There is evidence that PFO and ASD closure devices can be implanted without anticoagulation despite foreign material in the left atrium. The Amplatzer Cardiac Plug (St. Jude Medical) was a modification of the septal closure devices, and as a second-generation device, the instruction for use followed the same thromboprophylactic regime as the septal closure devices with only aspirin and clopidogrel advised at time of implantation. Several centres have reported that there were minimal or no thromboembolic events following this approach; however, the data were collected retrospectively, and the follow-up period was relatively short. The ASAP (ASA Plavix Feasibility Study With Watchman Left Atrial Appendage Closure Technology) study was a multi-centre, prospective non-randomised trial designed to assess the WATCHMAN device with 6 months of aspirin and clopidogrel followed by life-long aspirin in patients with a relative contraindication to anticoagulation [11]. The majority (93 %) of patients who were entered into the trial had a history of haemorrhagic/bleeding tendencies, which made them ineligible for warfarin. Six (4 %) of the 150 patients who had a Watchman device implanted in the ASAP study were found to have device-related thrombus on TEE . One of these patients had a stroke, and the others had thrombus detected on routine TEE screening at 3 or 12 months. This is very similar to the 3.8 % device-related thrombus rate in the PROTECT-AF study in which warfarin was continued post-procedure. Patients in the ASAP study who had thrombus were treated with 4–8 weeks of low molecular weight heparin. In terms of the primary efficacy outcomes, the ASAP study reported a rate of all-cause stroke or systemic embolism of 2.3 %, ischaemic stroke 1.7 % and haemorrhagic stroke 0.6 %. These compare favourably to the 3 %, 1.9 % and 0.3 % rates, respectively, for the PROTECT-AF trial although these were not done at similar training levels or on an intention-to-treat basis.

These studies suggest that DAPT for LAAO implantation is safe and feasible but does not establish whether it is superior to an initial period of anticoagulation. Furthermore, it is not known whether anticoagulation with warfarin, novel anticoagulants or DAPT is the safest in patients with risk factors or a history of bleeding. In the ACTIVE-W study, there was an unexpected finding that DAPT was not as effective as warfarin for stroke prevention and yet it caused significantly more minor bleeding events. Long-term aspirin use is associated with a two- to threefold increase in the risk of both peptic ulcer and non-ulcer upper and lower gastrointestinal (GI) bleeding events [12–14]. However, in patients receiving aspirin for secondary prevention of cardiovascular events, continuation of aspirin following an episode of serious upper GI bleed has been shown to be beneficial, i.e. the risk of recurrent GI bleed is still lower than the risk of cardiovascular events [15]. Similarly, patients with a GI bleed whilst on warfarin therapy for AF have a lower mortality risk if warfarin is restarted within a week without a significant increase in the rate of recurrent GI bleeds. Warfarin appears to be specifically linked to an increase in the risk of intracranial haemorrhage (ICH) [16]. Unfortunately, there is no consensus about restarting warfarin in patients who have suffered a haemorrhagic stroke whilst on warfarin for AF, although the risk of thromboembolic stroke and the location of the ICH are usually taken into account [17, 18]. When all types of bleeding are taken into account, there is no significant difference between the risk of bleeding with aspirin or warfarin therapy [16]; however, clopidogrel alone appears to carry a higher risk for non-fatal or fatal bleeds, as does DAPT and triple therapy, as would be expected [19]. Ideally, the uncertainties regarding anticoagulant and antiplatelet strategies in patients with a history of bleeding tendencies should be resolved by randomised studies; however, history of bleeding may be from a range of underlying pathology for both GI and ICH and may need to be taken into account when deciding on patient-specific therapies.

Incomplete closure is a potential outcome in any form of LAAO. Surgically closed LAA was shown to have residual flow of 60 % at TEE although surgical excision has a lower rate of incomplete LAA closure at 27 % compared to other forms of surgical LAA closure [20]. Intracardiac devices are circular in shape and the LAA is often oval leading to a leak along the side of the device. If a device is implanted deep in the appendage rather than at the ostium, then an ‘ante-chamber’ can be present and some of the device-related thrombi detected have been in devices implanted too deeply within the LAA. External ties such as the Lariat device can also result in a central channel with remnant flow. There is a concern that incomplete sealing of the LAA orifice may lead to the creation of a small pouch of stagnant blood, which could lead to thrombus formation and increase the risk of stroke post-device implant. On the other hand, these leaks may be too small to allow any large thrombi that develop to pass around the device, and thereby pose minimal risk of stroke. The PROTECT-AF investigators deemed a residual gap of <3 ± 2 mm, an acceptable amount of peri-device flow; however, the clinical implications of this cut-off were unknown.

A sub-study of PROTECT-AF investigated the frequency and clinical impact of incomplete left atrial appendage sealing and residual peri-device blood flow following implantation of the Watchman device (see Fig. 19.2). Peri-device flow was defined as minor (<1 mm), moderate (1–3 mm) or major (>3 mm), and its effect on the primary efficacy outcome of stroke, systemic embolism or cardiovascular death was assessed. The prevalence of any residual flow was 41 % at the 45-day TEE, 34 % at 6 months and 32 % at 12 months [21].

There was no significant difference in the rate of stroke, systemic embolism or cardiovascular death between patients with or without residual peri-device flow, nor between patients with minor, moderate or major peri-device flow [21]. In addition, there was no increase in primary efficacy events seen in those patients with residual peri-device flow who discontinued warfarin compared to those with no residual flow. Although these results suggest that it may be safe to stop warfarin at 45 days regardless of the presence of peri-device flow, they should be interpreted with caution since it is a retrospective analysis and may be confounded by continued anticoagulation therapy in some patients with residual flow.

The results of the PROTECT-AF sub-study and ASAP registry combined suggest that post-procedural warfarin may not be required unless device-related thrombus is present and even this does not invariably always cause stroke but certainly increases the risk.

Direct thrombin inhibition with dabigatran is approved for stroke prevention in patients with non-valvular AF and is superior (high-dose 150 mg bid) to warfarin in preventing thromboembolic events and associated with less ICH. Factor Xa inhibitors, rivaroxaban and apixaban, are another class of novel anticoagulants that have been shown to be non-inferior to warfarin for thromboprophylaxis and with lower ICH. These novel oral anticoagulants are increasingly being used to prevent stroke in patients with non-valvular AF in preference to warfarin given its multiple limitations of a narrow therapeutic window, many drug–drug and drug–food interactions and requirement for regular INR monitoring. Of note, all patients from the PROTECT-AF study who had a stroke or TIA, in whom there was an INR available, it was found to be sub-therapeutic. Theoretically, NOACs would maintain more stable anticoagulation and potentially limit these strokes in the early phase of endothelialisation over the device. However, NOACs come with a significant lifetime risk of major bleeding ranging from 1.4 to 3.0 % per year in clinical trials [22–26], and their use in patients with previous bleeds in particular would have to be undertaken with caution. The AVERROES (Apixaban Versus Acetylsalicylic Acid [ASA] to Prevent Stroke in Atrial Fibrillation Patients Who Have Failed or Are Unsuitable for Vitamin K Antagonist Treatment) study did investigate patients who had problems with warfarin, including inability to maintain the INR in the therapeutic range and medication compliance issues, but only a minority of these patients had a prior bleeding event. Furthermore, patients who had a recent serious GI bleeding event or active peptic ulcer disease were excluded from the trial [27]. The rates of bleeding were found to be similar in patients treated with aspirin or apixaban [28]. There is currently no study data for the use of NOACs in conjunction with LAAO devices; however, given the significant advantages of these agents over warfarin, a randomised study of the safety and efficacy of their use in the immediate post-procedural period following Watchman device implantation offers clear advantages over DAPT or warfarin thromboprophylaxis. The published series for the ACP device have used DAPT, and it is possible that short-term anticoagulation may not offer any advantage for this device and could possibly be used in those patients where device-related thrombus is identified.