Patients. There was a total of 1916 patients who underwent first catheter or surgical ablation between January 2008 and December 2013. There were only 411 patients that met the study criteria after review. Of these, 132 were treated by surgical ablation and TV repair, and 114 of whom had no history of left-sided valve surgery were included as the surgical cohort. Of the remaining 279 patients who were treated with ‘2C3L’ catheter ablation were selected as the catheter cohort. Baseline characteristics were shown in Table 1.
Procedural outcomes. All patients successfully underwent the procedure, and there was no in-hospital mortality. In both cohorts, 100% PVs that are targeted for ablation demonstrated entrance and exit block at the conclusion of the procedure. Intra-procedural AF termination was more (Table 2). In catheter cohort, the mean procedural time, fluoroscopy time and radiofrequency time were 215 ± 38 min, 48 ± 11 min and 116 ± 24 min, respectively. In surgical cohort, the mean duration of cardiopulmonary bypass and aortic cross-clamping was 97 ± 24 min and 46 ± 20 min, respectively.
Outcomes during follow-up. All included patients have completed a 5-year clinical follow-up. After a mean follow-up period of 77 months, there were 22 and 12 deaths in the catheter and surgical cohort (Table 3). During the follow-up, only 1 patient required pacemaker implantation for sinus node dysfunction in surgical cohort. On Kaplan-Meier analysis, the estimated actuarial 5-year survival rates were 96.8% (CI: 92.95-97.78) and 92.0% (CI: 85.26-95.78) in the catheter and surgical cohort, respectively (Figure 2-A). In contrast to the surgical cohort, the catheter cohort had a higher risk of stroke ( SHR: 0.3790; CI: 0.180-0.797; P = 0.011, Figure 2-C), but had a similar risk of recurrent AF ( SHR: 0.8458; CI: 0.637-1.124; P = 0.248, Figure 2-B) and moderate-severe TR ( SHR: 1.2950; CI: 0.924-1.814; P = 0.133, Figure 2-D) after considering death as a competing risk.
Heart rhythm and echocardiography during follow-up. All patients in both cohorts had their complete Holter, and echocardiogram follow-up results. According to the rhythm status at follow-up, study population were divided into the patients with sinus rhythm (SR) and with recurrent AF in each cohort.
In catheter cohort, at baseline, both groups had similar percentages of patients with moderate-severe TR. At follow-up, 13% of patients with SR had trace or no TR, compared with 0% in the recurrent group, and 69% with SR had mild TR compared with 32% in the recurrent group. Only 18% of patients with SR still had moderate-severe TR at follow-up, compared with 68% in the recurrent group (p <0.0001 for entire trend) (Figure 3-A). Moreover, SR group had a significant decrease in tethering height, tethering area, TR effective regurgitant orifice area (EROA), TR vena contracta (VC), TR grade and tricuspid annular dimension (TAD), with RV sphericity index (RVSI) and RV fractional area change (RVFAC) significantly increased at follow-up. And the recurrent group had a significant increase in right atrial (RA) area, TR EROA, TR grade and TAD, with RVSI and RVFAC significantly decreased at follow-up (Supplementary Table 1).
In surgical cohort, at baseline, both groups had similar percentages of patients with moderate-severe TR. At follow-up, all patients with recurrent AF had moderate-severe TR at follow-up when compared with 33% in the SR group (p<0.0001 for entire trend) (Figure 3-B). Moreover, SR group had a significant decrease in tethering height, tethering area, TR EROA, TR VC, TR grade and TAD, with RVSI and RVFAC significantly increased at follow-up. And the recurrent group had a significant increase in RA area and TAD, with RVSI and RVFAC significantly decreased at follow-up. (Supplementary Table 2)
Predictor for the recurrence of either AF or moderate-severe TR
Multivariate Cox hazard regression analysis revealed that RVSI and tethering height as independent risk factors for the recurrence of either AF or moderate-severe TR in both cohorts (Table 4 and 5). Next, the optimal cut-off value of the RVSI and tethering height in predicting the recurrence was identified using ROC curve analysis (Table 6). Interestingly, the tethering height showed a significantly better predictive performance than RVSI in the recurrence of either AF or moderate-severe TR in both cohorts (Figure 4).
According to the merits of tethering height mentioned above, we made a 3-D plot graph to demonstrate the relation between basal tethering height, change in TR severity at follow-up and SR restored duration after the procedure in total cohort (Figure 5). It showed the higher the basal tethering height was, the more changes in TR severity at follow-up were, and the less SR restored duration after the procedure was.
Propensity-score match analysis between two cohorts
A propensity-score matching was performed to adjust for differences in measured baseline characteristics between 2 cohorts (Table 1, Supplementary Figure 1). Among the 111 PSM pairs, the catheter group showed significantly higher rates of stroke (SHR = 0.286; 95% CI, 0.106–0.768; P=0.013), and AF recurrence (SHR = 0.669; 95% CI, 0.486-0.976; P=0.036) after considering death as a competing risk (Figure 6). We categorized matched patients into subgroups by tethering height < 6mm and ³ 6mm to determine whether a graded relation exists with tethering height and our further customizing and selecting the best treatment options for the patients with AF-caused moderate-severe TR. On Kaplan–Meier analysis, there were no difference in survival from mortality in patients with tethering height < 6mm and ³ 6mm (Figure 7-A, 8-A). On competing risk analysis, the cumulative incidence of stroke, recurrent AF and moderate-severe TR showed no difference between two groups (Figure 7 B-D). However, the cumulative incidence of stroke (SHR = 0.199; 95% CI, 0.047–0.849; P=0.029), recurrent AF (SHR = 0.589; 95% CI, 0.357–09738; P=0.039) and moderate-severe TR (SHR = 0.593; 95% CI, 0.359–0.981; P=0.042) were higher for patients with tethering height ³ 6mm in catheter group (Figure 8 B-D). Examples of two patients in each cohort with different TR severity at follow-up are shown in Supplementary Figure 2 and 3.