Results
Study population
As in Table 1 , a total of 139 patients with drug-refractory PVC referred to our institute for ablation were analyzed (40 in group 1 and 99 in group 2). The characteristics of the PVCs during the 24-hour Holter monitoring were documented, with the mean PVC burden over 24 hour was 20.6 ± 11.6%, and 18.0% of patients with sustained VT and 30.2% of those with non-sustained VT. The mean LVEF was 52.4 ± 11.2%. LV dysfunction (LVEF<50%), defined by echocardiography before the ablation, was identified in 40 patients (28.8%). The origin of the PVC was confirmed to be in the LV summit, with the earliest site in the epicardium and/or GCV/AIV.
Baseline patient and PVC characteristics
The comparison of the clinical characteristics of patients with (group 1) and without LV dysfunction (group 2) is summarized in Table 2 . The mean LVEF in group 1 was 37.5 ± 9.3% and 58.4 ± 4.1% in group 2. Compared to patients in group 2, there were more male (80.0% vs. 58.6%; P = 0.019), documented sustained VT (30.0% vs.13.1%;P = 0.027), RBBB pattern of the VA (53.4% vs. 33.5%; P = 0.006), longer CI (510.5 ± 66.4 ms vs.483.7 ± 58.8 ms; P = 0.021), wider QRS duration (163.7 ± 28.0 ms vs. 147.2 ± 20.4 ms;P = 0.001), and larger AEAD (12.0 ± 9.1 vs. 16.3 ± 8.2 ms; P= 0.012). There was no significant difference in underlying disease, clinical presentation, baseline medical therapy, PVC burden, or the successful ablation site.
Characteristics associated with the development of LV dysfunction
In univariate analysis, as in Table 3 , sex, presence of sustained VT, PVC morphology with RBBB pattern, CI, AEAD, and PVC QRS duration were associated with an increased risk of LV dysfunction. In multivariate analysis, only PVC QRS duration and AEAD were independently associated with the development of LV dysfunction.
Ablation outcome and changes in the LV function before and after the ablation
The overall acute success rate of the index ablation was 92.8%. The successful ablation site was most in the ASV (35.3%), followed by GCV/AIV (30.9%), subvalvular (20.9%), and epicardium (12.9%).
The changes in the PVC burden and the corresponding LVEF before and after ablation of the LV summit PVCs are shown in Figure 2 . In patients with and without LV dysfunction, the PVC burden after the ablation was significantly decreased than that before the ablation. In group 2 patients, the LVEF before (58.4 ± 5.0%) and after (58.4 ± 5.0%) the ablation was similar (P = 0.865). In contrast, the LVEF significantly improved after ablation in the group 1 patients (post-LVEF vs. before-LVEF: 48.5 ± 10.2% vs. 37.5 ± 9.3%; P< 0.001).
During a mean follow-up period of 27.6 ± 18.4 months, 26 (18.7%) patients had recurrences with a mean duration of 9.1 ± 5.8 months after the index procedure. Of these 26 patients with recurrence, 22 patients (84.6%) had the same PVC morphology as the PVC morphology during the index procedures, and the other four patients (15.4%) had recurrences with different PVC morphology. Compared with group 2, there was a higher recurrence rate in patients in group 1 (group 1 vs. group 2: 32.5% vs. 13.1%; P = 0.015). Although there was PVC recurrence during follow-up, the PVC burden decreased significantly (4.8 ± 7.4%) compared to the PVC burden before the index procedure (17.6 ± 8.8%; P< 0.001), and the LVEF also improved significantly (post-LVEF vs. before-LVEF: 53.8 ± 11.8% vs. 48.2 ± 14.6%; P = 0.037). For group 1 patients, despite PVC recurrences, the PVC burden significantly decreased (2.2 ± 3.7% vs.14.5 ± 7.9%; P < 0.001) and there were significant improvements for LVEF after the index procedures (post-LVEF vs. before-LVEF:47.5 ± 13.0% vs.35.9 ± 8.7%; P = 0.015).
For 26 patients with PVC recurrences, 11 patients (42.3%) received repeat procedures (9 patients with the same PVC morphology and 2 patients with different morphology as the PVC morphology during the index procedures). Of these 11 patients, 2 patients were in group 1 and the other 9 patients were in group 2. For these 2 patients in group 1, both had LV function recovery after the index procedure (one patient with LVEF improved from 27% to 42%, and the other LVEF improved from 25% to 43%), and the LV function were similar after the repeat procedures (LVEF was 45% in both patients after the repeat procedures).
Determinants associated without LV function recovery after ablation
Of 40 patients with LV dysfunction (group 1), 20 patients (50.0%) did not recover LV function after ablation. As in Supplemental Table I , of these 20 patients without recovery, there were poorer LVEF at baseline (34.3 ± 7.3% vs.40.7 ± 10.1%; P = 0.027), longer PVC QRS duration (178.3 ± 23.1 ms vs.149.0 ± 24.9 ms; P < 0.001), and with more PVC recurrences after ablation than patients with LV function recovery (50.0% vs.15.0%; P = 0.041).
After multivariate analysis (Supplement Table II) , QRS duration of the PVC and baseline LVEF before the index procedure was independently predictive for irreclaimable LV function after ablation.