Baseline LA strain and AF recurrence
Analysis of the 17 included studies revealed that the MD between baseline LA strain in patients with post-ablation SR maintenance ((N= 1147, pooled mean (SD): 22.22% (10.64)) and those with AF recurrence ((N= 557, pooled mean (SD): 15.96% (9.77)) was statistically significant (5.43%, 95%CI: 4.03-6.84%). Heterogeneity analysis showed substantial heterogeneity (I2: 82.7%) (Figure 2).
We performed a sensitivity analysis, by including only the 12 studies that were classified as high-quality, but no significant change was noted compared to the overall results (Supplementary Figure 2). Also, when we excluded two studies[15.27] which visually introduced substantial heterogeneity, a more homogenous effect size estimation emerged (mean difference: 6, 95%CI: 5.12-6.89, I2:42.77%). Subgroup analysis was also performed by dividing studies according to the methodology used for strain measurement (apical 4 chambers view only versus biplane and the number of LA segments assessed), a significant subgroup interaction was noted (p=0.02) (Figure 3). The pooled effect sizes presented minimal quantitative changes compared to the initial pooled effect size estimation. Heterogeneity was almost eliminated in the smaller subgroups of studies, but remained substantial in the larger subgroup of studies that used the biplane method. Moreover, subgroup analysis was performed based on the type of the ablation procedure that was used – 7 studies using radiofrequency catheter ablation PVI alone versus 9 studies using radiofrequency catheter ablation PVI plus additional lesions- but no significant interaction was found (p=0.6). Subgroup analysis based on the software used (9 studies using EchoPac and 8 studies using other softwares) no significant interaction was noted (p=0.5). Finally, after removing one subgroup of patients who underwent second radiofrequency[13] catheter ablation procedure and one study with mixed population (first and second radiofrequency catheter ablation PVI)[14] heterogeneity remained at the same levels (I2: 84.5%) with no significant change in the effect size (MD: 5.36, 95%CI: 3.84-6.87).
Meta-regression analysis revealed no confounding effect of age (p=0.67), male gender (p=0.34), number of PAF patients included (p=0.96), number of hypertensive patients included (p=0.56) and baseline pre-ablation strain (p=0.66). However a statistically significant negative correlation was documented between the pooled effect size and the mean baseline maximum LAVi (regression coefficient= -0.184, p=0.004, one study where LAVi was measured using 3D echocardiography was excluded[27]) (Figure 4).