Guidelines for clinical evaluation of chronic kidney disease in early stages : AMED research on regulatory science of pharmaceuticals and medical devices

Abstract

Background: For the development of pharmaceutical products in kidney field, appropriate surrogate endpoints which can predict long-term prognosis are needed as an alternative to hard endpoints, such as end-stage kidney disease. Though international workshop has proposed estimated glomerular filtration rate (GFR) slope reduction of 0.5-1.0 mL/min/1.73 m /year and 30% decrease in albuminuria/proteinuria as surrogate endpoints in early and advanced chronic kidney disease (CKD), it was not clear whether these are applicable to Japanese patients.

Methods: We analyzed J-CKD-DB and CKD-JAC, Japanese databases/cohorts of CKD patients, and J-DREAMS, a Japanese database of patients with diabetes mellitus to investigate the applicability of eGFR slope and albuminuria/proteinuria to the Japanese population. Systematic review on those endpoints was also conducted including the results of clinical trials published after the above proposal.

Results: Our analysis showed an association between eGFR slope and the risk of end-stage kidney disease. A 30% decrease in albuminuria/proteinuria over 2 years corresponded to a 20% decrease in the risk of end-stage kidney disease patients with baseline UACR ≥ 30 mg/gCre or UPCR ≥ 0.15 g/gCre in the analysis of CKD-JAC, though this analysis was not performed on the other database/cohort. Those results suggested similar trends to those of the systematic review.

Conclusion: The results suggested that eGFR slope and decreased albuminuria/proteinuria may be used as a surrogate endpoint in clinical trials for early CKD (including diabetic kidney disease) in Japanese population, though its validity and cutoff values must be carefully considered based on the latest evidence and other factors.

Keywords: Albuminuria; Chronic kidney disease; Diabetic kidney disease; Proteinuria; Surrogate endpoint; eGFR slope.

Fig. 1

Distribution of baseline estimated glomerular filtration rate (eGFR) and urinary albumin-creatinine ratio (UACR) in the population for whom 2-year slope could be calculated in the J-DREAMS cohort. The analysis population included a significant number of patients with eGFR ≥ 60 mL/min/1.73 m² or UACR < 30 mg/gCre, and only 53.8% of patients with eGFR < 60 mL/min/1.73 m² or UACR ≥ 30 mg/gCre were considered to have diabetic kidney disease (orange area)

Fig. 2

The association between the composite end-stage kidney disease (ESKD) events and 2-year estimated glomerular filtration rate (eGFR) slope. A Population distribution of change in eGFR slope and the association between eGFR slope and composite ESKD events. The upper panel shows the spline curve of the association between composite ESKD events and the 2-year eGFR slope. The lower panel shows the distribution of the number of cases in whom 2-year eGFR slopes were calculated. The green line corresponds to the slope calculated under the mixed-effects (ME) model, and the blue line corresponds to the slope calculated under ordinary least-squares (OLS) methods. B Association between composite ESKD risk and eGFR slope: subgroup analysis. Composite ESKD risk associations with eGFR slope reduction of 0.75 mL/min/1.73 m²/ year calculated under a ME model. eGFR slope was calculated over 2 years. HRs and 95% CIs for each subgroup divided by eGFR (G1-2: ≥ 60 mL/min/1.73 m², G3: 30–60 mL/min/1.73 m²), or urine albumin-creatinine ratio (A1: < 30 mg/g Cr, A2-3, ≥ 30 mg/g Cr) were shown. CI confidence interval, HR hazard ratio

Fig. 3

Adjusted sub-distribution hazard ratios (SHRs) for end-stage kidney disease (ESKD) occurrence by the change in estimated glomerular filtration rate (eGFR) slope. For each of the 1–3-year periods of the eGFR slope, the 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/year. A Adjusted SHRs for dialysis initiation and B adjusted SHRs for incident CKD stage G5. SHRs were estimated using a fine-gray proportional hazards regression model, with death as a competing risk. Multivariate analysis was adjusted for age, sex, eGFR, hemoglobin, serum albumin, C-reactive protein, antihypertensive medication, renin-angiotensin system inhibitor, and diabetes mellitus

Fig. 3

Adjusted sub-distribution hazard ratios (SHRs) for end-stage kidney disease (ESKD) occurrence by the change in estimated glomerular filtration rate (eGFR) slope. For each of the 1–3-year periods of the eGFR slope, the 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/year. A Adjusted SHRs for dialysis initiation and B adjusted SHRs for incident CKD stage G5. SHRs were estimated using a fine-gray proportional hazards regression model, with death as a competing risk. Multivariate analysis was adjusted for age, sex, eGFR, hemoglobin, serum albumin, C-reactive protein, antihypertensive medication, renin-angiotensin system inhibitor, and diabetes mellitus