Incidence of new-onset HFpEF across CKD stages and Its association with cardio-renal, inflammatory, and fibrotic biomarkers

Background Chronic kidney disease (CKD) substantially increases the risk of heart failure with preserved ejection fraction (HFpEF). In this study, we quantified prospective, stage-specific incidence and evaluated whether biologically anchored, multimarker domains improve prediction beyond clinical factors. Methods We conducted a single-center, prospective cohort study (2019–2024) of adults with CKD G1–G5 (not on dialysis) with a 36-month follow-up. Incident HFpEF was adjudicated by two cardiologists using Heart Failure Association—Pre-test assessment, Echocardiography and natriuretic peptide score, Functional testing, Final aetiology (HFA-PEFF) criteria with atrial fibrillation–adjusted natriuretic peptide thresholds, and death was treated as a competing event. Baseline biomarkers were grouped into five domains—cardiac stretch, myocyte injury, fibrosis/remodeling, renal injury/function, and inflammation—and standardized to domain z -scores. We estimated incidence per 100 person-years and adjusted associations using Cox and Fine–Gray models. Incremental value was assessed with changes in the C-statistic ( Δ C), continuous net reclassification improvement (NRI), integrated discrimination improvement (IDI), and calibration, with all metrics bootstrap-validated and false discovery rate control applied for domain tests. Results Among 920 participants (mean age 64 ± 12 years; 43% women), 68 HFpEF events occurred over 2,582 person-years, corresponding to 2.63 [95% confidence intervals (CI) 2.01–3.26] events per 100 person-years. Incidence increased with CKD stage—G1 0.29%, G2 1.16%, G3a 1.76%, G3b 3.14%, G4 4.46%, and G5 6.93%—and with albuminuria—A1 1.32%, A2 2.42%, and A3 4.87%. Adjusted hazard ratios were 1.32 (1.17–1.49) per CKD stage and 1.41 (1.20–1.66) per albuminuria category. Per 1 SD, domains associated with higher risk were stretch 1.45 (1.23–1.72), fibrosis/remodeling 1.34 (1.13–1.59), injury 1.26 (1.07–1.48), renal 1.22 (1.04–1.43), and inflammation 1.17 (0.99–1.37). Adding biomarker domains improved discrimination from 0.72 to 0.79 ( Δ C 0.07, 95% CI 0.03–0.10), with continuous NRI 0.20 and IDI 0.055. Conclusions In CKD, HFpEF incidence increases stepwise with CKD stage and albuminuria. Domain-based biomarker profiling, particularly cardiac stretch and fibrosis/remodeling, provides prognostic information beyond clinical factors, supporting a biologically anchored and parsimonious approach to risk stratification.