This systematic review and meta-analysis demonstrated that (1) patients with AIS who received EVT and were treated with VKAs had higher risks of sICH and mortality, but such an effect was not observed for patients treated with DOACs. (2) Patients with AIS who underwent EVT and received confirmed therapeutic doses of VKAs had a lower risk of sICH as compared with total VKA patients, but not lower than those without OACs. (3) Positive outcomes were achieved more commonly by patients who did not receive OACs than by those who did. (4) Finally, similar rates of successful reperfusion were achieved by patients who received OACs and those who did not.

sICH is a serious concern in patients receiving EVT and carries the risk of unanticipated complications that can increase mortality. Previous studies have reported varying sICH rates in those receiving EVT, ranging from less than 5% to more than 16% [21, 22]. Studies have reported positive associations of female sex, diabetic mellitus, high blood pressure, lower Alberta Stroke Program Early CT score, and longer treatment interval with the risk of hemorrhage [23,24,25]. Tirofiban, an antiplatelet agent, was also demonstrated to be associated with an increased risk of sICH in patients receiving EVT [26]. However, this increased risk was not discovered in patients with lower platelet counts [27]. OAC treatment, especially with VKAs, is also controversial because of its potential to increase the risk of hemorrhage.

In this pooled analysis, VKA was associated with a higher risk of sICH. Although some previous research has not observed such an effect, Meinel and Ramos observed an increased risk in studies conducted with large sample sizes [16, 18]. A possible explanation for the higher rate of sICH is that atrial fibrillation, which carries a high risk of hemorrhagic transformation, is more prevalent among patients receiving VKAs. In addition, strong anticoagulant activity may lead to more severe hemorrhages. In most enrolled studies, patients who received therapeutic doses of VKAs did not have a significantly higher risk of sICH. This could be because patients who received therapeutic doses of VKAs may have had lower stroke severity because they received appropriate treatment. However, a similar result was not observed in the DOAC group, which had a similar risk of sICH as patients who did not receive OACs. The steady pharmacokinetic activity of DOACs may explain this result; most patients receive DOACs within the therapeutic range and experience low stroke severity. The lower risk of systemic bleeding of DOACs than of VKAs may also be responsible for this effect [28]. Most patients currently receiving VKAs could gradually shift to DOAC use, except for some patients with contraindications such as antiphospholipid syndrome or receiving mechanical valve transplantation. Increasing DOAC use may diminish the risk of sICH, and operators should be aware of the risk of hemorrhage if patients received VKAs before undergoing EVT procedures.

Mortality after EVT is another safety concern that we assessed. A pooled analysis of 15 studies revealed a mortality rate after EVT of approximately 15% [29]. Factors affecting mortality included age, pretreatment collateral arteries status, baseline blood glucose level, sICH, and baseline NIHSS score [30]. Despite the lack of adjustment for confounding factors in the present meta-analysis, mortality was higher in patients who received VKAs, even among those who received doses within the therapeutic range. The higher mortality rate was correlated with higher sICH risk in patients who received VKAs; this result supports the risk of hemorrhage being related to mortality. Because the rate of sICH was lower in the DOAC group, mortality for patients who received DOACs and those who did not was similar. However, although mortality was higher in patients who received VKAs, the benefits of EVT in AIS with LVO were considerable. Currently, intravenous rtPA is the gold standard therapy for AIS within 4.5 h. Most patients receiving EVT had undergone prior intravenous rtPA treatment, which was once considered a risk factor for intracranial hemorrhage. In our subgroup analysis of patients receiving EVT without concomitant rtPA, the risks of sICH and mortality were higher in the VKA group than the DOAC and control arms. These results were similar to those of the pooled analysis and may indicate that intravenous rtPA is not a major contributor to sICH or mortality risk; however, this hypothesis requires further study.

In a German study of the outcomes of EVT, postprocedural functional independence was achieved by approximately 37% of patients [31]. A similar rate of positive outcomes was achieved in our pooled analysis, but the rate was significantly lower in those who received VKAs. Higher risks of sICH and morality in patients receiving VKAs possibly contributed to this result. Such associations have been reported in several previous studies [32, 33]. Although some studies have suggested that successful reperfusion is also an independent predictor of favorable outcomes, [34] the recanalization rates were similar in our OAC and control arms, indicating that successful reperfusion is not a major contributor.

EVT was introduced nearly 20 years ago, and numerous novel devices have been developed since then; therefore, the successful reperfusion rate has increased over time. In the past, recanalization was thought to be easily achieved in patients with AIS and atrial fibrillation because the blood clots were more friable than those in in patients with atherosclerosis. However, in the present meta-analysis, successful reperfusion was achieved at similar rates in all groups, including patients who received OACs, who were more likely to have atrial fibrillation. Additionally, no differences were observed in the treatment effects of EVT between patients with LVO with or without atrial fibrillation [35]. In our pooled analysis, the recanalization rate was 80%–85%, which is similar to the rates observed in most other studies. However, most studies did not report the number of passes; therefore, it is difficult to draw conclusions regarding successful reperfusion because a higher number of passes increases the risk of vessel damage and hemorrhage.

The heterogeneity of the present meta-analysis was minimal, both in the overall and subgroup analyses, which indicates that our results are reliable. However, the present study has several limitations. First, all enrolled studies were retrospective, indicating that many confounding factors may have required adjustment. However, the present study did not adjust for these confounding factors; thus, the results should be interpreted with caution. Second, heterogeneity across studies is inevitable. Differences in populations, underlying diseases, device usage, and outcome measurement tools could influence the results, but such influence was minimized by statistical methods with the use of the random-effects model. Third, atrial fibrillation, which carries a high risk of hemorrhage, was more common in patients who received OACs than in those who did not, and this may have influenced the results. Those who received OACs may also have had comorbidities such as autoimmune disorders. This effect could not be adequately adjusted for in most enrolled studies. Finally, the dosage of OAC was not frequently reported. A low therapeutic dose is correlated with poor disease control, which may lead to a high risk of hemorrhage and mortality. Although the present study conducted a subgroup analysis for patients with and without confirmed therapeutic doses of OACs, the available data were still inadequate. Besides, the use of idarucizumab for dabigatran reversal was also not reported in enrolled studies, which was a relevant contributor to the risk of hemorrhage. However, despite these limitations, a higher risk of sICH was observed in patients with AIS receiving EVT who used VKAs compared with those who used DOACs or did not receive OACs. Further research is warranted.