Diagnostic accuracy of cardiac magnetic resonance imaging in the detection and characterization of left atrial catheter ablation lesions: a multicenter experience.

Introduction: We tested the hypothesis that cardiovascular magnetic resonance (CMR) imaging can reliably distinguish the presence or absence of left atrial (LA) ablation lesions by blinded analysis of pre- and postablation imaging.

Methods: Consecutive patients at 2 centers undergoing pulmonary vein isolation (PVI) for paroxysmal atrial fibrillation by either wide area circumferential radiofrequency ablation (WACA) or ostial ablation with a cryoballoon underwent CMR late gadolinium enhancement (LGE) imaging pre- and 3 months postablation. Imaging was anonymized for blinded analysis of (1) LGE images, and (2) a 3D fusion image with LGE projected onto a segmented LA surface. Scans were categorized using both assessment techniques separately as pre- or postablation, and if postablation, whether lesions were in an ostial or WACA distribution.

Results: LGE imaging was performed in 50 patients (aged 60 ± 10 years, 68% male, 24 underwent WACA and 26 had cryoablation). Sensitivity and specificity for detection of ablation lesions was 60% and 96% on LGE imaging. Sensitivity was higher using 3D fusion imaging (88%; P = 0.003). The proportion in whom lesions were both detected and the distribution correctly assessed as WACA or ostial was higher with 3D fusion imaging compared to LGE imaging (54% vs 28%; P = 0.014). There was no difference in the detection of radiofrequency ablation lesions compared to cryoablation lesions (58% vs 62%; P = 1.000).

Conclusion: LGE imaging of atrial scar is not yet sufficiently accurate to reliably identify ablation lesions or to determine lesion distribution.