A total of 613 patients with persistent AF were included in our analyses based on the exclusion criteria (Fig. 1). Of these, 313 had been randomized to the PVI arm and 300 had been randomized to the MRI-guided ablation arm. The baseline characteristics of these patients including their medication use, comorbidities, and baseline imaging indices are shown in Tables 1 and 2. The mean age of the study cohort was 62.2 ± 9 years and females constituted 22.7% of the study population. A total of 60% of the study cohort had hypertension and 17.8% had CHF. Eighty-nine percent of the patients had a history of cardioversion and around 56% were taking antiarrhythmic medications.

Patient inclusion and exclusion from analysis. Legend: Among 843 patients randomized to pulmonary vein isolation or MRI-guided ablation, 613 patients met the final inclusion criteria

Atrial myopathy, atrial remodeling, and ablation-induced scar formation indices are shown in Table 2. At baseline, LAVI was 61.79 ± 18.58 mm3/m2. LAVI change from pre- to post-ablation in the whole cohort was 10.7 ± 12.8 mm3/m2 and was significantly higher in the MRI-guided ablation arm (9.38 vs. 12.06 mm3/m2, p < 0.001).

We also assessed baseline fibrosis, ablation-induced scar, and fibrosis coverage with ablation lesions. Baseline fibrosis was not significantly different in the treatment arms with an average baseline fibrosis of 18.43 ± 7.19%. Ablation-induced scar on post-ablation MRI constituted around 9.58 ± 5.05% of the left atrium. On average, 25% of fibrosis was covered in the MRI-guided arm compared to 19% in the PVI arm (p < 0.001).

Nearly half of the total cohort experienced the primary outcome of recurrence (n = 334, 54.5%). Consistent with the findings of the main trial, there was no significant difference in recurrence rates between patients who underwent conventional pulmonary vein isolation (PVI) ablation and those who underwent additional fibrosis-guided ablation (168, 46.3% vs. 134, 44.7%; p = 0.680). The average time to recurrence after the blanking period was 290 ± 191.94 days, with no significant difference observed between the two treatment arms (291.13 ± 189 days vs. 288.9 ± 194 days).

AFSS burden and total AFSS score were 18.94 ± 7.35 and 12.24 ± 8.17 at baseline, respectively (Table 3). All AFSS and SF-36 QoL items improved following CA compared to pre-ablation assessment (Tables 3 and 4) both in the PVI and fibrosis-guided ablation arms. As shown in Table 4, for all subscales of the AFSS and SF-36, there was no significant difference in improvement between both treatment arms. On univariable analysis, multiple associations between LGE-MRI indices and QoL outcomes were observed (Supplementary Table 1). A lower pre-ablation LAVI was associated with post-ablation improvements in AF frequency (β = 0.03, p < 0.001) and shortness of breath on activity (β = 0.009, p = 0.01).

Post-ablation improvement in LAVI was significantly associated with improvements in physical functioning (β =  − 0.1359, p = 0.035), vitality (β =  − 0.172, p = 0.035), general health (β =  − 0.114, p = 0.04), health change (β =  − 0.18, p = 0.034), AF burden (β =  − 0.056, p = 0.006), shortness of breath at rest (β =  − 0.009, p = 0.01), and dizziness (β =  − 0.008, p = 0.035) (Supplementary Table 2).

Post-ablation reduction in LAVI was significantly associated with total AFSS improvement (β = 0.085, p = 0.001) to a greater extent than baseline or post-ablation LAVI or substrate-related parameters such as baseline fibrosis or scar formation (Fig. 2).

Correlation between LAVI change and AFSS change. Legend: This scatter plot displays the relationship between the decrease in left atrial (LA) volume index post-ablation and the reduction in AFSS (Atrial Fibrillation Severity Scale) score. The results suggest an improvement in the severity of atrial fibrillation symptoms with favorable remodeling

In the multivariable linear regression model, various clinical confounders were included, such as patient demographics, comorbidities, treatment arm, and myopathy indices. The association between a reduction in LAVI and an improvement in AFSS remained significant, independent of potential confounding factors (Fig. 3). Additionally, baseline fibrosis was significantly associated with attenuated AFSS improvement in this model. Baseline AFSS score emerged as the strongest predictor of improvement, indicating that patients with more severe symptoms at baseline experienced the greatest benefit.

Multivariable linear regression model for AFSS change. Legend: This forest plot shows the results of the multivariable analysis for the outcome of AFSS score change. The multivariable analysis shows that baseline AFSS and LA volume index decrease are associated with improved AFSS score. Moreover, the association of LAVI decrease with AFSS improvement is still significant after controlling for the rest of the factors including the occurrence of atrial fibrillation recurrence post-ablation

Furthermore, atrial fibrillation recurrence was significantly associated with less symptom improvement but did not negate the significant effects of LAVI reduction and baseline fibrosis. LAVI reduction was also linked to improvements in specific metrics of the AFSS questionnaire, including global well-being (β =  − 0.0164, p = 0.018), AF frequency (β = 0.108, p < 0.001), and AF burden (β = 0.15, p < 0.001).

With respect to improvement in outcomes measured by the SF-36 questionnaire, LAVI decrease was associated with improvements in physical functioning (β = 0.27, p = 0.001), limitations due to physical health (β = 0.53, p < 0.001), vitality (β = 0.197, p = 0.01), and patient perspective on health change (β = 0.519, p < 0.001).

Baseline and residual fibrosis were associated with vitality (β =  − 0.34, p = 0.008 and β =  − 0.41, p = 0.006, respectively) and general health improvement on SF-36 (β =  − 0.26, p = 0.0034; β =  − 0.28, p = 0.04, respectively).To further understand our findings, we compared the AFSS change in patients who had an increase in LAVI (adverse remodeling) versus a reduction in LAVI (positive remodeling). Although both groups experienced improvements in AFSS score, those with positive remodeling experienced a greater improvement (4.95 vs. 7.02, p = 0.006, Fig. 4).

Change in AFSS based on increase or decrease in LAVI. Legend: This box plot compares the improvement in AFSS scores between two patient groups: those with an increased left atrial volume index (LAVI) and those with a decreased LAVI after ablation. More symptom severity improvement is seen in patients who had a decrease in LAVI (favorable remodeling) compared to patients with an increase in LAVI post-ablation