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

Area of Science:

• Cardiovascular Imaging and Radiology
• Biomedical Data Science
• Predictive Modeling in Medicine

Background:

• Accurate heart failure (HF) risk prediction is vital for timely intervention.
• Traditional HF risk models use clinical factors; cardiac MRI offers potential for enhanced prediction.
• The PREVENT equations are established tools for cardiovascular disease risk assessment.

Purpose of the Study:

• To assess if integrating multidimensional cardiac MRI parameters improves HF risk prediction accuracy.
• To evaluate the added value of cardiac MRI variables to the PREVENT equations.
• To explore sex-specific contributions of cardiac MRI parameters to HF risk prediction.

Main Methods:

• Secondary analysis of UK Biobank participants undergoing cardiac MRI.
• Development of sex-specific Fine-Gray competing risk models using PREVENT equations.
• Variable selection via LASSO regression and bootstrap resampling for 82 cardiac MRI parameters.
• Model performance evaluated using C-index and calibration charts on training and internal test sets.

Main Results:

• Sixteen cardiac MRI parameters identified by LASSO significantly improved HF risk prediction.
• C-index increased from 0.753 to 0.812 in the training set (P < .001) and 0.760 to 0.821 in the test set (P = .007).
• In men, left atrial functional parameters (e.g., LA ejection fraction) were key; in women, left ventricular ejection fraction and myocardial strain were most impactful.

Conclusions:

• Integration of cardiac MRI parameters into the PREVENT equations substantially enhances HF risk prediction.
• The study highlights significant sex-specific differences in cardiac MRI parameters influencing HF risk.
• Cardiac MRI provides valuable multidimensional data for refining cardiovascular risk stratification.