Recurrence of atrial fibrillation correlates with the extent of post-procedural late gadolinium enhancement: a pilot study

Abstract

Objectives: We sought to evaluate radiofrequency (RF) ablation lesions in atrial fibrillation (AF) patients using cardiac magnetic resonance (CMR), and to correlate the ablation patterns with treatment success.

Background: RF ablation procedures for treatment of AF result in localized scar that is detected by late gadolinium enhancement (LGE) CMR. We hypothesized that the extent of scar in the left atrium and pulmonary veins (PV) would correlate with moderate-term procedural success.

Methods: Thirty-five patients with AF, undergoing their first RF ablation procedure, were studied. The RF ablation procedure was performed to achieve bidirectional conduction block around each PV ostium. AF recurrence was documented using a 7-day event monitor at multiple intervals during the first year. High spatial resolution 3-dimensional LGE CMR was performed 46 +/- 28 days after RF ablation. The extent of scarring around the ostia of each PV was quantitatively (volume of scar) and qualitatively (1: minimal, 3: extensive and circumferential) assessed.

Results: Thirteen (37%) patients had recurrent AF during the 6.7 +/- 3.6-month observation period. Paroxysmal AF was a strong predictor of nonrecurrent AF (15% with recurrence vs. 68% without, p = 0.002). Qualitatively, patients without recurrence had more completely circumferentially scarred veins (55% vs. 35% of veins, p = NS). Patients without recurrence more frequently had scar in the inferior portion of the right inferior pulmonary vein (RIPV) (82% vs. 31%, p = 0.025, Bonferroni corrected). The volume of scar in the RIPV was quantitatively greater in patients without AF recurrence (p < or = 0.05) and was a univariate predictor of recurrence using Cox regression (p = 0.049, Bonferroni corrected).

Conclusions: Among patients undergoing PV isolation, AF recurrence during the first year is associated with a lesser degree of PV and left atrial scarring on 3-dimensional LGE CMR. This finding was significant for RIPV scar and may have implications for the procedural technique used in PV isolation.

Figure 1. Methodology for measuring scar volumes

(A) original LGEimages of scar, with hyperenhanced scar. (B) Scar is highlighted (in red) using signal thresholding, separating scar from normal tissues. (C) ROIs are used to identify scar on each slice as belonging to the LIPV, RIPV, LSPV, RSPV, and posterior wall. (D) Each slice is analyzed using thresholding ands ROIs to identify scarred pixels in each region. (E) Total scar is summed over all slices.

Figure 2. 3D LGE images in three patients, with a volume-rendered view

Three patients, two without recurrence of AF (A,B) and one with recurrence of AF (C). Top panels show an LGE slice from the 3D volume. Bottom panels show planar reformats across PV ostia. The orientation is shown in bottom panel B. The reformats provide examples of qualitative gradings for the RIPV of (Fig. 2A) 3 – complete; (Fig. 2B) 2–partial; and (Fig. 2C) 1 –no scar. D) A 3D volume-rendering of the scar for Patient B. A mitral isthmus line is observed (arrow). Movie Legend (associated with Figure 2D): 3D visualization showing RF ablation scar fused with the MR angiogram, in another patient, providing evaluation of the scar in the context of the pulmonary vein anatomy.

Figure 3. Kaplan-Meier AF free survival curve stratified by RIPV scar

Kaplan-Meier survival curve showing AF free survival time for patients using median scar volume cutoffs (1.98 mls), for patients with greater and lesser RIPV scar volume measurement. More extensive scarring was associated with longer AF-free survival (p = 0.029).