Post-ablation arrhythmogenic channels predict atrial fibrillation recurrence – iatrogenic substrate revisited

Fibrosis is a hallmark of atrial fibrillation (AF) arrhythmogenic substrate, and the extent of atrial fibrosis, as determined by late gadolinium enhancement (LGE)-MRI, predicts AF recurrence after ablation. However, ablation fundamentally alters the individual composition and distribution of fibrotic tissue – ideally, isolating arrhythmogenic foci and homogenizing pre-existing fibrosis. However, incomplete ablation lesions may create an arrhythmogenic substrate themselves. 

Against this background, we investigated post-ablation arrhythmogenic substrate using LGE-MRI. Of note, from ventricular arrhythmias (VA) we know that it is not the extent of scar tissue that determines arrhythmogenicity, but rather the scar-pervading channels of surviving tissue with residual conduction. LGE-MRI-based algorithms to automatically detect such scar channels are well-established in the context of VA; here we adapt and validate these tools for AF. 

 

150 patients with PVI-only AF ablation were included. All patients received a systematic 12-months follow-up and an LGE-MRI 3 months post-ablation. Gradient echo MR sequences were acquired in sinus rhythm and 3D-reconstruction of left atrium performed using a dedicated software (ADAS-3D). Scar channels were defined as corridors of borderzone tissue (residual conduction), protected by dense scar (non-conductive). Channels were automatically identified using a software algorithm that has been extensively validated for the ventricle.

To reproducibly distinguish healthy tissue, borderzone and dense scar, LGE was quantified based on the signal intensity ratios of each voxel relative to the blood pool. Signal intensity ratio thresholds defining dense scar and borderzone, were then empirically tested to enhance the predictive value of the identified channels. The algorithm was then applied to an independent validation cohort using the empirically determined thresholds.

 

150 patients with PVI-only AF ablation were analysed. 10 different combinations of signal intensity ratio thresholds defining borderzone tissue (1.1, 1.2) and dense scar (1.32, 1.45, 1.6, 1.7, 1.8) were empirically tested regarding the prediction of potentially arrhythmogenic channels in 50 patients (derivation cohort). For each of the threshold combinations, the number of predicted channels was linked to 12-months arrhythmia-free survival using logistic regression. The computed channels best predicted 12-months arrhythmia recurrence, when based on signal intensity ratios of 1.1-1.32 (borderzone) and >1.32 (dense scar).

 

This threshold combination was then validated in an independent cohort of 100 patients. Patients with scar channels had significantly lower arrhythmia-free survival than those without (55% vs. 88%, p=0.016). Logistic regression analyses found the number of detected scar channels to be predictive of 12-months arrythmia recurrence (OR 4.9; p=0.012) – independent of other factors like LA diameter, AF type or the extent of native atrial fibrosis (pre-ablation LGE-MRI). Of note, the number of gaps in the PV-encircling lesions was not predictive of AF recurrence.

 

LGE-MRI-detected post-ablation scar channels predict recurrent AF after PVI and may be a surrogate of iatrogenic substrate. In light of large-area ablations with PFA and other single-shot devices fueling a trend towards more and more extensive lesions, the concept of iatrogenic arrhythmogenic substrate may have to be reconsidered.