The anthropometric, environmental and genetic determinants of right ventricular structure and function

Abstract

BACKGROUND Measures of right ventricular (RV) structure and function have significant prognostic value. The right ventricle is currently assessed by global measures, or point surrogates, which are insensitive to regional and directional changes. We aim to create a high-resolution three-dimensional RV model to improve understanding of its structural and functional determinants. These may be particularly of interest in pulmonary hypertension (PH), a condition in which RV function and outcome are strongly linked. PURPOSE To investigate the feasibility and additional benefit of applying three-dimensional phenotyping and contemporary statistical and genetic approaches to large patient populations. METHODS Healthy subjects and incident PH patients were prospectively recruited. Using a semi-automated atlas-based segmentation algorithm, 3D models characterising RV wall position and displacement were developed, validated and compared with anthropometric, physiological and genetic influences. Statistical techniques were adapted from other high-dimensional approaches to deal with the problems of multiple testing, contiguity, sparsity and computational burden. RESULTS 1527 healthy subjects successfully completed high-resolution 3D CMR and automated segmentation. Of these, 927 subjects underwent next-generation sequencing of the sarcomeric gene titin and 947 subjects completed genotyping of common variants for genome-wide association study. 405 incident PH patients were recruited, of whom 256 completed phenotyping. 3D modelling demonstrated significant reductions in sample size compared to two-dimensional approaches. 3D analysis demonstrated that RV basal-freewall function reflects global functional changes most accurately and that a similar region in PH patients provides stronger survival prediction than all anthropometric, haemodynamic and functional markers. Vascular stiffness, titin truncating variants and common variants may also contribute to changes in RV structure and function. CONCLUSIONS High-resolution phenotyping coupled with computational analysis methods can improve insights into the determinants of RV structure and function in both healthy subjects and PH patients. Large, population-based approaches offer physiological insights relevant to clinical care in selected patient groups.