Performance of the updated 3C-HF risk score across ejection fraction phenotypes in acute heart failure: a multicohort study with external validation
E Fiori, C Basile, L Fazzini, C Giaccherini, A Gavazzi, E D'elia, E Barbato, M Senni, G Savarese, M GoriAbstract
Background
Risk stratification in patients hospitalized with acute heart failure (AHF) is critical for tailoring post-discharge follow-up. There is limited evidence on whether scores derived and validated in chronic HF maintain their prognostic role in AHF and across the EF spectrum.
Purpose
We aimed to repurpose the Cardiac and Comorbid Conditions Heart Failure (3C-HF) score for the prediction of 1-year survival in the AHF setting across the EF spectrum. We also assessed the incremental prognostic value of variables not included in the score, as mineralocorticoid receptor antagonist (MRA) therapy, natriuretic peptides, and the neutrophil-to-lymphocyte ratio (NLR).
Methods
Data from consecutive patients hospitalized for AHF at a tertiary Italian center in the OPPORTUNITIES registry were retrospectively collected between January 2017 and June 2023. Patients with AHF from the Swedish HF registry, with no missing data for the variables needed to calculate the 3C-HF score, served as the comparative cohort. EF was dichotomized at 40% to define HF with reduced EF (HFrEF) and HF with preserved EF (HFpEF). BNP and NT-proBNP data were available in the Italian and Swedish cohorts, respectively. Prognostic performance was assessed using Harrell’s C-index. Kaplan-Meier curves with log-rank testing evaluated risk stratification across 3C-HF tertiles. Incremental and independent prognostic value was estimated using likelihood-ratio testing and multivariable Cox regression models.
Results
In the Italian (n=1,030) and Swedish (n=1,570) cohorts, HFrEF comprised 567 (55%) and 1,002 (64%) patients, while HFpEF comprised 463 (45%) and 568 (36%) patients, respectively. In the two cohorts KM analysis across 3C-HF tertiles showed decreasing survival together with increasing score value in both HFrEF and HFpEF (Figure 1). The 3C-HF score demonstrated good prognostic discrimination in HFrEF (C-index 0.77) and modest discrimination in HFpEF (C-index 0.68), with consistent performance across the two cohorts. In the Italian cohort, MRA therapy was not associated with survival. BNP was associated with the outcome only in HFrEF, and its addition to the 3C-HF score modestly improved model fit in this phenotype (C-index 0.78; LRT p<0.001). In the Swedish cohort, NT-proBNP was associated with survival across EF phenotypes and provided incremental prognostic discrimination in both HFrEF and HFpEF. NLR, available only in the derivation cohort, was associated with survival only in HFpEF, and its addition to the 3C-HF score significantly improved model fit (C-index 0.68 vs 0.70, LRT p<0.001). After multivariable adjustment, the 3C-HF score remained independently associated with survival irrespective of HF phenotype in both cohorts (Figure 2).
Conclusions
The 3C-HF score is independently associated with 1-year survival after hospitalization for AHF across the full spectrum of EF. Natriuretic peptides and NLR provide only modest additional, phenotype-specific, prognostic information.KM analysisFor image description, please refer to the figure legend and surrounding text.Prognostic performanceFor image description, please refer to the figure legend and surrounding text.