Background: Identifying significant coronary artery disease (CAD) in patients with reduced left ventricular ejection fraction (rLVEF) is essential for guiding therapeutic decisions, including medical management, device implantation and potential revascularisation. Prior studies suggested that rest cardiac MRI (CMR) with late gadolinium enhancement (LGE) could reliably detect significant CAD. We aimed to evaluate the diagnostic accuracy of rest LGE-CMR for predicting significant CAD in rLVEF patients.
Methods: In this prospective, multicentre cohort study across 10 centres, adults with new-onset rLVEF≤45% without obvious cause were included. All patients underwent rest CMR and coronary angiography. Independent, blinded committees reviewed images. Significant CAD was defined as ≥70% stenosis in major coronary arteries. Ischaemic scars were identified on CMR as subendocardial LGE. The primary outcome was the sensitivity of CMR in detecting significant CAD.
Results: Among 380 patients (median age 63 years, 68% male), significant CAD was present in 49 (13%). CMR identified ischaemic scars in 106 (28%). The sensitivity of CMR for detecting significant CAD was 57% (95% CI: 43% to 71%), specificity 76% (95% CI: 72% to 81%), positive predictive value 26% (95% CI: 18% to 35%) and negative predictive value 92% (95% CI: 89% to 95%). A CMR-first strategy would have missed 43% of significant CAD cases, many requiring revascularisation (86% of missed cases).
Conclusions: In this large, prospective multicentre study with independent image review, rest LGE-CMR demonstrated limited sensitivity for detecting significant CAD in patients with rLVEF. Relying solely on CMR could lead to missed diagnoses and undertreatment. CMR should be integrated with other diagnostic tools to optimise care in this population.
Trial registration number: NCT03231189.
Keywords: Coronary artery disease; Health Care Economics and Organizations; Heart Failure, Systolic; Magnetic Resonance Imaging.
© Author(s) (or their employer(s)) 2025. Re-use permitted under CC BY. Published by BMJ Group.