This study is based on data from previously conducted studies and does not include any new human participants or patients enrolled in this study. Therefore, this study complies with ethical guidelines and did not require an ethics review. The FINE-CKD model uses a Markov modeling approach to compare the cost-effectiveness of adding finerenone to SoC versus SoC alone for the treatment of CKD in patients with T2D. The Markov approach provides a framework for modeling the relationship between CKD progression and the incidence of CV events in regular cycles over a period. Finerenone reduces the risk of clinically important CV and renal outcomes among patients with CKD and T2D [24] and, thus, the study uses the Markov modeling approach to focus on two important dimensions of the clinical value of finerenone.

The modeled population was based on the intention-to-treat (ITT) population from the FIDELITY analysis, with a mean age of 64.8 years; the proportion of male patients was 69.8% [24]. The trials predominantly included patients with CKD and T2D across the spectrum of CKD severity. Most patients (53.3%) were diagnosed with CKD stage 3 comorbid with T2D, excluding patients undergoing ESKD or dialysis (Table 1).

Finerenone was added to SoC and compared with SoC alone. SoC was based on the weighted average of background treatment over the time horizon. Patients with CKD and T2D used RAAS blockers, including ACEi or ARBs, as first-line treatment options.

Model health states were defined according to the stage of kidney disease and history of CV events (Fig. 1). Patients changed health states after experiencing one of the main health events, defined on the basis of the key outcomes of finerenone clinical trials and covering transitions between CKD stages, starting renal replacement therapy, as well as selected CV events and mortality. Additionally, the model included other health events (recurrence of CV events, sustained decrease in eGFR of at least 40% from baseline, new onset of atrial fibrillation/flutter, and hyperkalemia), encompassing a range of clinically meaningful outcomes as an additional layer.

Model structure. CKD chronic kidney disease, CV cardiovascular, HF heart failure, MI myocardial infarction, RRT renal replacement therapy

Four stages of CKD progression were considered: CKD 1/2, CKD 3, CKD 4, and CKD 5, based on eGFR levels and without renal replacement therapy. In addition, two stages were considered for ESKD after dialysis and post-renal transplantation (Fig. 1). Patients entered the model in one of the CKD stages in the presence or absence of CV events. Modeled patients remained at the same CKD stage or moved to a more/less advanced CKD stage and/or experienced a first modeled CV event (non-fatal myocardial infarction, non-fatal stroke, and hospitalization for heart failure) or death. Once patients experienced a first CV event, they moved to the post-CV event health state and were not able to move back to the health state without CV events. At any point in the model, patients could experience death, which was appropriately disaggregated to capture the different causes of death (CV or renal).

Patients remained in the same CKD stage for a cycle duration of 4 months, based on the duration of evaluating outcomes in the pivotal trials [21, 22]. The analysis was performed for a lifelong time horizon (up to 100 years of age) to account for the chronic nature of CKD in the context of T2D. A half-cycle correction was applied in the model. Both costs and outcomes were discounted at 2.0% [27]. The cost-effectiveness model calculated quality-adjusted life years (QALYs), total costs, and the incremental cost-effectiveness ratio (ICER) per QALY gained. A willingness-to-pay (WTP) threshold of Japanese yen (¥) 5 million per QALY gained was used to assess the cost-effectiveness results [28].

The health-state transition probabilities for SoC were estimated from the FIDELITY data analysis [24] (Supplementary Table S1). Transition probabilities included CKD progression, development of ESKD, incidence of the first CV event (transition to the corresponding health state with CV event), both the incidence of the first CV event and deterioration in renal function, and death (renal death, CV death, and death from other causes) (Supplementary Table S2). The transition probabilities for finerenone were estimated by applying the relevant hazard ratios (HRs), assumed to be constant over time. HRs were sourced from the FIDELITY analysis [24] (Supplementary Table S3). HRs were applied to the first and subsequent CV events.

For other health events, transition probabilities for SoC were sourced separately for patients with or without CV events from the FIDELITY analysis [24]. For finerenone, the risk for other health events was estimated by applying the HRs sourced from the FIDELITY analysis [24] (Supplementary Table S4).

The model applied a two-step approach to calculate the utilities. First, the results of a multivariate regression (multilevel mixed repeated measurements) model, based on FIDELITY, were used to estimate the utility values for main health events and utility decrements for other health events. Next, the age adjustment multiplier was included to better reflect the impact of age on baseline utility [29] (Supplementary Table S5).

The model considered only direct medical costs, including costs for medications, health states, and other health events (Table 1 and Supplementary Table S6), and assumed that patients received the same dose of medication for the duration of the model. Costs of other health events were applied to the proportion of patients experiencing those events to calculate total average costs over the time horizon of the model. All cost-related data were obtained by performing a cost analysis from the MDV (Medical Data Vision Co., Ltd.) database [30] and identifying the cohort of patients between October 2013 and August 2019 with at least 1 year of follow-up data for CKD and T2D. Healthcare resource utilization data and the costs of all four CKD stages (CKD 1/2, 3, 4, and 5), dialysis (hemodialysis and peritoneal dialysis), kidney transplantation (acute and post-acute), CV events (myocardial infarction, cerebral infarction, intracerebral hemorrhage, and hospitalization due to heart failure), death (CV and renal death), and other health events were obtained from this cost analysis (Supplementary Table S6).

The price per day for finerenone was calculated as a weighted average of the prices of two different doses (10 mg and 20 mg), according to the percentage of their use in FIDELITY [24] (Supplementary Table S7). For SoC, all commonly used therapies among patients with CKD with T2D in Japan were included and based on the claims database analysis [30]. The cost of SoC was the sum of all treatments weighted by the percentage of patients who used each therapy in the FIDELITY analysis [24] (Supplementary Table S7). All drug costs were obtained from the drug price list, April 2022 [31].

The base-case analysis compared finerenone plus SoC with SoC alone from the Japanese payer perspective. The cost for finerenone considered the brand drug price, and the total ITT population from FIDELITY was taken into consideration.

One-dimensional sensitivity analysis was performed to assess the uncertainty of the results due to variations in individual parameters. The analysis was performed with a 95% confidence interval (CI) based on the probability distribution set for each parameter. A variation of 20% of the average estimate was assumed where parameters did not achieve the 95% CI. The top ten parameters with the greatest impact on the ICER were depicted as a tornado diagram.

A probabilistic sensitivity analysis using 1000 Monte Carlo simulations was performed to assess the uncertainty of the analytical results. The probability distribution and range of parameters assumed in the stochastic sensitivity analysis were based on their parameter distribution. A scatter plot of the cost-effectiveness plane based on the results of the probabilistic sensitivity analysis and the cost-effectiveness acceptability curve was used to depict the results.

The model considered two scenarios: First, the price of the SoC was set to the cost of the generic drug (Supplementary Table S7). Second, only the Japanese patient population from the FIDELITY analysis was included in the model instead of the overall ITT population. The model inputs for this scenario are shown in Supplementary Tables S8–S12.