From glomerular filtration rate to renal composite endpoint: an integrated time-to-event analysis of finerenone in phase 3 studies.
Martijn van Noort (1), Martijn Ruppert (1), Sebastiaan C. Goulooze (1), Paul van den Berg (1), Nelleke Snelder (1), Hiddo J. Heerspink (2), Dirk Garmann (3), Jörg Lippert (3), Roland Heinig (3), Meike Brinker (3), Thomas Eissing (3)
(1) LAP&P Consultants, Leiden, The Netherlands (2) University of Groningen, Groningen, The Netherlands (3) Bayer AG, Leverkusen/Wuppertal, Germany
Objectives: Finerenone (Kerendia) is used to treat chronic kidney disease in patients with type 2 diabetes. In addition to effects on cardiorenal endpoints, finerenone affects urine albumin-to-creatinine ratio (UACR) and estimated glomerular filtration rate (eGFR), two surrogate biomarkers for kidney failure [1,2]. A previously developed PKPD model characterized finerenone’s effects on UACR and eGFR in the Phase 3 studies FIDELIO-DKD and FIGARO-DKD. In this analysis we aim to bridge between eGFR over time and the renal endpoint by using predictions from the eGFR biomarker model in a time-to-event (TTE) analysis of the renal composite endpoint.
Methods: This study analysed the renal composite endpoint data from two randomised, placebo-controlled Phase 3 studies: FIDELIO-DKD and FIGARO-DKD [3,4]. The renal composite endpoint consisted of time to first (1) kidney failure, (2) sustained decrease in eGFR ≥57% versus baseline or (3) renal death. For each subject, either an interval-censored event (between two consecutive eGFR observations) or a right-censored data point (at the end of the follow-up time) was included in the analysis dataset. All modelling was performed in NONMEM 7.5.0.
For each subject, the individual posthoc predicted eGFR over time (and derived metrics, such as relative change from baseline) were used as potential predictors of the hazard for experiencing the renal composite event. The predicted eGFR was obtained from a previously developed PKPD model of UACR and eGFR in FIDELIO-DKD [2] and FIGARO-DKD. The UACR model feeds into the eGFR model, as model-predicted UACR is included as a strong predictor of the rate of chronic eGFR decline. Finerenone has three treatment effects in these models: 1) a sustained reduction of UACR, 2) an acute (and reversible upon treatment discontinuation) decline of eGFR, and 3) a reduction in the rate of chronic eGFR decline. The first two effects are characterized with power functions linked to the exposure in an effect compartment, while the latter effect in the eGFR model is fully characterized (or mediated) via finerenone’s reduction of UACR.
Based on the role of eGFR in the first two elements of the renal composite endpoint, it was hypothesized that a TTE analysis could be used to bridge the model-predicted eGFR to the renal endpoint by characterizing the link between eGFR and TTE, via optimization among a set of potential effect drivers and effect functions. A covariate analysis was performed to explore if there were any eGFR-independent effects of finerenone or covariates on the hazard of the renal endpoint.
Results: A total of 13,026 subjects were included in the analysis, in which 75,503 UACR observations and 186,756 eGFR observations were collected to inform the individual model-predicted eGFR over a period of up to 5 years.
The TTE analysis of the renal outcome data in these 13,026 subjects identified two eGFR-dependent effects, (1) increased hazard with increasing relative decrease in eGFR versus baseline (2) increased hazard with decreasing absolute eGFR. The first effect was dependent on the time: for example, a patient with a 40% eGFR decrease compared to baseline at 2 years would have a higher hazard compared to a patient with 40% decrease at 4 years. We were not able to identify any eGFR-independent effects of covariates or finerenone treatment on the renal outcome.
Conclusions: The two eGFR-dependent effects in the renal TTE analysis intuitively represent different components in the definition of the renal composite outcome; the effect driven by relative decrease in eGFR explaining the increased hazard of triggering a sustained decrease in eGFR ≥57%, while the effect driven by low absolute eGFR explains the increased hazard of triggering the event of kidney failure as eGFR gets closer to 15 ml/min/1.73m2.
The fact that eGFR effects in the model could capture the effects of both finerenone treatment and known covariates on the renal outcome, i.e. no eGFR-independent effects were identified, supports eGFR as a surrogate for the renal outcome. Additionally, this finding also implies relevance of the covariate effects identified in the UACR/eGFR PKPD models to the renal composite endpoint.
References:
[1] Levey et al. Am J Kidney Dis (2020) 75 (1) 84-104.
[2] Goulooze et al. Clin Pharmacokinet (2022) 61 (7) 1013-1025.
[3] Bakris et al. N Engl J Med (2020) 383 (23) 2219-2229.
[4] Ruilope et al. Nephrol Dial Transplant (2023) 38 (2) 372-383.