Background
Late gadolinium enhancement MRI (LGE-MRI) is increasingly used to detect native as well as ablation-induced atrial fibrosis in the context of atrial fibrillation (AF). However, cardiac fibrotic tissue is subject to sustained remodeling, and neither the development of atrial fibrosis over time nor the capability of LGE-MRI to detect fibrosis at different stages of scar formation have been defined. We sought to define the long-term development of ablation-induced atrial fibrosis and to validate the LGE-MRI for the assessment of ablation lesions.

Methods
Consecutive patients with first-time AF ablation and serial post-ablation LGE-MRI at least 12 months apart were included if they had been scheduled for a repeat ablation procedure. LGE-MRI data were postprocessed for quantification of fibrotic tissue using the ADAS 3D software. High-density electroanatomical mapping (EAM) served as a reference.

Results
In the 22 patients analysed, LGE-MRI-detected ablation-induced fibrosis decreased from 91.7 ± 7.0% of the anticipated ablation line at a median of 3 months post ablation to 62.8 ± 25.0% at a median of 28 months (p<0.0001) (see figure for representative examples and individual development of fibrosis over time). This was also reflected by an increase in the median number of LGE-MRI-predicted gaps in the ablation lesion from 4 (3 months) to 10 (28 months). However, EAM subsequent to the second post-ablation LGE-MRI confirmed only 5% of the new gaps detected by the 28-months LGE-MRI, but agreed with the 3-months LGE-MRI in 95% of these cases. Overall agreement of LGE-MRI with EAM regarding the prediction of ablation-induced fibrosis and functional gaps was good after three months (κ 0.74; p<0.0001, negative predictive value 98%, positive predictive value 76%), but weak after 28-months (κ 0.29; p<0.0001, negative predictive value 96%, positive predictive value 31%).

Conclusions

Our results indicate that while ablation-induced atrial fibrosis appears to remain constant over time, LGE-MRI loses its capability to detect it. Thus, while agreement of LGE-MRI with EAM appears to be very good at 3-months post-ablation with a very high predictive value regarding functional gaps, the predictive value of LGE-MRI was rather weak after a median of 28 months following ablation.