Late gadolinium enhancement cardiac magnetic resonance (LGE-CMR) permits characterizing ischaemic scars, detecting heterogeneous tissue channels (HTCs) which constitute the arrhythmogenic substrate (AS). Late gadolinium enhancement cardiac magnetic resonance also improves the arrhythmia-free survival when used to guide ventricular tachycardia (VT) substrate ablation. However, its availability may be limited. We sought to evaluate the performance of multidetector cardiac computed tomography (MDCT) imaging in identifying HTCs detected by LGE-CMR in ischaemic patients undergoing VT substrate ablation.
Thirty ischaemic patients undergoing both LGE-CMR and MDCT before VT substrate ablation were included. Using a dedicated post-processing software, two blinded operators, assigned either to LGE-CMR or MDCT analysis, characterized the presence of CMR and computed tomography (CT) channels, respectively. Cardiac magnetic resonance channels were classified as endocardial (layers < 50%), epicardial (layers ≥ 50%), or transmural. Cardiac magnetic resonance- vs. CT-channel concordance was considered when showing the same orientation and American Heart Association (AHA) segment. Mean age was 69 ± 10 years; 90% were male. Mean left ventricular ejection fraction was 35 ± 10%. All patients had CMR channels (n = 76), whereas only 26/30 (86.7%) had CT channels (n = 91). Global sensitivity (Se) and positive predictive values for detecting CMR channels were 61.8% and 51.6%, respectively. MDCT performance improved in patients with epicardial CMR channels (Se 80.5%) and transmural scars (Se 72.2%). In 4/11 (36%) patients with subendocardial myocardial infarction (MI), MDCT was unable to identify the AS.
Compared to LGE-CMR, myocardial wall thickness assessment using MDCT fails to detect the presence of AS in 36% of patients with subendocardial MI, showing modest sensitivity identifying HTCs but a better performance in patients with transmural scars.

Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2020. For permissions, please email: journals.permissions@oup.com.

Author