Baseline clinical characteristics
Out of 387 consecutive patients with NVAF who underwent LAAC, six patients were switched to the Amplatzer Cardiac Plug device (St. Jude Medical, Golden Valley, MN) because of unsuitable LAA anatomy for WATCHMANTM (WM) device. Four cases of implantation with WM device were halted due to 3 cases of significant pericardial effusion / cardiac tamponade and 1 case of repeated device-associated thrombus. 377 (97.4%) cases were successfully implanted with WM occluders, including 137 cases in the bleeding group and 240 cases in the non-bleeding group. In the bleeding group, prior major bleedings were intracranial in 20 patients (14.6%), gastrointestinal in 86 patients (62.8%), and other major bleeding in 31 patients (22.6%). No significant difference in procedural success rate of WATCHMANTM occluders was found between the two groups (97.6% vs 96.8%, P=0.156).
Table 1 shows the baseline demographic and clinical characteristics of the study population. The bleeding group more often had individuals ≥ 75 years old and had significantly higher mean CHA2DS2-VASc and HAS-BLED scores, but was less likely to take oral warfarin compared to the non-bleeding group. The remaining clinical variables were comparable between the two groups (Table 1).
The total adverse events over the peri-procedure period (i.e. within 7 days of the index procedure) occurred in 11 cases (2.9%). Stroke, major bleeding, pericardial effusion/ cardiac tamponade, and severe vascular complication occurred in 0.3%, 0.5%, 0.8%, and 1.3% of subjects respectively, with no device-related death. All 2 cases with major bleeding were attributed to gastrointestinal bleeding. No significant differences were found in the aforementioned complications between the bleeding and non-bleeding groups (Table 2).
Long-term follow-up outcomes
For 377 patients, a clinical follow up was documented through outpatient service or telephone visits. The average follow-up was 744 days (2.04 years), which yielded 768.5 patient-years in the overall cohort. As shown in Table 3, the average length of follow-up was comparable between the two groups (P=0.072). All the subjects received TEE examination and no difference was presented in TEE examination rate between groups.
Table 3 illustrates the clinical outcomes of an average 2-year follow-up. The incidence of ischemic stroke, TIA, systemic embolism, DRT, intracranial hemorrhage, gastrointestinal bleeding, other major bleedings, CV / unexplained death, or non-CV / unexplained death in the bleeding group was similar to that in the non-bleeding group, respectively. Additionally, there were no significant differences in major bleeding (P=0.917) and all-cause death (P=0.313) between the two groups. Patients with prior bleeding showed nonsignificant trends toward lower co-primary efficacy events compared to those with no prior bleeding (13.9% versus 21.7%, P=0.063). However, the overall rate of composite thromboembolic events in the cohort was 5.6%, with the bleeding group presenting a significantly lower rate compared with the non-bleeding group throughout the follow-up (2.2% versus 7.5%, P=0.031) (Table 3). After the adjustment of confounders which were chosen based on clinical relevance, such as age, ≥75 years, sex, hypertension, CHD, diabetes, chronic heart failure, previous stroke/TIA, liver dysfunction, and impaired renal function by using propensity score matching (PSM) in the overall cohort, 258 out of 377 patients remained, including 129 in the bleeding group and 129 in the non-bleeding group. Comparison of baseline clinical variables between groups before and after PSM are presented in Table S1 as an online data supplement. Kaplan–Meier survival curves and a log rank test demonstrated that the cumulative ratio of freedom from all-cause death was comparable between the bleeding and non-bleeding groups (HR: 1.065; 95% CI: 0.601-1.889; P=0.828) (Figure 2).
Comparisons between the observed annual risk of thromboembolic or bleeding events and the estimated annual risk
The estimated annual rate of thromboembolic events, based on CHA2DS2-VASc score, was 7.9 per 100 patient-years in the bleeding group and 6.6 per 100 patient-years in the non-bleeding group. However, 3 thromboembolic events (ischemic stroke in 1 patient and TIA in 2 patients) were observed in the bleeding group, and 18 (ischemic stroke in 11 patients and TIA in 7 patients) in the non-bleeding group. Thus, the observed annual rate of thromboembolic events was 1.1 per 100 patient-years, resulting in a relatively risk reduction (RRR) of 86.1% (χ2=15.8898, P<0.005) in the bleeding group, and 3.7 per 100 patient- years, resulting in a RRR of 43.9% (χ2=4.3758, P<0.05) in the non-bleeding group (Figure 3). Meanwhile, the expected annual rate of major bleeding events based on HAS-BLED score was 7.7 per 100 patient-years in the bleeding group and 6.9 per 100 patient-years in the non-bleeding group. In fact, during follow-up, 13 major bleeding events in the bleeding group and 22 in the non-bleeding group were observed. The observed annual rate of bleeding was 4.7 per 100 patient-years and 4.5 per 100 patient-years, which conferred a RRR of 39.0% (χ2=3.89, P<0.05) and a RRR of 34.8% (χ2=3.86, P<0.05) in the bleeding and non-bleeding groups, respectively (Figure 4).
Comparisons of LAAC efficacy in decreasing thromboembolic and bleeding risks between the bleeding and non-bleeding groups
The patients in the bleeding group had a statistically significantly higher RRR in the annual rate of thromboembolic events compared with those in the non-bleeding group (86.1% vs. 43.9%, χ2 = 7.03, P<0.01), whereas no significant difference was found for RRR in the annual rate of major bleeding between the two groups (39.0% vs. 34.8%, χ2 = 0.29, P>0.05).