Predictive value of late gadolinium enhancement MRI in patients with persistent atrial fibrillation. Multicentric validation of a standardised method

• Published in: European heart journal Vol. 43; no. Supplement_2
• Main Authors: Althoff, T F, Eichenlaub, M, Padilla, D, Brugada, J, Arentz, T, Mont, L, Jadidi, A
• Format: Journal Article
• Language: English
• Published: 03-10-2022
• ISSN: 0195-668X; 1522-9645
• DOI: 10.1093/eurheartj/ehac544.424

Abstract
Background
With recurrence rates after PVI in persistent atrial fibrillation (AF) as high as 50%, predictive tools to improve patient selection are clearly needed. Compared to paroxysmal AF, persistent AF is more substrate-dependent. While endocardial low voltage substrate detected by invasive mapping has been shown to predict AF recurrence, to date the only non-invasive method to assess arrhythmogenic substrate is late gadolinium enhancement (LGE)-MRI. In fact, patient selection for PVI based on left atrial (LA) LGE extent according to the UTAH stages has been proposed, However, this approach has not been widely established, in part owed to the lack of standardisation and thus limited reproducibility of the MRI postprocessing method. Moreover, in the recent DECAAF-2 multicenter trial, the UTAH stages failed to predict outcome in patients with persistent AF.
Purpose
We have recently established a standardised MRI method that aims at reproducible quantification of LGE. Here, the ability of this method to detect arrhythmogenic substrate and to predict outcome after PVI in patients with persistent AF was evaluated by two centers independently.
Methods
This dual center study consisted of a prospective derivation cohort at centre1 and a validation cohort at centre 2. All patients received an LGE-MRI prior to ablation (PVI only) and were followed systematically with holter ECG at 3, 6 and 12 months. Gradient echo MR sequences were acquired in sinus rhythm and LA 3D-reconstruction performed using ADAS-3D software.
LGE was quantified based on the signal intensity ratio of each voxel relative to the blood pool, applying a uniform threshold of >1.2 to define LGE indicative of fibrotic tissue. Invasive mapping served as a reference defining arrhythmogenic substrate (>2 cm2 LA area with bipolar voltage <0.5 mV).
Results
A ROC analysis of 37 consecutive patients with persistent AF in the prospective derivation cohort yielded a cutoff-value of 12% LGE to predict relevant low-voltage substrate with the highest accuracy. This cutoff-value was also predictive of AF recurrence after PVI (Fig. 1). When applied to the external validation cohort of 182 consecutive patients, the cutoff-value of 12% was also predictive of AF recurrence after PVI – both in the full cohort (odds ratio 3.3, p=0.003) and in the subgroup of 65 patients with persistent AF (odds ratio 3.7, p=0.038). Of note, among patients with persistent AF, recurrence rate was 54% in those with >12% LGE but only 24% in those with <12% (Fig. 2).
Conclusion
This dual center study established and validated the predicitive value of LGE as determined by a standardised MRI method in patients with persistent AF that can be reproducibly applied across different centers. A cutoff-value of 12% LGE was able to discriminate between responders and non-responders to PVI and may thus guide selection of suitable candidates that are likely to benefit from ablation.
Funding Acknowledgement
Type of funding sources: None.