A parsimonious echocardiography-based model for long-term mortality risk stratification in heart failure with preserved ejection fraction
L Y Mi, J S K Chan, W T Wong, G Tse, F FangAbstract
Background
Heart failure with preserved ejection fraction (HFpEF) is heterogeneous with substantial long-term mortality risk. Diagnostic algorithms such as HFA-PEFF and H2FPEF facilitate HFpEF diagnosis but may not optimally stratify mortality risk.
Purpose
To develop and internally validate a parsimonious, echocardiography-based model to predict long-term all-cause mortality in HFpEF and compare its prognostic performance head-to-head with HFA-PEFF and H2FPEF.
Methods
Adults with HFpEF (left ventricular ejection fraction 50% or higher) diagnosed between 2010 and 2016 from a real-world echocardiography database linked to longitudinal electronic health records were included (n=792) and randomly split into training (n=554) and internal validation (n=238) cohorts. The primary endpoint was all-cause mortality. Candidate predictors were screened using Cox models; a parsimonious model was derived using LASSO-penalised Cox regression and refitted as a multivariable Cox model, presented as a nomogram.
Results
The final nomogram incorporated five routinely measured predictors: age, LV posterior wall thickness at end-systole, mitral E velocity, E/e’ ratio, and pulmonary artery systolic pressure. During median follow-up of 5.17 years (IQR 2.26–9.14) and 5.75 years (IQR 2.17–9.25), 393/554 (70.9%) and 165/238 (69.3%) deaths occurred in training and validation cohorts, respectively. In the validation cohort, the nomogram outperformed HFA-PEFF and H2FPEF for discrimination (C-index 0.860 vs 0.815 and 0.803; time-dependent AUCs at 1/3/5 years: 0.658/0.706/0.713 vs 0.507/0.561/0.642 and 0.516/0.533/0.607). Calibration at 1, 3, and 5 years was acceptable. Risk-score tertiles separated survival in both cohorts (log-rank P<0.001).
Conclusion
A compact echocardiography-based nomogram enables robust long-term mortality risk stratification in HFpEF and provides superior prognostic discrimination compared with two commonly used diagnostic scores. This tool may support pragmatic risk communication and risk enrichment in routine care and research settings.For image description, please refer to the figure legend and surrounding text.For image description, please refer to the figure legend and surrounding text.