Extending the PREVENT Equations with Cardiac MRI: Prediction of 10-year Heart Failure Risk

Background Accurate prediction of heart failure (HF) risk is crucial for early intervention and prevention. Traditional models rely on clinical risk factors, whereas cardiac MRI parameters may offer additional predictive value. Purpose To evaluate whether integrating multidimensional cardiac MRI parameters with the Predicting Risk of Cardiovascular Disease EVENTs (PREVENT) equations improves HF risk prediction. Materials and Methods This secondary analysis of a prospective study included participants from the UK Biobank who underwent cardiac MRI. Participants were randomly divided into 70% for a training set and 30% for an internal test set. Eighty-two representative cardiac MRI variables were log-transformed and standardized. Variable selection was performed using least absolute shrinkage and selection operator (LASSO) regression combined with bootstrap resampling. Sex-specific Fine-Gray competing risk models were constructed based on the PREVENT equations. Model performance was evaluated using the C-index and calibration charts. Results This study included 39 069 participants (mean age, 55 years ± 7.5 [SD]; 20 520 women). Sixteen cardiac MRI parameters identified by LASSO enhanced HF risk prediction. In the training set (n = 27 531), the C-index increased from 0.753 (95% CI: 0.728, 0.777) to 0.812 (95% CI: 0.787, 0.837) (ΔC-index = 0.059; P < .001); in the internal test set (n = 11 538), it increased from 0.760 (95% CI: 0.719, 0.801) to 0.821 (95% CI: 0.779, 0.862) (ΔC-index = 0.061; P = .007). In men, left atrial (LA) functional parameters contributed most to the improvement in C-index, including LA ejection fraction (ΔC-index = 0.022), whereas in women, the gain was primarily driven by left ventricular (LV) ejection fraction (ΔC-index = 0.011) and myocardial strain (LV circumferential strain: ΔC-index = 0.008; LV radial strain: ΔC-index = 0.006). Conclusion Integrating cardiac MRI parameters into the PREVENT equations improved HF risk prediction and highlighted sex-specific pathophysiologic differences. © RSNA, 2026 Supplemental material is available for this article. See also the editorial by Hanneman and Moayedi in this issue.