Abstract
Background: Left atrial appendage closure (LAAC) is a safe and effective method for preventing embolic events in patients with non-valvular atrial fibrillation. However, peri-device leaks (PDLs) are sometimes unavoidable. Controversy exists regarding whether PDLs lead to embolic events. Objectives: This study aimed to explore the association between PDLs and embolic events, including ischaemic stroke, transient ischaemic attacks (TIAs), and systemic embolism (SE). Methods: We conducted a systematic search of the PubMed, Web of Science, MEDLINE, and Cochrane Library databases for studies published up to September 25, 2022, to compare the rate of ischaemic stroke/TIA/SE between the PDL group and the non-PDL group after LAAC. Results: Thirteen studies comprising 54,405 patients were included in the meta-analysis. The PDL group detected by transoesophageal echocardiography (TEE) had a significantly higher rate of ischaemic stroke/TIA/SE than the non-PDL group (OR: 1.20, 95% CI: 1.08–1.33, p = 0.0009). However, no difference in ischaemic stroke/TIA/SE was found between the PDL and non-PDL subgroups of the cardiac computed tomography angiography (CCTA) group (OR: 1.12, 95% CI: 0.51–2.50, p = 0.77). CCTA and TEE showed different rates of PDL detection, with the CCTA group having a higher rate of PDL detection (p < 0.0001), especially for trivial leaks. Conclusions: PDL detected by TEE increases the risk of embolic events after LAAC. However, no association was found between PDL and ischaemic stroke/TIA/SE in the CCTA group, which showed a higher rate of PDL detection than TEE, particularly for trivial leaks. In the future, CCTA may be used to explore the relationship between PDL size and ischaemic stroke/TIA/SE.
Introduction
Non-valvular atrial fibrillation (NVAF) is a common arrhythmia that poses a high risk for thrombotic events. Oral anticoagulants, such as warfarin and non-warfarin oral anticoagulants (NOACs), are the primary therapy for stroke prevention in NVAF patients [1, 2]. However, in patients with a history of bleeding or high risk for bleeding, left atrial appendage closure (LAAC) is a viable alternative strategy, as over 90% of thrombi occur in the left atrial appendage in NVAF patients [3, 4]. LAAC can be achieved through surgery or with devices implanted via catheterization. The PROTECT-AF, CAP trial, and CAP2 trial demonstrated a reduction in haemorrhagic stroke in the LAAC group compared to treatment with warfarin [5]. Additionally, LAAC was shown to be noninferior to NOACs in preventing major AF-related cardiovascular, neurological, and bleeding events [6]. However, peri-device leaks (PDLs) are a common occurrence, with a 1-year rate ranging from 12.5% to 40% [7‒9], varying between different devices [10]. The main cause of PDLs is a mismatch between the left atrial appendage and the closure device [11]. Although evidence suggests that patients in the ischaemic stroke/transient ischaemic attacks (TIAs)/systemic embolism (SE) group have a higher likelihood of PDL [12], the relationship between the occurrence of embolism events and PDL is still a topic of debate. This article aimed to evaluate the combination of ischaemic stroke/TIA/SE and PDL after LAAC.
Methods
Search Strategy
The present meta-analysis adhered to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) and the Cochrane Handbook. Two independent reviewers, Yin-Ge He and Shao-Hua Yang, conducted a comprehensive literature search in a range of online databases, including the Cochrane Library, Web of Science, MEDLINE, and PubMed, up to September 25, 2022, using the search terms “left appendage closure,” “left atrial appendage closure,” “atrial fibrillation,” “peri-device leak,” and “leak.” Moreover, potential publications were manually screened for relevant literature.
Study Design
The inclusion criteria for this study were as follows: (1) multicentre or single clinical trials; (2) participation of individuals diagnosed with the assistance of researchers Yin-Ge He and Shao-Hua Yang, with any disagreements being resolved through discussion with a third researcher, Yu-Jie Zhao. We considered articles that demonstrated the impact of PDLs on ischaemic stroke/TIA/SE and included non-English papers while excluding data that could not be extracted. Reviews were also excluded.
The Result of Extracting Data and Assessing Its Quality
Two independent researchers, Yin-Ge He and Shao-Hua Yang, extracted data for each eligible study, including study characteristics such as the first author, publication year, study design, number of patients, and follow-up duration, as well as variables such as study size, mean age, percentage of male patients, comorbidities (hypertension, diabetes, previous ischaemic stroke/TIA/SE), CHA2DS2-VASc score, HAS-BLED score, time for PDL detection, cardiac computed tomography angiography (CCTA) or transoesophageal echocardiography (TEE) usage, ischaemic stroke/TIA/SE after LAAC, type of device, and follow-up days. Any potential disagreements were discussed with a third researcher, Yu-Jie Zhao. Additionally, the risk of bias was assessed independently by Yin-Ge He and Shao-Hua Yang for each eligible study using the Newcastle-Ottawa Quality Assessment Scale [13], as shown in Table 1. Any disagreements were resolved with the help of a third researcher.
Newcastle-Ottawa Scale for Quality Assessment of the included studies
| Study | Selection | Comparability | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| representativeness of exposed cohort | selection of nonexposed cohort | ascertainment of expose | demonstration that outcome of interest was not present at start of study | adjust for the most important risk factors | adjust for other risk factors | assessment for outcome | fellow-up length | loss of follow-up rate | total quality score | |
| Bai et al. [16] (2011) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Viles-Gonzalez et al. [17] (2012) | 1 | 0 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 7 |
| Viles-Gonzalez et al. [18] (2012) | 1 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 5 |
| Pillarisetti et al. [19] (2015) | 1 | 0 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 7 |
| Sievert et al. [20] (2015) | 1 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 5 |
| Saw et al. [21] (2017) | 1 | 0 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 6 |
| Fink et al. [22] (2018) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Nguyen et al. [23] (2019) | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 5 |
| Staubach et al. [24] (2020) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Agudelo et al. [25] (2021) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Afzal et al. [26] (2021) | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 5 |
| Alkhouli et al. [27] (2022) | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 8 |
| Korsholm et al. [28] (2022) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Study | Selection | Comparability | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| representativeness of exposed cohort | selection of nonexposed cohort | ascertainment of expose | demonstration that outcome of interest was not present at start of study | adjust for the most important risk factors | adjust for other risk factors | assessment for outcome | fellow-up length | loss of follow-up rate | total quality score | |
| Bai et al. [16] (2011) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Viles-Gonzalez et al. [17] (2012) | 1 | 0 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 7 |
| Viles-Gonzalez et al. [18] (2012) | 1 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 5 |
| Pillarisetti et al. [19] (2015) | 1 | 0 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 7 |
| Sievert et al. [20] (2015) | 1 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 5 |
| Saw et al. [21] (2017) | 1 | 0 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 6 |
| Fink et al. [22] (2018) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Nguyen et al. [23] (2019) | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 5 |
| Staubach et al. [24] (2020) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Agudelo et al. [25] (2021) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
| Afzal et al. [26] (2021) | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 5 |
| Alkhouli et al. [27] (2022) | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 8 |
| Korsholm et al. [28] (2022) | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 4 |
Statistical Analysis
The statistical analyses were conducted using REVMAN software (version 5.4) and PRISM (version 9). Categorical variables are reported herein as frequencies or percentages, and mean values and standard deviations were calculated as described in the study by Wan et al. [14]. Pooled results are presented as the odd ratios (OR) of events with a 95% confidence interval (CI). The I2 statistic was used to measure between-study heterogeneity, with values of 0%, <25%, 25–49%, and >50% denoting no, low, moderate, and high heterogeneity, respectively. If the I2 value was greater than 50%, the between-study heterogeneity was considered significant, and a random effect model was adopted [15]. The χ2 test was used to compare differences in PDL and ischaemic stroke/TIA/SE rates among multiple groups. PRISM (version 9) was used to prepare the χ2 test and bar chart.
PDL Definition
The study used colour-flow Doppler to determine if there was any residual blood flow around the device in the LAA. In this analysis, a PDL was defined as a coloured jet seen around the device in at least two frames, regardless of the edges [16]. For identification by CCTA, the presence of contrast in the LAA beyond the device was considered indicative of a PDL [17].
Results
Search Results
The literature search retrieved 768 articles from PubMed, the Cochrane Library, Web of Science, and MEDLINE that were potentially relevant (see Fig. 1). After reviewing and assessing the inclusion and exclusion criteria, only thirteen articles [16‒28] met the criteria and were included in the meta-analysis. These articles, which included a total of 54,405 patients, were all observational trials.
Flow chart of the study selection procedure. Search and screening were used in this approach. The trial reports were identified and screened in this analysis, resulting in the retrieval of 768 reports. After removing duplicates and irrelevant ones, 24 reports were screened. Thirteen reports that met the criteria were included, and reasons for exclusion were also provided.
Flow chart of the study selection procedure. Search and screening were used in this approach. The trial reports were identified and screened in this analysis, resulting in the retrieval of 768 reports. After removing duplicates and irrelevant ones, 24 reports were screened. Thirteen reports that met the criteria were included, and reasons for exclusion were also provided.
Study Characteristics
The baseline characteristics of the individual studies are summarized in Table 2. The sample sizes of the studies varied, with the smallest being 22 and the largest being 51,333. Our analysis included 54,405 patients, of whom 54,270 underwent CCTA or TEE detection for PDLs across thirteen studies. We compared the rates of ischaemic stroke/TIA/SE with the PDL rate after LAAC postoperatively.
Baseline characteristics of the included studies
| Study | Design | Study size | With detected (PDL/non-PDL) | Mean age, years | Male | Hypertension | Diabetes | Previous ischaemic stroke/TIA/SE | CHA2DS2-VASc score, mean±SD | HAS-BL ED score | TEE/CCTA time for PDL, days | CCTA/TEE | Without anti-thrombus therapy rate | Ischemic stroke/TIA/SE after LAAC (PDL/non-PDL) | Type of device | Follow-up | Study |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Bai et al. [16] (2011) | 2011/China | Non-RCT | 58 | 20/38 | 74±9 | 64% | N.A | N.A | N.A | 2.2±1.0 | N.A | 45 | TEE | 94% | 1/0 | Watchman | 25.9±13.4 months |
| Viles-Gonzalez et al. [17] (2012) | 2012/America | Non-RCT | 485 | 182/263 | 72±9 | 14% | N.A | N.A | N.A | 2.2±1.2 | N.A | 45 | TEE | 87% | 5/11 | Watchman | 365 day |
| Viles-Gonzalez et al. [18] (2012) | 2012/America | Non-RCT | 22 | 13/9, 11/2 | 68±5 | 77.2% | N.A | N.A | 22 | 3.6±0.6 | N.A | 2.1 month | TEE, CT | 0% | 3/1, 3/1 | PLAATO | 58±9 month |
| Pillarisetti et al. [19] (2015) | 2015/America | Non-RCT | 479 | 79/395 | N.A | 27% | 150 | 48 | 60 | N.A | N.A | 365 | TEE | 40% | 2/4 | Watchman, Lariat | 365 days |
| Sievert et al. [20] (2015) | 2015/Germany | Non-RCT | 139 | 13/114 | 67.4±10.8 | 61% | 131 | 32 | 45 | 2.4±1.2 | 2.8±1.2 | 45 | TEE | 32% | 2/2 | Lariat | 2.9±1.1 years |
| Saw et al. [21] (2017) | 2017/Europe and Canada | Non-RCT | 344 | 39/272 | 74.4±7.5 | 67% | 290 | 98 | 122 | 4.3±1.5 | 3.0±1.2 | 134 | TEE | 24.7% | 1/8 | ACP | 355 days |
| Fink et al. [22] (2018) | 2018/Germany | Non-RCT | 76 | 12/23 | 70±9 | 52% | N.A | N.A | 6 | 3.0±1.0 | 3.0±1.0 | 115 | TEE | 21% | 0/1 | Lariat | 443 days |
| Nguyen et al. [23] (2019) | 2019/France | Non-RCT | 77 | 50/23 | 75.3±8.3 | 64% | N.A | N.A | N.A | 4.4±1.5 | 3.4±1.1 | 90 | CT | 25% | 2/0 | Amulet/ACP/Watchman | 236 days |
| Staubach et al. [24] (2020) | 2020/Germany | Non-RCT | 63 | 29/34 | 77 | 54% | 60 | 17 | 14 | 4.0±1.0 | 3.0±1.0 | 6 week to 3.1 years TEE | TEE | 22% | 3/2 | Watchman/Amulet/Occlutech/LAmbre | 3.1 years |
| Agudelo et al. [25] (2021) | 2021/Spain | Non-RCT | 137 | 78/59 | 76.8±7.2 | 64.2% | N.A | N.A | N.A | N.A | N.A | 3–6month | CT | 18% | 4/2 | Amulet/ACP | 638 days |
| Afzal et al. [26] (2021) | 2021/America | Non-RCT | 1,039 | 108/931 | 73±9 | 72% | N.A | N.A | N.A | 4.2±1.3 | N.A | 45 | TEE | 78% | 9/25 | Watchman | 208±137 days |
| Alkhouli et al. [27] (2022) | 2022/America | Non-RCT | 51,333 | 13,637/37,696 | N.A | 59% | 47,356 | 19,353 | 12,588 | 4.9±1.5 | 3.0±1.1 | 45 | TEE | 83% | 475/1,123 | Watchman | 425 days |
| Korsholm et al. [28] (2022) | 2022/Denmark | Non-RCT | 153 | 93/60 | 72.9±8.5 | 67% | 125 | 26 | 69 | 4.1±1.6 | 3.8±1.0 | 365 | CT | 0% | 11/7 | Amulet | 365 days |
| Study | Design | Study size | With detected (PDL/non-PDL) | Mean age, years | Male | Hypertension | Diabetes | Previous ischaemic stroke/TIA/SE | CHA2DS2-VASc score, mean±SD | HAS-BL ED score | TEE/CCTA time for PDL, days | CCTA/TEE | Without anti-thrombus therapy rate | Ischemic stroke/TIA/SE after LAAC (PDL/non-PDL) | Type of device | Follow-up | Study |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Bai et al. [16] (2011) | 2011/China | Non-RCT | 58 | 20/38 | 74±9 | 64% | N.A | N.A | N.A | 2.2±1.0 | N.A | 45 | TEE | 94% | 1/0 | Watchman | 25.9±13.4 months |
| Viles-Gonzalez et al. [17] (2012) | 2012/America | Non-RCT | 485 | 182/263 | 72±9 | 14% | N.A | N.A | N.A | 2.2±1.2 | N.A | 45 | TEE | 87% | 5/11 | Watchman | 365 day |
| Viles-Gonzalez et al. [18] (2012) | 2012/America | Non-RCT | 22 | 13/9, 11/2 | 68±5 | 77.2% | N.A | N.A | 22 | 3.6±0.6 | N.A | 2.1 month | TEE, CT | 0% | 3/1, 3/1 | PLAATO | 58±9 month |
| Pillarisetti et al. [19] (2015) | 2015/America | Non-RCT | 479 | 79/395 | N.A | 27% | 150 | 48 | 60 | N.A | N.A | 365 | TEE | 40% | 2/4 | Watchman, Lariat | 365 days |
| Sievert et al. [20] (2015) | 2015/Germany | Non-RCT | 139 | 13/114 | 67.4±10.8 | 61% | 131 | 32 | 45 | 2.4±1.2 | 2.8±1.2 | 45 | TEE | 32% | 2/2 | Lariat | 2.9±1.1 years |
| Saw et al. [21] (2017) | 2017/Europe and Canada | Non-RCT | 344 | 39/272 | 74.4±7.5 | 67% | 290 | 98 | 122 | 4.3±1.5 | 3.0±1.2 | 134 | TEE | 24.7% | 1/8 | ACP | 355 days |
| Fink et al. [22] (2018) | 2018/Germany | Non-RCT | 76 | 12/23 | 70±9 | 52% | N.A | N.A | 6 | 3.0±1.0 | 3.0±1.0 | 115 | TEE | 21% | 0/1 | Lariat | 443 days |
| Nguyen et al. [23] (2019) | 2019/France | Non-RCT | 77 | 50/23 | 75.3±8.3 | 64% | N.A | N.A | N.A | 4.4±1.5 | 3.4±1.1 | 90 | CT | 25% | 2/0 | Amulet/ACP/Watchman | 236 days |
| Staubach et al. [24] (2020) | 2020/Germany | Non-RCT | 63 | 29/34 | 77 | 54% | 60 | 17 | 14 | 4.0±1.0 | 3.0±1.0 | 6 week to 3.1 years TEE | TEE | 22% | 3/2 | Watchman/Amulet/Occlutech/LAmbre | 3.1 years |
| Agudelo et al. [25] (2021) | 2021/Spain | Non-RCT | 137 | 78/59 | 76.8±7.2 | 64.2% | N.A | N.A | N.A | N.A | N.A | 3–6month | CT | 18% | 4/2 | Amulet/ACP | 638 days |
| Afzal et al. [26] (2021) | 2021/America | Non-RCT | 1,039 | 108/931 | 73±9 | 72% | N.A | N.A | N.A | 4.2±1.3 | N.A | 45 | TEE | 78% | 9/25 | Watchman | 208±137 days |
| Alkhouli et al. [27] (2022) | 2022/America | Non-RCT | 51,333 | 13,637/37,696 | N.A | 59% | 47,356 | 19,353 | 12,588 | 4.9±1.5 | 3.0±1.1 | 45 | TEE | 83% | 475/1,123 | Watchman | 425 days |
| Korsholm et al. [28] (2022) | 2022/Denmark | Non-RCT | 153 | 93/60 | 72.9±8.5 | 67% | 125 | 26 | 69 | 4.1±1.6 | 3.8±1.0 | 365 | CT | 0% | 11/7 | Amulet | 365 days |
PDL, peri-device leaks; LAAC, left atrial appendage closure; TIA, transient ischaemic attack; SE, systemic embolism; TEE, transoesophageal echocardiography; CCTA, cardiac computed tomography angiogram; RCT, randomized controlled trial.
PDL Rates
In the thirteen research articles, the PDL rates ranged from 10.2% to 68.4%, as shown in Figure 2 [16‒28]. Of the patients in the TEE group, 26.2% had a PDL, while 61.7% of patients in the CCTA group had a PDL. The χ2 test revealed a significant difference in PDL rates among the thirteen observations (F = 4.503, p = 0.005). Two of the thirteen reports showed a significant difference in detecting PDLs using CCTA versus TEE (p < 0.0001), as illustrated in Figure 3 [23, 28].
The difference in PDL rate in the thirteen observations. The central illustration shows that Prism version 9 was used for mapping and data analysis in this study. The research articles included in the study reported PDL rates ranging from 10.2% to 68.4%. To compare the PDL rates between the observations, we conducted a χ2 test. The analysis revealed a significant difference in the PDL rates among the thirteen reports (F = 4.503, p = 0.005). PDL, peri-device leaks.
The difference in PDL rate in the thirteen observations. The central illustration shows that Prism version 9 was used for mapping and data analysis in this study. The research articles included in the study reported PDL rates ranging from 10.2% to 68.4%. To compare the PDL rates between the observations, we conducted a χ2 test. The analysis revealed a significant difference in the PDL rates among the thirteen reports (F = 4.503, p = 0.005). PDL, peri-device leaks.
![The difference in PDL rate between CCTA and TEE. Nguyen et al. [23] and Korsholm et al. [28] both utilized CCTA and TEE to assess the prevalence of PDL in patients with NVAF. CCTA identified PDL in 121 out of 181 patients, resulting in a PDL rate of 66.9%, while TEE identified PDL in 60 out of 181 patients, resulting in a PDL rate of 33.1%. The disparity in PDL rates between the two detection methods is significant (p < 0.0001). Therefore, it can be inferred that there is a considerable difference in the PDL rates detected through the two procedures. NVAF, non-valvular atrial fibrillation; CCTA, cardiac computed tomography angiogram; TEE, transoesophageal echocardiography; PDL, peri-device leaks.](/view-large/figure/14441202/000532019_f03.jpg)
The difference in PDL rate between CCTA and TEE. Nguyen et al. [23] and Korsholm et al. [28] both utilized CCTA and TEE to assess the prevalence of PDL in patients with NVAF. CCTA identified PDL in 121 out of 181 patients, resulting in a PDL rate of 66.9%, while TEE identified PDL in 60 out of 181 patients, resulting in a PDL rate of 33.1%. The disparity in PDL rates between the two detection methods is significant (p < 0.0001). Therefore, it can be inferred that there is a considerable difference in the PDL rates detected through the two procedures. NVAF, non-valvular atrial fibrillation; CCTA, cardiac computed tomography angiogram; TEE, transoesophageal echocardiography; PDL, peri-device leaks.
The difference in PDL rate between CCTA and TEE. Nguyen et al. [23] and Korsholm et al. [28] both utilized CCTA and TEE to assess the prevalence of PDL in patients with NVAF. CCTA identified PDL in 121 out of 181 patients, resulting in a PDL rate of 66.9%, while TEE identified PDL in 60 out of 181 patients, resulting in a PDL rate of 33.1%. The disparity in PDL rates between the two detection methods is significant (p < 0.0001). Therefore, it can be inferred that there is a considerable difference in the PDL rates detected through the two procedures. NVAF, non-valvular atrial fibrillation; CCTA, cardiac computed tomography angiogram; TEE, transoesophageal echocardiography; PDL, peri-device leaks.
Anticoagulant Regimen and Stroke Rate in the 13 Observations
After implantation, patients received anticoagulant treatment to allow for device endothelialization. However, due to the lack of a consistent guideline consensus on whether and for how long anticoagulant regimens should be continued after LAAC, the thirteen observational studies showed varying anticoagulant regimens (as shown in Table 3). Although there is no universal anticoagulant regimen, the effectiveness of the drugs can be evaluated by the ischaemic stroke/TIA/SE rate after LAAC (as seen in Fig. 4 [16‒28]). The χ2 test indicated that there was no significant difference in the ischaemic stroke/TIA/SE rate among the thirteen evaluated reports (F = 1.122, p = 0.42).
Anticoagulant regimen in the thirteen observations
| Change anticoagulant strategy after finding PDL | At hospital | At 3 months | At 12 months | |
|---|---|---|---|---|
| Bai et al. [16] (2011) | Yes | Warfarin + aspirin: 100% | Warfarin: 5% (45 days); APT: 95% | N.A |
| Viles-Gonzalez et al. [17] (2012) | Not in all PDL group | Warfarin: 100% | Warfarin: 13% (45 days); DAPT: 87% | Aspirin (life-long) |
| Viles-Gonzalez et al. [18] (2012) | No | Clopidogrel or ticlopidine | Clopidogrel or ticlopidine | Aspirin 325 mg/day indefinitely |
| Pillarisetti et al. [19] (2015) | No | NOACs: 62.5%; APT: 16.2% | N.A | NOACs: 5.6%; APT: 59.7% |
| Sievert et al. [20] (2015) | No | Aspirin: 60%; clopidogrel: 2%; aspirin plus clopidogrel: 6%; no treatment: 32% | N.A | N.A |
| Saw et al. [21] (2017) | No | Aspirin: 64.0%; clopidogrel: 17.4%; warfarin: 25.7%; NOACs: 2.7%; heparin: 18.3%; no treatment: 5.6% | N.A | DAPT: 62.3%; SAPT: 30.9%; NOACs: 6.2%; no treatment: 0.3% |
| Fink et al. [22] (2018) | Yes | Aspirin: 4.0%; NOACs: 46%; warfarin: 40%; heparin: 2%; no treatment: 8% | N.A | NOACs: 37%; no treatment: 63% |
| Nguyen et al. [23] (2019) | No | DAPT: 48.1%; SAPT: 39.0%; NOACs: 9.1%; NOACs + aspirin: 1.3%; no treatment: 2.6% | DAPT: 13.2%; SAPT: 51.3%; no treatment: 17% | DAPT: 10.2%; SAPT: 60.7%; no treatment: 25% |
| Staubach et al. [24] (2020) | Yes (leak >5 mm) | Leak >5 mm were detected: vitamin K antagonist together with aspirin; the remaining occludes dual antiplatelet therapy | Leak >5 mm were detected: vitamin K antagonist together with aspirin; the remaining occludes aspirin therapy | Aspirin indefinitely |
| Agudelo et al. [25] (2021) | No | N.A | SAPT: 41.6%; DAPT: 5.1%; NOACs: 30.6%; NOACs + APT: 6.5%; no treatment: 14.5% | N.A |
| Afzal et al. [26] (2021) | Yes | PDL >3 mm: OAC; PDL ≤3 mm: stop anticoagulation repeat imaging at 6–12 months to assess progress | N.A | N.A |
| Alkhouli et al. [27] (2022) | N.A | Aspirin: 35.5%; P2Y12 inhibitor: 12.8%; NOACs: 47.2%; warfarin: 45.1%; no treatment: 4.7% | N.A | Aspirin: 12.7%; P2Y12 inhibitor: 10.7%; NOACs: 2.4%; warfarin: 0.8%; no treatment: 83.5% |
| Korsholm et al. [28] (2022) | No | Aspirin: 100% | N.A | N.A |
| Change anticoagulant strategy after finding PDL | At hospital | At 3 months | At 12 months | |
|---|---|---|---|---|
| Bai et al. [16] (2011) | Yes | Warfarin + aspirin: 100% | Warfarin: 5% (45 days); APT: 95% | N.A |
| Viles-Gonzalez et al. [17] (2012) | Not in all PDL group | Warfarin: 100% | Warfarin: 13% (45 days); DAPT: 87% | Aspirin (life-long) |
| Viles-Gonzalez et al. [18] (2012) | No | Clopidogrel or ticlopidine | Clopidogrel or ticlopidine | Aspirin 325 mg/day indefinitely |
| Pillarisetti et al. [19] (2015) | No | NOACs: 62.5%; APT: 16.2% | N.A | NOACs: 5.6%; APT: 59.7% |
| Sievert et al. [20] (2015) | No | Aspirin: 60%; clopidogrel: 2%; aspirin plus clopidogrel: 6%; no treatment: 32% | N.A | N.A |
| Saw et al. [21] (2017) | No | Aspirin: 64.0%; clopidogrel: 17.4%; warfarin: 25.7%; NOACs: 2.7%; heparin: 18.3%; no treatment: 5.6% | N.A | DAPT: 62.3%; SAPT: 30.9%; NOACs: 6.2%; no treatment: 0.3% |
| Fink et al. [22] (2018) | Yes | Aspirin: 4.0%; NOACs: 46%; warfarin: 40%; heparin: 2%; no treatment: 8% | N.A | NOACs: 37%; no treatment: 63% |
| Nguyen et al. [23] (2019) | No | DAPT: 48.1%; SAPT: 39.0%; NOACs: 9.1%; NOACs + aspirin: 1.3%; no treatment: 2.6% | DAPT: 13.2%; SAPT: 51.3%; no treatment: 17% | DAPT: 10.2%; SAPT: 60.7%; no treatment: 25% |
| Staubach et al. [24] (2020) | Yes (leak >5 mm) | Leak >5 mm were detected: vitamin K antagonist together with aspirin; the remaining occludes dual antiplatelet therapy | Leak >5 mm were detected: vitamin K antagonist together with aspirin; the remaining occludes aspirin therapy | Aspirin indefinitely |
| Agudelo et al. [25] (2021) | No | N.A | SAPT: 41.6%; DAPT: 5.1%; NOACs: 30.6%; NOACs + APT: 6.5%; no treatment: 14.5% | N.A |
| Afzal et al. [26] (2021) | Yes | PDL >3 mm: OAC; PDL ≤3 mm: stop anticoagulation repeat imaging at 6–12 months to assess progress | N.A | N.A |
| Alkhouli et al. [27] (2022) | N.A | Aspirin: 35.5%; P2Y12 inhibitor: 12.8%; NOACs: 47.2%; warfarin: 45.1%; no treatment: 4.7% | N.A | Aspirin: 12.7%; P2Y12 inhibitor: 10.7%; NOACs: 2.4%; warfarin: 0.8%; no treatment: 83.5% |
| Korsholm et al. [28] (2022) | No | Aspirin: 100% | N.A | N.A |
PDL, peri-device leaks; NOACs, non-warfarin oral anticoagulants; SAPT, single antiplatelet therapy; DAPT, dual-drug antiplatelet therapy; APT, antiplatelet therapy.
Different ischaemic stroke/TIA/SE rates in the thirteen observations. We used Prism version 9 for data analysis and mapping to evaluate the impact of various anticoagulant therapies on rates of ischaemic stroke/TIA/SE across 13 observations. The χ2 test was employed to compare the ischaemic stroke/TIA/SE rate among these observations. Results of the test indicated no significant difference in the rates of ischaemic stroke/TIA/SE among the 13 observations (F = 1.122, p = 0.42). TIA, transient ischaemic attack; SE, systemic embolism.
Different ischaemic stroke/TIA/SE rates in the thirteen observations. We used Prism version 9 for data analysis and mapping to evaluate the impact of various anticoagulant therapies on rates of ischaemic stroke/TIA/SE across 13 observations. The χ2 test was employed to compare the ischaemic stroke/TIA/SE rate among these observations. Results of the test indicated no significant difference in the rates of ischaemic stroke/TIA/SE among the 13 observations (F = 1.122, p = 0.42). TIA, transient ischaemic attack; SE, systemic embolism.
Primary Efficacy Endpoint
The group with PDLs showed a significantly higher rate of ischaemic stroke/TIA/SE than the group without PDLs, as detected by TEE (OR: 1.20, 95% CI: 1.08–1.33, p = 0.0009). However, there was moderate heterogeneity observed in the study (I2 = 37%) (as shown in Fig. 5 [16‒22, 24, 26, 27]). In the CCTA group, there was no significant difference in the rate of ischaemic stroke/TIA/SE observed between the subgroup with PDLs and the subgroup without PDLs (OR: 1.12, 95% CI: 0.51–2.50, p = 0.77) (as shown in Fig. 6 [18, 23, 25, 28]).
Forest plot of ischaemic stroke/TIA/SE in the TEE group. The forest plot depicts the outcome for the PDL and non-PDL groups regarding the occurrence of ischaemic stroke/TIA/SE in the TEE group, based on 10 observations. The PDL group had a significantly higher rate of ischaemic stroke/TIA/SE when compared to the non-PDL group, as detected by TEE (OR: 1.20, 95% CI: 1.08–1.33, p = 0.0009). There was moderate heterogeneity observed in the trial (I2 = 37%). M-H, Mantel-Haenszel; PDL, peri-device leaks; TEE, transoesophageal echocardiography; TIA, transient ischaemic attack; SE, systemic embolism; OR, odd ratio.
Forest plot of ischaemic stroke/TIA/SE in the TEE group. The forest plot depicts the outcome for the PDL and non-PDL groups regarding the occurrence of ischaemic stroke/TIA/SE in the TEE group, based on 10 observations. The PDL group had a significantly higher rate of ischaemic stroke/TIA/SE when compared to the non-PDL group, as detected by TEE (OR: 1.20, 95% CI: 1.08–1.33, p = 0.0009). There was moderate heterogeneity observed in the trial (I2 = 37%). M-H, Mantel-Haenszel; PDL, peri-device leaks; TEE, transoesophageal echocardiography; TIA, transient ischaemic attack; SE, systemic embolism; OR, odd ratio.
![Forest plot of ischaemic stroke/TIA/SE in the CCTA group. Forest Plot of the Outcome: The forest plot displays the endpoint for PDL and ischaemic stroke/TIA/SE in the CCTA group, based on four observations (one study examined the correlation between PDL and ischaemic stroke/TIA/SE using both TEE and CCTA [18]). The findings indicate no significant difference in the rate of ischaemic stroke/TIA/SE between PDL and non-PDL in the CCTA group (OR: 1.12, 95% CI: 0.51–2.50, p = 0.77). Additionally, the trial showed low heterogeneity (I2 = 0%). M-H, Mantel-Haenszel; PDL, peri-device leaks; CCTA, cardiac computed tomography angiogram; TIA, transient ischaemic attack; SE, systemic embolism.](/view-large/figure/14441210/000532019_f06.jpg)
Forest plot of ischaemic stroke/TIA/SE in the CCTA group. Forest Plot of the Outcome: The forest plot displays the endpoint for PDL and ischaemic stroke/TIA/SE in the CCTA group, based on four observations (one study examined the correlation between PDL and ischaemic stroke/TIA/SE using both TEE and CCTA [18]). The findings indicate no significant difference in the rate of ischaemic stroke/TIA/SE between PDL and non-PDL in the CCTA group (OR: 1.12, 95% CI: 0.51–2.50, p = 0.77). Additionally, the trial showed low heterogeneity (I2 = 0%). M-H, Mantel-Haenszel; PDL, peri-device leaks; CCTA, cardiac computed tomography angiogram; TIA, transient ischaemic attack; SE, systemic embolism.
Forest plot of ischaemic stroke/TIA/SE in the CCTA group. Forest Plot of the Outcome: The forest plot displays the endpoint for PDL and ischaemic stroke/TIA/SE in the CCTA group, based on four observations (one study examined the correlation between PDL and ischaemic stroke/TIA/SE using both TEE and CCTA [18]). The findings indicate no significant difference in the rate of ischaemic stroke/TIA/SE between PDL and non-PDL in the CCTA group (OR: 1.12, 95% CI: 0.51–2.50, p = 0.77). Additionally, the trial showed low heterogeneity (I2 = 0%). M-H, Mantel-Haenszel; PDL, peri-device leaks; CCTA, cardiac computed tomography angiogram; TIA, transient ischaemic attack; SE, systemic embolism.
Discussion
Main Finding
Our report investigates the influence of PDLs on ischaemic stroke/TIA/SE after LAAC. Our findings are as follows: First, the PDL subgroup had a significantly higher rate of stroke/TIA/SE than the non-PDL subgroup of the TEE group. Second, there was no difference in the rate of ischaemic stroke/TIA/SE between the PDL and non-PDL subgroups of the CCTA group. Third, there was a significant difference between CCTA and TEE in detecting PDLs. The CCTA group demonstrated a higher detection rate of PDLs. Finally, our study suggests that ischaemic stroke/TIA/SE is more likely caused by an insufficient initial anticoagulant regimen and lack of an individualized anticoagulant regimen after LAAC.
Risk Factors Associated with PDL
A PDL, which is a significant limitation of LAAC, can occur for various reasons. The personalized orifice of the LAA and the circular closure device result in a mismatch that leads to PDLs between the LAA and the systemic circulation [29, 30]. Other factors that contribute to PDLs include the contractile nature of the LAA, suboptimal placement of the device during implantation, and anatomic remodelling of the LAA after LAAC [31]. Our research indicates that factors such as lobe-LAA axis misalignment, permanent NVAF, a larger landing zone diameter, a lower ratio of device compression, and more frequent off-axis positioning of the device increase the risk of PDLs. Apart from the operator’s experience and the degree of matching between the LAA orifice and the device, LAA remodelling also significantly contributes to PDLs. LAA remodelling may lead to reduced LAA compliance, particularly when a compressible LAAC device is deployed at a relatively low radial force, which increases the risk of PDLs due to the mismatch between the LAAC device and the LAA orifice [32]. These findings emphasize the importance of closely monitoring patients at high risk of LAA remodelling. In summary, a PDL is a major limitation of LAAC that arises due to various factors, including LAA orifice customization, device placement, and LAA remodelling. Identifying risk factors and monitoring patients at high risk of LAA remodelling can help mitigate the risk of PDLs.
PDL and Stroke
LAAC has implemented a strategy for preventing stroke in patients with atrial fibrillation. Studies have shown that LAAC is more effective than warfarin and NOACs in preventing haemorrhagic stroke and just as effective in preventing ischaemic stroke [5, 33‒36]. However, some studies have shown that LAAC is less effective than warfarin in preventing ischaemic stroke [37, 38]. Further analysis is needed to determine the causes of ischaemic stroke after LAAC. The management of PDLs after LAAC is still a topic of debate. Complete LAAC is a logical option for reducing the severity of cardioembolic strokes, but PDLs resulting from incomplete LAAC may lead to stagnant blood flow, thrombus formation, and embolization [39]. Freixa et al. [40] found that most cerebrovascular events after LAAC are not disabling and may be caused by small thrombi resulting from PDLs. A systematic review has summarized the current literature on PDL closure after LAAC and demonstrated its effectiveness in preventing ischaemic stroke/TIA/SE. The strategy of closing PDLs before discontinuing NOACs is a reasonable approach in clinical practice [41].
CCTA and TEE
The PDL incidence varies significantly and ranges from 3% to 53% between 45 days and 12 months post-LAAC in TEE studies [8, 10, 17, 28, 42, 43]. In contrast, the PDL rate in the CCTA studies ranges from 39% to 62% [44, 45]. In our study, the PDL rate ranged from 10.2% to 68.4%, with 26.2% of patients in the TEE group and 61.7% in the CCTA group having PDLs. The PDL incidence in our study was consistent with that in previous reports [46‒48]. We also found that the rate of ischaemic stroke/TIA/SE was significantly higher in the group with PDLs detected by TEE, but there was no difference in this rate between the PDL and non-PDL groups diagnosed by CCTA.
Unlike TEE, CCTA has a higher spatial resolution and can provide detailed 3D visualization, enabling the identification of incomplete endothelialization of the LAA [49]. However, the definition of a PDL as contrast enhancement on the delayed scan may require modification since it may represent flow through the device face, depending on the level of endothelialization [49]. This may result in an increase in false-positive rates. CCTA was found to be more sensitive in detecting PDLs, even trivial leaks (≤1 mm) that were missed on TEE [47, 50, 51]. Amulet IDE found that PDLs ≥3 mm were significantly associated with an increased risk for the composite of stroke or SE [52]. Furthermore, small PDLs, detectable by CCTA but not TEE, may not increase the risk of ischaemic stroke/TIA/SE. Future studies could investigate the discrepancy in PDL detection between CCTA and TEE to identify PDLs that more readily lead to embolic events.
Anticoagulant Regimen
After undergoing LAAC, anticoagulants and antiplatelet therapy should be administered to lower the risk of embolic events. However, it is common for anticoagulant programmes to be customized for individual patients. In two comprehensive observational LAAC registries, both including high-risk NVAF cohorts, corresponding annualized ischaemic stroke/TIA/SE rates of 2.3% and 2.0% were recently reported [53, 54]. Furthermore, our research uncovered three studies reporting increased ischaemic stroke/TIA/SE rates following LAAC. Viles-Gonzalez et al. prescribed single antiplatelet therapy (SAPT) following LAAC and continued anticoagulant treatment for 12 months. Staubach et al. initiated warfarin with aspirin solely for patients with PDLs >5 mm, while the other patients received dual-drug antiplatelet therapy (DAPT). Korsholm et al. [28] applied a uniform SAPT strategy following LAAC.
Endothelialization of the LAA device was observed to be complete within 90 days in dog models. However, the duration of endothelialization in humans remains unstudied and could potentially be prolonged compared to that in dogs, with significant interindividual variability. Therefore, physicians should personalize and closely monitor the duration of antithrombotic therapy [55]. According to studies conducted by Pillarisetti et al. and Bai et al., patients with a lower risk of ischaemic/TIA/SE tended to utilize warfarin and NOACs more frequently in the early stages of treatment.
The patients were treated with warfarin and aspirin (81 mg/day) for 45 days. After this period, they were treated with aspirin (325 mg/day) and clopidogrel (75 mg/day) for 6 months. Following the 6-month period, they were treated with aspirin (325 mg/day) alone, as observed in both the PROTECT-AF and PREVAIL trials. This treatment procedure was primarily based on empirical evidence and supported by preclinical dog studies, with the aim of providing antithrombotic coverage until complete endothelialization was achieved. It is worth noting that both the PROTECT-AF and PREVAIL trials were designed at a time when NOACs were not readily available. Among the non-warfarin regimens, DAPT is the most commonly used. Recent research indicates that patients treated with SAPT after left LAAC had a significantly higher rate of device-related thrombosis (DRT) than those treated with a warfarin regimen [55]. Our study found that patients treated with higher doses of warfarin or NOACs after LAAC had a lower rate of stroke. Thus, the ideal early anticoagulation regimen should involve warfarin or NOACs followed by DAPT, while the SAPT regimen may be recommended for patients who are at high risk of bleeding.
Conclusion
The detection of PDLs by TEE increases the risk of embolic events after LAAC, while no association was found between PDLs and ischaemic stroke/TIA/SE in the CCTA group. Furthermore, CCTA shows a higher rate of PDL detection than TEE, especially in cases with trivial leaks. Going forward, CCTA has the potential to be used to investigate the correlation between PDL size and the incidence of ischaemic stroke/TIA/SE.
Statement of Ethics
An ethics statement is not applicable because this study is based exclusively on published literature.
Conflict of Interest Statement
The authors disclosed no relevant relationships.
Funding Sources
This research was funded by the Henan Key Laboratory of Arrhythmia Medicine.
Author Contributions
Yin-Ge He and Yu-Jie Zhao conceived and designed the study. Shao-Hua Yang, Liang Xu, Yan Wang, Xu-Tan Qin, and Pan-Pan Chen independently assessed studies for possible inclusion and collected the data. Shao-Hua Yang extracted data from research. Yin-Ge He analysed the data and drafted the manuscript. All authors revised and approved the final version of the manuscript.
Data Availability Statement
All data generated or analysed during this study are included in this article. Futher enquiries can be directed to the corresponding author.
