Myocardial contraction fraction derived from cardiovascular magnetic resonance cine images-reference values and performance in patients with heart failure and left ventricular hypertrophy.

Aims: Left ventricular hypertrophy (LVH) has strong prognostic implications and is associated with heart failure. Recently, myocardial contraction fraction (MCF) was identified as a useful marker for specifically identifying cardiac amyloidosis (CA). The purpose of this study was to evaluate the diagnostic accuracy of MCF for the discrimination of different forms of LVH.
Methods and Results: We analysed cardiovascular magnetic resonance (CMR) scans of patients with CA (n = 132), hypertrophic cardiomyopathy (HCM, n = 60), hypertensive heart disease (HHD, n = 38) and in 100 age- and gender-matched healthy controls. MCF was calculated by dividing left ventricular (LV) stroke volume by LV myocardial volume. The diagnostic accuracy of MCF was compared to that of LV ejection fraction (EF) and the mass index (MI). Compared with controls (136.3 ± 24.4%, P < 0.05), mean values for MCF were significantly reduced in LVH (HHD:92.6 ± 20%, HCM:80 ± 20.3%, transthyretin CA:74.9 ± 32.2% and light-chain (AL) CA:50.5 ± 21.4%). MCF performed better than LVEF (AUC = 0.96 vs. AUC = 0.6, P < 0.001) and was comparable to LVMI (AUC = 0.95, P = 0.4) in discriminating LVH from controls. There was a significant yet weak correlation between MCF and LVEF (r = 0.43, P < 0.0001). MCF outperformed LVEF and LVMI in discriminating between different etiologies of LVH and between AL and other forms of LVH (AUC = 0.84, P < 0.0001). Moreover, cut-off values for MCF <50% and LVEF <60% allowed to identify patients with high probability for CA.
Conclusion: In patients with heart failure MCF discriminates CA from other forms of LVH. As it can easily be derived from standard, non-contrast cine images, it may be a very useful marker in the diagnostic workup of patients with LVH.
(Published on behalf of the European Society of Cardiology. All rights reserved. © The Author 2017. For permissions, please email: journals.permissions@oup.com.)