Summary: | Background Cardiac structure and function result from complex interactions between genetic and environmental factors. Population-based studies have relied on 2-dimensional cardiovascular magnetic resonance as the gold-standard for phenotyping. However, this technique provides limited global metrics and is insensitive to regional or asymmetric changes in left ventricular (LV) morphology. High-resolution 3-dimensional cardiac magnetic resonance (3D-CMR) with computational quantitative phenotyping, might improve on traditional CMR by enabling the creation of detailed 3D statistical models of the variation in cardiac phenotypes for use in studies of genetic and/or environmental effects on cardiac form or function. Purpose To determine whether 3D-CMR is applicable at scale, and provides methodological and statistical advantages over conventional imaging for large-scale population studies and to apply 3D-CMR to anthropometric and genetic studies of the heart. Methods 1530 volunteers (54.8% females, 74.7% Caucasian, mean age 41.3±13.0 years) without self-reported cardiovascular disease were recruited prospectively to the Digital Heart Project. Using a cardiac atlas-based software, these images were computationally processed and quantitatively analysed. Parameters such as myocardial shape, curvature, wall thickness, relative wall thickness, end-systolic wall stress, fractional wall thickening and ventricular volumes were extracted at over 46,000 points in the model. The relationships between these parameters and systolic blood pressure (SBP), fat mass, lean mass and genetic variationswere analysed using 3D regression models adjusted for body surface area, gender, race, age and multiple testing. Targeted resequencing of titin (TTN), the largest human gene and the commonest genetic cause of dilated cardiomyopathy, was performed in 928 subjects while common variants (~700.000) were genotyped in 1346 subjects. Results Automatically segmented 3D images were more accurate than 2D images at defining cardiac surfaces, resulting in fewer subjects being required to detect a statistically significant 1 mm difference in wall thickness. 3D-CMR enabled the detection of a strong and distinct regionality of the effects of SBP, body composition and genetic variation on the heart. It shows that the precursors of the hypertensive heart phenotype can be traced to healthy normotensives and that different ratios of body composition are associated with particular gender-specific patterns of cardiac remodelling. In 17 asymptomatic subjects with genetic variations associated with dilated cardiomyopathy, early stages of ventricular impairment and wall thinning were identified, which were not apparent by 2D imaging. Conclusions 3D-CMR combined with computational modelling provides high-resolution insight into the earliest stages of heart disease. These methods show promise for population-based studies of the anthropometric, environmental and genetic determinants of LV form and function in health and disease.
|