Atrial Fibrosis Imaging: Correlation of Late Gadolinium Enhancement Magnetic Resonance Imaging With High-Definition Voltage Electroanatomical Mapping Across Different Thresholds.

INTRODUCTION: Atrial fibrosis is an adverse factor for arrhythmia recurrences and thromboembolic events in atrial fibrillation (AF). Cardiac magnetic resonance imaging (CMRi) has emerged as a noninvasive tool for its assessment, though evidence remains conflicting.

METHODS AND RESULTS: Sixty ablation-naïve patients with AF scheduled for first-time catheter ablation were enrolled (78% men, 67% paroxysmal AF, 73% heart failure reduced ejection fraction [HFrEF]). Prior to the procedure, CMRi was performed in sinus rhythm (SR) on a 1.5T system using a high-resolution (1.3 mm isotropic) free-breathing 3D LGE sequence with echo-navigator end-expiratory gating and whole-heart coverage. Left atrium (LA) fibrosis was quantified using the commercially available ADAS 3D software, with fibrosis defined at four image intensity ratios (IIR) thresholds (0.97, 1.2, 1.32, and 1.61 times mean blood-pool signal intensity). High-density electroanatomical mapping (EAM) was performed in SR, and low-voltage substrates (LVS) were quantified at three bipolar voltage thresholds (0.5, 0.25, and 0.1 mV). LVS-EAM were identified in 97%, 88%, and 45% of patients at 0.5, 0.25, and 0.1 mV, respectively. LGE-CMRi was detected in 100%, 97%, 63%, and 17% of patients at IIR thresholds of 0.97, 1.2, 1.32, and 1.61, respectively. Quantitative correlation between CMR- and EAM-derived fibrosis was negligible to weak across all threshold combinations. When fibrosis was analyzed dichotomously, agreement was highest between EAM 0.5 mV and CMR IIR 0.97 (0.97, 95% CI [0.92, 1.00]; p < 0.001) and IIR 1.2 (1, 95% CI [1.00, 1.00]; p < 0.001), and between EAM 0.25 mV and CMR IIR 0.97 (0.87 95% CI [0.76, 0.97]; p < 0.001) and IIR 1.2 (0.9, 95% CI [0.81, 0.99]; p < 0.001).

CONCLUSIONS: In an ablation-naïve AF population with a predominance of HFrEF patients, CMR- and EAM-derived measures of atrial fibrosis showed poor quantitative correlation across multiple threshold combinations. Although agreement improved when fibrosis was analyzed dichotomously (presence vs. absence of fibrosis), the highest agreement was observed at the most sensitive threshold combinations. The findings of our study highlight the need for standardized, validated, and reproducible CMR protocols for native atrial fibrosis assessment in order to provide a noninvasive alternative to EAM.