OBJECTIVES: Finerenone has shown to reduce the risk of cardiorenal events in patients with chronic kidney disease and type 2 diabetes, irrespective of sodium-glucose cotransporter 2 inhibitor (SGLT2i) use. Enrollment of SGLT2i users in cardiorenal trials has remained low due to their limited use in this population over the past years, impacting statistical power. To increase precision, we aimed to augment the pooled SGLT2i subgroups from the FIDELIO-DKD and FIGARO-DKD randomized clinical trials (RCTs) with an external control arm (ECA), comprised of matching patients from US Optum electronic health records (EHR), to recalculate effect estimates.

METHODS: We extracted EHR patients if they had at least one qualifying estimated glomerular filtration rate (25-90 ml/min/1.73m²) and one urine albumin-creatinine ratio value (≥30 mg/g) separated by 90 to 540 days and if other translated RCT selection criteria were satisfied. A constrained optimization algorithm was used to match these patients to the RCT subgroup on 42 covariates. Incidence rates (IR) and hazard ratios (HR) of the kidney and cardiovascular (CV) composite endpoints were calculated for the ECA-augmented study population.

RESULTS: We successfully matched 877 patients to the RCT subgroup, achieving a 1:3 treatment-to-control ratio (median standardized mean difference <0.01). IRs per 100 patient-years were comparable between the ECA and the RCT controls, with 4.12 [95% CI 3.35, 5.02] vs. 4.08 [3.05, 5.35] (p=0.955), and 1.04 [0.68, 1.53] vs. 1.37 [0.80, 2.19] (p=0.383) for the CV and kidney composite endpoints, respectively. Recalculation of the ECA-augmented treatment effect estimates between patients using the Finerenone + SGLT2i combination vs. those using SGLT2is alone lead to higher precision in the evaluated composite endpoints, CV: HR 0.70 [95% CI 0.49, 0.99] (p=0.048) and kidney: 0.66 [0.32, 1.36] (p=0.263).

CONCLUSIONS: ECA-augmentation of the pooled SGLT2i subgroups from cardiorenal trials was feasible and enabled increased precision for the treatment effect estimates.