFR slope serves as a surrogate endpoint for predicting long-term prognosis in Japanese early CKD populations.

Methods: The data source was the J-CKD-Database, which contains real-world data on patients with CKD in Japan. eGFR slope was calculated from the eGFR of each period, 1-year (1-year slope), 2-year (2-year slope), and 3-year (3-year slope), for participants with a baseline eGFR ≥ 30 ml/min/1.73 m². The outcome was ESKD (defined as dialysis initiation or incidence of CKD stage G5). The relationship between eGFR slope and the sub-distribution hazard ratio (SHR) of ESKD with death as a competing event was investigated using a Fine-Gray proportional hazard regression model.

Results: The number of participants and mean observation periods were 7768/877 ± 491 days for 1-year slope, 6778/706 ± 346 days for 2-year slope, and 5219/495 ± 215 days for 3-year slope. As the eGFR slope decreased, a tendency toward a lower risk of ESKD was observed. Compared with the 1-year slope, there was a smaller variation in the slope values for the 2-year or 3-year slope and a greater decrease in the SHR; therefore, a calculation period of 2 or 3 years for the eGFR slope was considered appropriate.

Conclusion: Even in Japanese patients with early stage CKD, a slower eGFR slope calculated from eGFR values over 2-3 years was associated with a decreased risk of ESKD.

Keywords: Chronic kidney disease; End stage kidney disease; Surrogate endpoint; eGFR slope.

Fig. 1

Calculation period and observation period for eGFR slope. The extraction period for J-CKD-DB-Ex was from January 1, 2014, to December 31, 2018, and data for up to five years were extracted. The 1-year slope was calculated using data from days 1 to 365 for participants with eGFR data for 366 days or more from baseline. The 2-year slope was calculated using data from days 1 to 730 for participants with eGFR data for 731 days or more from baseline. The 3-year slope was calculated using data from day 1 to day 1095 for participants with eGFR data for 1096 days or more from baseline. The observation period was defined as the period from the end of the eGFR slope calculation to the occurrence of the outcome or final eGFR measurement

Fig. 2

Distribution of slope values for each eGFR slope calculation period. The number of participants for each eGFR slope value is shown as a histogram. (a) 1-year slope calculated using the least-squares method, (b) 2-year slope calculated using the least-squares method, (c) 3-year slope calculated using the least-squares method, (d) 1-year slope calculated using the mixed-effects model, (e) 2-year slope calculated using the mixed-effects model, and (f) 3-year slope calculated using the mixed-effects model

Fig. 3

Adjusted sub-distribution hazard ratios for ESKD occurrence by change in eGFR slope. For each of the 1–3 year periods of the eGFR slope, the sub-distribution hazard ratios (SHRs) and 95% confidence intervals (CIs) for ESKD occurrence were shown for the range of change in eGFR slope of + 0.25 to + 1.50 ml/min/1.73 m². (a) Adjusted SHRs for dialysis initiation and (b) Adjusted SHRs for the incident of CKD stage G5. The estimation of SHRs was performed using a Fine-Gray proportional hazards regression model, with death as a competing risk. The multivariate analysis was adjusted for age, sex, eGFR, hemoglobin, serum albumin, C-reactive protein, antihypertensive medication prescription, renin-angiotensin system inhibitor prescription, and diabetes mellitus