Predictors of heart failure with improved ejection fraction: a real world analysis
M Moreira, J L Ferraro, I Bastos Castro, I Gomes Campos, A Rodrigo Costa, J Ponte Monteiro, I Almeida, A Leal Neto, A Pereira, P Silva, A AndradeAbstract
Introduction
Heart failure with improved ejection fraction (HFimpEF) is increasingly recognized as a meaningful phenotype. The factors associated with this improvement warrants further investigation.
Methods
Single-center retrospective study with consecutive patients with chronic heart failure with reduced ejection fraction (HFrEF) from an HF clinic between 2014-2024. Previous diseases, medication, clinical status, biomarkers, electrocardiogram and echocardiogram findings were recorded. HFimpEF (left ventricle ejection fraction ≥40% with improvement of ≥10% from baseline) was the endpoint at 1-year post-admission. Backwards Wald logistic regression was used to estimate independent predictor factors of HFimpEF.
Results
We included 292 patients (73.3% male; mean age 59.9±11.8 years-old), ischemic etiology in 35.3%, median LVEF of 25.0±10.2%. HFimpEF occurred in 24.0% of patients. It was more prevalent in current smokers (25.7% vs 11.7%; p=0.013); atrial fibrillation (35.7% vs 20.0%; p<0.001); absence of coronary artery disease (CAD) (85.7% vs 49.4%; p<0.001); non-ischemic etiology (88.6% vs 58.3%; p<0.001); shorter QRS (117.9ms vs 131.0ms; p=0.002), absence of late gadolinium enhancement (LGE) (53.3% vs 33.8%; p=0.028), and in patients on sodium-glucose cotransporter 2 inhibitors (SGLT2i) at baseline (62.9% vs 46.1%; p=0.017). After applying these variables in logistic regression, the resulting module has a very good prediction accuracy for HFimpEF (AUC=0.835). Previous CAD (OR=0.06; p=0.013), LGE with ischemic pattern (OR=0.196; p=0.035) and QRS duration (OR=0.98; p=0.026) were identified as independent predictors of the endpoint. Paradoxically, smoking was the strongest predictor of HFimpEF (OR=39.1; p<0.001).
Conclusion
This study highlights the role of previous CAD and LGE with ischemic pattern in predicting HFimpEF, regardless of HF etiology, as well as QRS duration. Unexpectedly, HFrEF therapy did not predict HFimpEF. The relationship between smoking and HFimpEF needs further investigation.