Atrial fibrillation (AF) is one of the most common arrhythmias in clinical practice among the elderly (1). Moreover, AF is related to increased risk of several serious adverse events such as ischemic stroke and systemic embolism (SE) (2). Research have shown that patients with AF are four to five times more likely to develop thrombotic and embolic events than the general population (3, 4). Therefore, stroke prevention with oral anticoagulation (OAC) and direct oral anticoagulants (DOACs) is an important part of the treatment regime in patients with AF (5). However, it might be difficult to use these drugs in the clinic, for drug interactions may increase the bleeding risk in patients. Due to these limitations, left atrial appendage closure (LAAC) is recommended for patients who are intolerant to OAC (6). Growing operator experience coupled with technical improvements and device modification have reduced procedural complications and accelerated continued growth in LAAC (7). However, device related thrombus (DRT) is still a known complication of LAAC. Until now, the prevalence and possible risk factors of DRT were discussed based on the randomized trials and large registries (8–11). Saw et al. (9) concluded that independent predictors of DRT were smoking and female sex, while Plicht et al. (10) thought CHADS₂ and CHA₂DS₂-VASc scores, platelet count, and ejection fraction were risk factors of DRT formation. Nevertheless, whether DRT increases the risk of stroke and SE remains controversial. In Cochet’s study (12), a non-significant trend for a lower rate of stroke and SE was observed in DRT group rather than normal patients; while in a large multicenter RCT led by Dukkipati et al. (13) concluded that a non-significant trend for higher occurrence rate of stroke and SE was observed in DRT group. It has been less clear if and to what degree the discovery of this finding increases the stroke risk (14). Recently, major published studies focused on the incidence and consequence of DRT, hence we conducted an updated meta-analysis of both latest observational cohort and randomized controlled trials (RCT) to investigate the association of DRT following LAAC with stroke and SE. Furthermore, the subgroup analysis was performed to explore whether there was a causal relationship between DRT and device types and post-implant anti-thrombotic strategy.

We performed a systematic review to assess the incidence of DRT and whether detected DRT is associated with a significantly elevated risk of ischemic events. The clinical trials in our meta-analysis included more than 12,800 patients (largest-to-date) and have established a solid evidence base supports that DRT is associated with a fivefold increase in ischemic events. The incidence of DRT after LAAC in included studies was 2.8%, which is lower than previous meta-analysis describing the occurrence rate of DRT (9). In the subgroup analysis of anti-thrombotic therapy, the incidence of DRT was similar between patients received post-implant APT or OAC. While the subgroup analyses based on occlusion mechanism showed the incidence of DRT was not statistically different in patients who underwent LAAC using DS or SS device. And the elevated risk of ischemic events associated DRT was not related to occlusion mechanisms of the device.

On the basis of the early successes of PROTECT-AF and PREVAIL trial, LAAC became rapidly adopted as stroke preventive strategy (20). Several major complications have been reported including pericardial tamponade and device embolization (29). However, DRT remains a major concern, with an estimated incidence of 4% (range 0–16% in eligible studies) (12, 30). In previous studies, plenty studies set 45 days after LAAC for DRT detection (31–35). However, DRT requires a certain development time after device implantation. Therefore, in combination with the definition of DRT in multiple literatures, the occurrence time of DRT was defined as at least 75 days after LAAC in our study. Studies which DRT reported within 3 months after LAAC were not included. Additionally, our meta-analysis included 59 eligible studies enrolled 12,845 patients with sample size ranging from 30 to 1,739. Compared with the previous meta-analysis, our study has included more literature published in the past 3 years. Also, more strictive definition of occurrence time was used in our study to calculate the pooled incidence of DRT. And this increased the credibility of our DRT incidence results. Therefore, the major procedure-related complication incidence has improved over time and with increasing physician. Furtherly, improvements can be made to the current device to further reduce the incidence of DRT (WATCHMAN FLX vs. WATCHMAN 2.5; Amulet vs. ACP) (36).

Current published reports suggests that DRT is associated with a significantly elevated risk of ischemic events (9, 13, 26). A significant positive association between the incidence of DRT and subsequent ischemic events was also observed in our meta-analysis, documented approximately fivefold greater rates of ischemic events in patients with DRT compared with those without DRT. Among patients who developed DRT and stroke, majority of them were diagnosed with DRT prior to the occurrence of stroke. Although different clinicians did not have a uniform opinion on the relationship between DRT and stroke, the persistent signal of increased ischemic events warrants attention. However, current result together indicate that DRT after LAAC represents a significant danger in patients with AF who are already at high risk for stroke or TIA. Ding et al. (37) suggested that the high-risk AF patients have a high rate of mortality and an ongoing significant risk of bleeding and thrombotic events during follow-up. While Teiger et al. (38) concluded that LAAC in high-risk patients seems reasonable to decrease the rate of stroke. The limitations of the uncertain definition of DRT occurrence time reduce the credibility of their results. It is noteworthy that the thrombus development after LAAC is not clear since DRT is mostly silent and its follow-up is highly depended on imaging tools. As a result, it is difficult for clinicians to diagnose the DRT occurrence time precisely. It is important to detect them in a timely manner during follow up through appropriate diagnostic imaging techniques such as coronary angiography and trans-esophageal echocardiography. Our findings suggest that DRT is associated with more severe complications (stroke and SE), so early diagnosis and treatment of DRT may lead to better prognosis of patients. However, some studies have shown that anticoagulant therapy has an effect on the endothelialization process (39–41), and the occurrence of DRT is related to the incomplete endothelialization of device, so the early use of anticoagulant to treat DRT may hinder the completion of endothelialization. Recent studies indicated a reduced risk of DRT in patients receiving post-LAAC anticoagulation (42).

Within the last decade, several LAAC devices have been developed and introduced into clinical practice. The included clinical trials in our study also covered 3–4 types of devices. In our study, the devices are divided into two categories, one is SS including Watchman, FLX, Occlutect, and the other is DS including ACP, Amulet, LAmbre (16). The subgroup analyses showed that the incidence of DRT was not statistically different between SS and DS group, and the occurrence rate of stroke and SE associated DRT was similar between two groups. Based on these results, it seemed that the safety for LAAC was comparable between devices under different occlusion mechanisms. As more clinical trials comparing the clinical outcomes between SS and DS devices are underway, it is important to consider not only the baseline characteristics of the patient but also the differences in device mechanisms when formulating post-operative therapy.

Identifying predictors or risk factors of DRT is important for DRT prevention. As mentioned in the preceding text, the independent predictors of DRT occurrence remained unknown. In our current meta-analysis, the predictors of DRT were history of stroke/TIA and CHA₂DS₂–VASc score, while the conclusion among studies were inconsistent. Saw et al. (9) concluded that smoking and female sex were independent predictors of DRT, while Plicht et al. (10) suggested that predictors might be CHADS₂ and CHA₂DS₂–VASc scores, platelet count, etc. In our previous real-world study, peri-device leak was considered as a risk factor of DRT (43). In recent two studies deep implantation depth was found to be a risk factor for DRT (7, 42). Though other predictors such as pericardial effusion and renal insufficiency were found, these predictors should be confirmed in larger studies (7). Until more data become available, researchers should raise more concern on the treatment of DRT, though the selection of treatment is still in debate. Simard et al. (44) concluded that 8–12 weeks Vitamin K antagonist (VKA), NOAC, and 2–4 weeks LMWH were recommended for treatment of DRT. Sedaghat et al. (45) revealed that patients underwent the three regimens mentioned above showed no statistical difference in clinical outcomes. Patients diagnosed with DRT may have persistent DRT or DRT recurrence after receiving anti-thrombotic therapies. Recent studies reported that only approximate a quarter of cases demonstrating persistent DRT presence in clinical follow-up (7, 45). And patients with persistent DRT showed higher stroke rates and increased mortality rates (45). And Asmarats et al. (46) firstly evaluated the recurrence of DRT, suggesting that thrombus recurrence was common, thus long-term OAC was encouraged after a first DRT. More specified anti-thrombotic regimens and more RCT are needed in prevention of DRT. Except for DRT, late DRT is thought to be less frequent because of device sealing, which remains variable on an individual basis. Furthermore, the collection of late DRT in this study is not comprehensive enough to carry out effective analysis. Consistent with our results, Sedaghat et al. (45) suggested that no clinical or echocardiographic predictors for late DRT formation could be identified in their study.

This meta-analysis has several limitations. First, because of limited randomized data, this meta-analysis included both randomized and observational studies. The observational studies are subjected to unmeasured confounding and selection bias. Therefore, consistent with previous meta-analysis of DRT (47), we conducted sensitivity analysis by including MCR and RCT only to test the inconsistency, which could verify the feasibility of the inclusion of multiple studies. Second, the definition of some clinical events such as major procedure-related complications was not unanimous across studies; however, it was less likely to have a huge impact on our final conclusion. Third, the follow-up duration in each study varied, which may have negative influence on outcomes. Last but not least, due to the paucity of individual data in each eligible study, no subgroup analysis was done according to different generations of SS and DS.

Device related thrombus is an infrequent complication of LAAC, associated with increased rates of ischemic events. The incidence of DRT was comparable between SS and DS devices. The occurrence rate of stroke and SE associated DRT was similar in devices with different occlusion mechanisms. Further large multicenter prospective studies are needed to confirm the true prevalence of DRT and to evaluate the risk factors, associated complications and treatment regimens.

The original contributions presented in this study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

SZ: data curation (lead), formal analysis (equal), investigation (equal), methodology (equal), software (lead), validation (equal), visualization (equal), writing – original draft (equal), and writing – review and editing (equal). S-HX: data curation (equal), formal analysis (equal), investigation (equal), methodology (equal), validation (equal), visualization (equal), writing – original draft (equal), and writing – review and editing (equal). Y-GG: methodology (equal), software (lead), and writing – review and editing (equal). X-XZ and Z-FG: supervision (equal), investigation (equal), methodology (equal), and writing – review and editing (equal). Y-WQ: supervision (equal), investigation (equal), and writing – review and editing (equal). YB: conceptualization (lead), resources (lead), writing – original draft (equal), writing – review and editing (equal), supervision (lead), and taken responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. All authors contributed to the article and approved the submitted version.