#2546 CKD-PC risk model for CKD progression: validation and association with outcomes in the FIDELITY population

Abstract

Abstract Background and Aims The Chronic Kidney Disease Prognosis Consortium (CKD-PC) recently developed risk models to predict the 3-year risk of a ≥40% decrease in estimated glomerular filtration rate (eGFR) or to predict kidney failure in the general population. Unlike the kidney failure risk equation, the CKD-PC model is valid for all levels of eGFR (above and below 60 mL/min/1.73 m2) and includes chronic kidney disease (CKD) progression risk factors besides age, sex, eGFR and albuminuria; namely: systolic blood pressure, antihypertensive medication use, history of heart failure, coronary heart disease, atrial fibrillation, smoking status, body mass index, glycated hemoglobin, insulin use, and oral diabetes medication use [1]. We evaluated whether the CKD-PC model, developed using electronic health record data, is valid for use in a contemporary clinical population using the FIDELITY pooled dataset from international randomized clinical trials of finerenone [2]. Additionally, to inform clinical practice, we evaluated the efficacy of finerenone using categories of 3-year CKD progression risk as predicted by the model. Method FIDELITY was an individual, patient-level, prespecified, pooled efficacy and safety analysis of the phase III FIDELIO-DKD and FIGARO-DKD trials of patients with CKD and type 2 diabetes (T2D), randomized 1:1 to finerenone or placebo. The 3-year CKD progression risk of these patients was predicted and the validity of the CKD-PC model was evaluated using area under the curve (AUC) score (with higher scores indicating better risk prediction), Brier score (lower scores indicating better prediction), and by calibration plots of observed versus predicted risk by deciles of 3-year CKD progression risk. To evaluate the efficacy of finerenone, 3-year risk was categorized in quartiles and the effect of finerenone versus placebo on a kidney composite outcome, defined as kidney failure, a sustained ≥40% decrease in eGFR from baseline over ≥4 weeks, or renal death, was assessed. Results The FIDELITY population comprised 13,026 individuals with a mean eGFR of 58 mL/min/1.73 m2 and urine albumin-to-creatinine ratio of 515 mg/g. The median follow-up was 3 years and there were 1849 total kidney outcome events. Of 12,968 patients analyzed, the median 3-year risk of kidney failure was 15% (Q1–Q3: 9–24%). Overall, the model demonstrated good discrimination to predict a ≥40% kidney composite outcome for patients with CKD and T2D in FIDELITY: AUC score 0.726 (95% confidence interval [CI] 0.712–0.739) and Brier score 0.100 (95% CI 0.097–0.103). The predicted 3-year risk was generally slightly higher than the observed risk (Fig. 1). Finerenone was effective in preventing CKD progression across the risk quartiles (p-interaction = 0.094) with a trend towards greater efficacy in the higher risk quartiles (3-year risk of CKD progression ≥10%) (Fig. 2). All risk groups showed benefit with finerenone compared to placebo regarding eGFR decline in the chronic phase. Conclusion The CKD-PC risk model for CKD progression performed well in a large global clinical trial population of patients with CKD and T2D. Finerenone was effective across the range of baseline risk of CKD progression.