Abstract: FR-PO490

Importance of eGFR Change as a Surrogate End Point of ESRD in a Randomized Controlled Trial

Session Information

Category: Chronic Kidney Disease (Non-Dialysis)

  • 304 CKD: Epidemiology, Outcomes - Non-Cardiovascular


  • Kanda, Eiichiro, Tokyo Kyosai Hospital, Meguro, TOKYO, Japan
  • Imai, Enyu, Nakayamadera Imai Clinic, Takarazuka, Hyōgo, Japan
  • Kobayashi, Fumiaki, Daiichi Sankyo Inc, Somerset, New Jersey, United States
  • Kashihara, Naoki, Kawasaki Medical School, Kurashiki City, Japan
  • Nangaku, Masaomi, the University of Tokyo School of Medicine, Tokyo, Japan

A use of a validated surrogate endpoint instead of a clinical endpoint could make a sample size small and shorten trial period. We evaluated a usefulness of an estimated glomerular filtration rate (eGFR) change as a surrogate endpoint using data from the randomized controlled clinical trial [Olmesartan Reducing Incidence of End-Stage Renal Disease (ESRD) in Diabetic Nephropathy Trial].


ESRD was defined as the true endpoint. eGFR changes over 1 to 3 years by 10% were defined as the surrogate endpoints. The relationship between eGFR changes and ESRD or the surrogate endpoints was evaluated using Cox proportional hazard models, which were adjusted for baseline characteristics. An effect of olmesartan on ESRD was compared with those on surrogate endpoints in terms of the ratio of the adjusted hazard ratio (aHR) of olmesartan to ESRD to those to surrogate endpoints by the bootstrap method.


Diabetic kidney disease patients (n=566; male, 69.1%) were included in this analysis; average age±SD, 59.1±8.1 years; eGFR 37.1±9.8 ml/min. The Cox proportional hazard models with spline curves showed the relationships between eGFR changes over 1 and 2 years and ESRD (Figure 1). The ratios of aHR to ESRD to aHRs to surrogate endpoints near 1 were -30% or -40% over 2 years: -30%, 1.01 (95%CI 0.69-1.41); -40%, 1.0 (0.71-1.34). The events of eGFR changes of more than -30% showed good agreement with ESRD (0.6 < Cohen’s kappas). The total sample sizes were 1004 for estimating eGFR change of -40% over 1 year; 800 for that of -40% over 2 years, which were smaller than that (1922) for one estimating ESRD.


eGFR change of more than -30% to -40% over 2 years would be a useful surrogate endpoint of ESRD in DKD patients.