This was the first initial worldwide experience with the WATCHMAN device, which consisted of 75 patients enrolled in three European and four US centers [11]. In this open-label non-randomized pilot study, patients >18 years of age with a life expectancy of at least 2 years, with documented chronic or paroxysmal non-valvular AF, who had CHADS2 score of ≥1, and who were eligible for warfarin therapy were included. In this initial experience, a few complications occurred with the first-generation device (n = 16) with 3 device failures (2 embolizations, 1 delivery system failure). The device and the delivery system were altered with the second-generation device, and no further embolization occurred with the updated device/system (n = 59). Of the 75 patients, 66 (88 %) had successful implantation (7 had unsuitable anatomy, 1 core wire malfunction, and 1 unsuccessful transseptal sheath placement in LAA). Pericardial effusions occurred in 2 patients (2.6 %); one was related to an overly vigorous “tug test ,” and the tug technique had since been modified as the LAA is thin. At 45 days, 93 % (54/58) devices showed successful sealing of LAA on TEE (LAA completely sealed with absence of flow or with minimal flow around the device with jet of <3 mm). At a mean follow-up of 24 months, no ischemic stroke or systemic embolism occurred. There were two transient ischemic attacks (TIA): 1 at 4 months without thrombus visible on the device, and 1 at 6 months who had a smooth layer of thrombus on device surface. There were two deaths that were not device-related, and there were four device-associated thrombus (5.3 %). Overall, this pilot study provided preliminary data suggesting that LAA occlusion with WATCHMAN was safe and feasible, setting the stage for randomized studies.

Following the pilot feasibility study, the WATCHMAN device was studied in the multi-centre PROTECT AF (Watchman Left Atrial Appendage System for Embolic Protection in Patients With Atrial Fibrillation) [12] study. This prospective, randomized controlled trial involved Bayesian sequential design and was conducted at 59 sites in the U.S. and Europe from February 2005 to June 2008. There were 707 patients with non-valvular AF and ≥1 risk factor (age >75 years, hypertension, heart failure, diabetes, or prior stroke/TIA) who were randomized to WATCHMAN (n = 463) or continued warfarin (n = 244) in a 2:1 ratio. The mean CHADS2 score was 2.2 and mean age was 72 years. WATCHMAN was successfully implanted in 90.9 %. Warfarin was continued for 45 days with WATCHMAN and switched to clopidogrel for 4.5 months (if there was no peri-device leak, or leak was <5 mm on TEE at 45 days), with aspirin lifelong after implant. The composite primary efficacy event-rates (stroke, systemic embolism, and cardiovascular death) were 3.0 and 4.9 % (per 100 patient-years; relative risk 0.62) at 1065 patient-years follow-up with WATCHMAN and warfarin, respectively, meeting the criteria for non-inferiority (Table 11.1) [13]. However, the primary adverse outcome (procedure-related events and major bleeding) was higher with WATCHMAN (5.5 %/year) compared to warfarin (3.6 %/year; RR, 1.53; 95 % CI, 0.95–2.70). The incidence of serious pericardial effusion was 4.8 %, procedure-related stroke 1.3 % (majority related to procedural air embolism), and device embolization 0.6 %. Warfarin was discontinued in 86 % of patients at 45 days and 92 % at 6 months. At follow-up TEE, device-associated thrombus was seen in 4.2 %; however, device thrombus-associated stroke rate was only 0.3 % per 100 patient-years [14].

With longer follow-up at 1588 patient-years (mean 2.3 years), the primary efficacy event-rates were 3.0 % and 4.3 % in the WATCHMAN and warfarin groups, respectively (RR 0.71; 95 % CI 0.44–1.30 %), still meeting the non-inferiority criteria [15].

At 2621 patient-years (3.8 years) follow-up, the primary efficacy event-rates were 2.3 per 100 patient-years (95 % CI 1.7–3.2) with WATCHMAN and 3.8 per 100 patient-years (95 % CI 2.5–4.9) with warfarin (Table 11.2) [16]. This met both the superiority and non-inferiority criteria, demonstrating a 40 % risk reduction (rate ratio 0.6, 95 % CI 0.41–1.05) of all-cause stroke, systemic embolism, cardiovascular, and unexplained death with WATCHMAN. There was also statistically significant 85 % reduction in hemorrhagic stroke (RR 0.15, 95 % CI 0.03–0.49), 63 % reduction in disabling stroke (RR 0.37, 95 % CI 0.15–1.00), 60 % reduction in cardiovascular death (RR 0.4, 95 % CI 0.23–0.82), and 34 % reduction in all-cause mortality (RR 0.66, 95 % CI 0.45–0.98). The longer-term safety events of procedural safety events and subsequent major bleeding were not significantly different (RR 1.21, p = 0.41). Major bleeding occurred in 4.8 % with WATCHMAN versus 7.7 % with warfarin. In terms of ischemic stroke, beyond the peri-procedural period, the ischemic stroke events that accrued during follow-up were similar in both groups.