Systemic immune-inflammation index as a prognostic marker in heart failure with non-dialysis chronic kidney disease: a two-cohort derivation and validation study with head-to-head comparison against natriuretic peptides

Introduction Heart failure (HF) and chronic kidney disease (CKD) frequently coexist and share chronic low-grade inflammation, oxidative stress, and thrombotic activation. Natriuretic peptides remain the cornerstone of HF prognostication, but their interpretation is confounded by reduced renal clearance, leading to false elevations and attenuated prognostic specificity in CKD. The systemic immune-inflammation index (SII), derived from routine blood counts, integrates neutrophil-driven inflammation, lymphocyte-mediated immunity, and platelet-associated thrombosis, and may provide a renal-function–independent adjunct to natriuretic peptides among patients with non-dialysis CKD. We aimed to evaluate whether SII provides incremental prognostic information to BNP and NT-proBNP in HF, with particular attention to the non-dialysis cardiorenal spectrum. Methods We conducted a two-cohort study. The derivation cohort included 1,284 adults hospitalized for chronic HF at Guangdong Medical University Affiliated Hospital (2018–2023). The external validation cohort comprised adults with a first HF hospitalization from MIMIC-IV v3.1 (2008–2022). After excluding patients without SII data and those with incomplete covariates, 11,767 patients were included in the primary analysis. The systemic immune-inflammation index (SII), a composite inflammatory marker derived from routine complete blood count, was calculated as platelet count (×10 9 /L) × neutrophil count (×10 9 /L) / lymphocyte count (×10 9 /L). Patients were stratified by CKD-EPI estimated glomerular filtration rate (eGFR) into >=60, 30–59, and 2 . Cox proportional hazards regression, C-index comparisons, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and SII × CKD-stage interaction tests were performed. Results SII predicted mortality in the derivation cohort (HR 1.012, 95% CI 1.007–1.017, P  = 7.5 × 10⁻⁶) and added modest incremental value to NT-proBNP. Conclusion SII is an inexpensive, universally available prognostic marker in HF patients with non-dialysis CKD. It does not replace NT-proBNP, but provides a statistically significant yet clinically modest incremental signal in CKD—across the non-dialysis CKD spectrum. SII may serve as a practical adjunct, not a substitute, for risk stratification across the non-dialysis cardiorenal spectrum.