Anesthesia for endovascular treatment in anterior circulation stroke: A systematic review and meta‐analysis

Abstract Background Endovascular treatment in patients with acute anterior circulation stroke could be performed under either conscious sedation (CS) or general anesthesia (GA). Although several studies have investigated the association between the clinical outcomes and the two anesthesia methods, consensus is lacking. Methods PubMed and EMBASE searches were used to select full‐text articles comparing the effects of GA and CS on functional outcome and complications in patients with anterior circulation ischemic stroke. Enrolled patients were assigned to receive endovascular treatment with CS or GA, with a primary outcome of functional independency within 90 days. Secondary outcomes included intracranial hemorrhage, all‐cause mortality at 90 days, pneumonia, and intraprocedural complications. Results Thirteen studies (3 RCTs and 10 observational studies), which included 3,857 patients (CS = 2,129, GA = 1,728), were eligible for the analysis. The overall analysis including the RCTs and observational studies demonstrated that the functional independence within 90 days occurred more frequently among patients with CS compared with GA (OR, 1.42; 95% CI, 1.05–1.92, p = 0.02); and the risk of mortality was higher with GA compared with CS; furthermore, CS was associated with lower rate of intracranial hemorrhage. In RCTs, GA was associated with increased functional independence (OR, 0.55; 95% CI, 0.34–0.89, p = 0.01) and successful reperfusion (OR, 0.51; 95% CI, 0.30–0.89, p = 0.02). Conclusions In the overall analysis and observational studies, CS was associated with improved functional outcomes and relatively safe for anterior ischemic stroke compared with GA. While the pooled data from RCTs suggested that GA was associated with improved outcomes. The inconsistency indicated that more large‐scale RCTs are required to evaluate what factors influenced the effect of the anesthesia methods on clinical outcomes.


| INTRODUC TI ON
Endovascular treatment (EVT) with mechanical thrombectomy is safe and effective in patients with acute anterior circulation stroke, compared with intravenous tissue plasminogen activator (IV-tPA) (Badhiwala et al., 2015;Elgendy, Kumbhani, Mahmoud, Bhatt, & Bavry, 2015;Kim, Jeon, Kim, Choi, & Cho, 2018;Marmagkiolis et al., 2015). However, the primary clinical outcomes are affected by many factors, such as the site of occlusion, stroke severity, and patient management factors including blood pressure during thrombectomy (Adams et al., 2007;Hungerford et al., 2016). Previous research has demonstrated that the anesthesia types would also impact the hemodynamic change (Jagani, Brinjikji, Rabinstein, Pasternak, & Kallmes, 2016), thereby influencing the outcomes of endovascular therapy. Currently used anesthetic techniques primarily include conscious sedation (CS) and general anesthesia (GA). However, there is a debate over which type of anesthesia is more beneficial to patients.
During endovascular treatment with GA, the airway is more protected, and the intraprocedural complications are less observed due to patient immobility (Li et al., 2014;Slezak et al., 2017). The unfavorable hemodynamic changes including hypotension and treatment delay are potential disadvantages of GA (Jagani et al., 2016). While the advantages of CS include that interventionalists can continuously monitor patient neurological functions during the procedure and the duration of time to complete endovascular treatment can be reduced (Li et al., 2014).
Previous retrospective studies comparing anesthesia methods during mechanical thrombectomy for anterior circulation ischemic stroke have concluded that CS is preferable to GA (Slezak et al., 2017;Whalin et al., 2014;Berg et al., 2015;Berkhemer et al., 2016;Jumaa et al., 2010;Abou-Chebl et al., 2014;Abou-Chebl et al., 2010;John et al., 2014;Nichols et al., 2010), other than one study (Bracard et al., 2016) which found that there was no difference in the functional independence of the two anesthesia methods. Currently, three RCTs compared the clinical outcomes of the various anesthesia methods for anterior circulation ischemic stroke, one of which, ANSTROKE (Löwhagen Hendén et al., 2017) showed that the clinical outcomes of the two anesthetic techniques were similar. The other two RCTs, SIESTA (Schönenberger et al., 2016) and GOLIATH (Simonsen et al., 2018), demonstrated that GA did not result in worse clinical outcomes compared with CS. Although a meta-analysis, Ilyas et al., (2018) found that there was no significant difference between the CS group and GA group for acute anterior circulation ischemic stroke using Solitaire stent retriever, there were some limitations such as few studies and the results from a mixture of prospective and retrospective studies. Furthermore, no meta-analysis has separately analyzed the current data of RCTs and observational studies for anterior circulation ischemic stroke. Therefore, we performed a meta-analysis of complete results from RCTs and observational studies to evaluate the association between the clinical outcomes and the anesthesia types during endovascular treatment for anterior circulation ischemic stroke.

| Search strategy and selection criteria
The major online databases, PubMed and EMBASE, were searched to identify the comparative studies on CS versus GA during endovascular treatment for acute anterior circulation ischemic stroke, from inception to January 2018, using the Medical Subject Heading (MeSH) terms and the keywords as follows: (a) the terms pertinent to the anesthesia methods including general anesthesia, conscious sedation, and local anesthesia; (b) the terms pertinent to the intervention of interest including endovascular, thrombectomy, intra-arterial, thromboembolism, fibrinolysis, and thrombolysis; and (c) the terms pertinent to the patient conditions including anterior circulation, ischemic, stroke, cerebrovascular accident, and infarct. The search terms were used in relevant combinations. In addition, previous systematic reviews and meta-analyses related to anesthesia management during mechanical thrombectomy were critically reviewed (Brinjikji et al., 2015;Campbell et al., 2018;Erickson & Cole, 2005;Ilyas et al., 2018;John, Mitchell, Dowling, & Yan, 2013).
The following inclusion criteria for this meta-analysis were used: (a) the studies that only included anterior circulation infarct; (b) the articles that compared the clinical results of CS with that of GA during endovascular treatment; (c) the researches that reported the modified Rankin scale (mRS) at 90 days in both CS and GA groups.
We also included post hoc analyses (Abou Chebl et al., 2015;Berg et al., 2015;Nichols et al., 2010) except for one study, Pfaff JAR et al (Pfaff et al., 2018), the results of which were duplicate with that of the SIESTA (Schönenberger et al., 2016). We excluded duplicate reports, abstracts that were not published as full-text reports in a journal and articles without mRS at 90 days in both CS and GA groups. Moreover, studies that included GA or CS only and studies that reported the posterior circulation stroke were also excluded.
Two investigators independently examined each study to determine whether to be included or excluded based on the selection criteria.
RCTs are required to evaluate what factors influenced the effect of the anesthesia methods on clinical outcomes.

K E Y W O R D S
acute ischemic stroke, anterior circulation, conscious sedation, endovascular treatment, general anesthesia, meta-analysis Disagreements between the two investigators were resolved by a third investigator. All researches obtained ethics approval from the local institutional boards at participating sites. Although this meta-analysis was not registered, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Moher, Liberati, Tetzlaff, & Altman, 2009) were followed.

| Quality assessment
Two reviewers independently used the Cochrane Collaboration's tool to assess the risk of selection bias, performance bias, detection bias, attrition bias, reporting bias, and other sources of bias (Higgins et al., 2011) among the RCTs. We assessed the quality of the observational studies using the Newcastle-Ottawa Scale (NOS), including selection, comparability, and outcomes (Stang, 2010).

| Data extraction and outcome definitions
Data were independently extracted by two investigators. The following characteristics were examined: (a) descriptive summary of each study (study name, author, year of publication, and total number of patients) and (b) patient characteristics (age, hypertension, atrial fibrillation, hyperlipidemia, diabetes mellitus, smoking, site of occlusion, and baseline NIHSS).
Our prespecified clinical endpoints included both primary and secondary outcomes. The primary outcome was functional independence, as defined by mRS scores (from 0 to 6) of 0-2 within 90 days.
Our secondary efficacy outcome was the proportion of patients with successful revascularization indicated by a modified thrombolysis in cerebral infarction (mTICI) score ≥2b (perfusion with distal branch filling ≥50%). Our secondary safety outcomes were intracranial hemorrhage (as defined by each trial), all-cause mortality at 90 days, pneumonia, and intraprocedural complications (including device-related complications, vessel perforation, dissection, and groin hematoma).

| Statistical analyses
The extracted data were analyzed by meta-analysis software including STATA version 11 (StataCorp, College Station, Texas, USA) and Review Manager (RevMan) version 5.3. Characteristics of patients are presented as numbers and percentages for categorical variables, and continuous data were expressed as means ± standard deviations (SDs). When the median, range, and sample size were provided, we estimated the mean and variance according to a formula (Hozo, Djulbegovic, & Hozo, 2005). Mean differences (MDs) and 95% confidence intervals (CIs) were calculated for pooled continuous variables.
Random-effects summary odds ratios (ORs) with corresponding 95% CIs were also constructed for the prespecified primary and secondary clinical endpoints, using RevMan with the DerSimonian and Laird random-effects model (DerSimonian & Laird, 2015).
Statistical heterogeneity was assessed by the I 2 statistic, with values <25%, 25%-50%, and >50% as low, moderate, and high degree of heterogeneity, respectively (Higgins, Thompson, Deeks, & Altman, 2003). To further estimate heterogeneity of the primary outcome, we performed subgroup and sensitivity analyses and meta-regressions. Funnel plots were used to visually evaluate publication bias, and Egger regressions were simultaneously used for quantification (Egger, Davey Smith, Schneider, & Minder, 1997). A two-tailed value of p<0.05 was considered statistically significant. To evaluate the features of patients in the GA and CS groups, we analyzed the demographic data (Supporting information Table   S1). In the overall analysis, hypertension and heart disease were  Table S1).

| RE SULTS
Overall, risk of bias was rated as low for RCTs, as assessed by the Cochrane Collaboration's tool (Supporting information Figure S1) and Newcastle-Ottawa Scale quality scores were at least 7 stars for the observational studies, indicating high quality, except for three studies (Jumaa et al., 2010, Abou-Chebl et al., 2015, and John et al., 2014 with 4 stars, revealing middle quality).

| Primary outcome
Compared with GA, the pooled data from 13 studies indicated that patients receiving CS had higher rates of functional independence within 90 days (OR, 1.42; 95% CI, 1.05-1.92, p = 0.02; I 2 = 74%; Figure 2a). For the observational studies, the primary outcome was in accordance with the total combined effect (OR, 1.79; 95% CI, 1.42-2.24; p < 0.0001, I 2 = 49%, Figure 2a and Supporting information Table S4). For the RCTs, GA was associated with significantly higher rate of functional independence than CS (OR, 0.55; 95% CI, 0.34-0.89, p = 0.01) by random-effects models with low heterogeneity (I 2 = 15%, Figure 2a and Supporting information  Table S3). Data of NIHSS scores and ASPECT scores were insufficient, so we did not conduct meta-regressions. Furthermore, there were only 3 RCTs, and there were no significant differences in the pooled demographic characteristics other than NIHSS scores; therefore, meta-regressions were not performed.

| Secondary outcomes
The secondary clinical endpoints in the present study included efficacy and safety outcomes in both groups (Figure 2 which were in accordance with that of the overall analysis ( Figure 2 and Supporting information Table S4). Other clinical endpoints were similar between two groups. There was no significant heterogeneity in any endpoints other than revascularization (I 2 = 63%, p = 0.01, Table S4).

Supporting information
There was no evidence of systematic bias, as visually assessed by funnel plots and quantitatively assessed by Egger tests, with details as follows: intracranial hemorrhage (p = 0.808), mortality endpoint (p = 0.134), and revascularization (p = 0.824 Supporting information Table S5).

| D ISCUSS I ON
This meta-analysis reports detailed analyses of 3 RCTs and 10 observational studies that compared GA with CS in patients with endovascular treatment for anterior circulation ischemic stroke. Our results indicated that CS was associated with improved functional outcome within 90 days, lower rates of intracranial hemorrhage, and mortality in the overall analysis and observational studies, compared with GA. However, for the RCTs, the rates of successful reperfusion and functional independence were higher in the GA group, with no differences between CS and GA in other secondary efficacy or safety outcomes.
Contrary to the four previous meta-analysis (Brinjikji et al., 2017;Campbell et al., 2018;Gravel et al., 2018;Ilyas et al., 2018) that  onset to EVT and no limit to the NIHSS scores (Slezak et al., 2017)), thus, they are more powered to estimate the safety endpoints.  (Campbell et al., 2016;Saber, Rajah, Kherallah, Jadhav, & Narayanan, 2018). However, except for the EKOS devices used in IMS II (Nichols et al., 2010), the other first thrombectomy devices were still used in many studies (Kim, Son, Kang, Hwang, & Kim, 2017;Lapergue et al., 2016). Thus, unlike the previous meta-analysis, Ilyas et al. (2018); Gravel et al. (2018), only including the modern devices, some observational studies (Abou Chebl et al., 2015;Jumaa et al., 2010) in our study utilized the first devices. Although we performed subanalysis to adjust for the thrombectomy devices, the heterogeneity was difficult to be eliminated. Patients with the different sites of occlusion may be related to the various severity of stroke.
Studies reported that the improved functional outcomes were more observed in the proximal arterial occlusion (Badhiwala et al., 2015).
The functional outcomes were not adjusted for the sites of occlusion due to the lacking data.
Systematic bias should be critically considered in a meta-analysis of published literature, and detection and adjustment for publication bias in statistical methods are common (Jin, Zhou, & He, 2015;Sedgwick, 2015). The fact that our meta-analysis only included patients with anterior circulation ischemic stroke, which was different from the previous study, Brinjikji et al. (2017), with a mixture of patients with anterior and posterior circulation ischemic stroke, could reduce the bias of results. Egger tests revealed no relationship between the assessments of OR and study size for the most clinical outcomes.
Our conclusions are limited by variability in the designs and study reporting, which is inherent to meta-analyses. Our meta-analysis only included patients with anterior circulation stroke, which may lead to insufficiency of some data. Thus, some factors could not be evaluated, such as the usage of intra-arterial thrombolytic agents and hemodynamics during anesthesia.

| SUMMARY
Our meta-analysis indicated that the results of the observational studies were in contrast to that of RCTs. By analyzing the inconsistencies, we found that patients with CS were associated with the improved functional outcomes when patients were in good conditions, but when the basic features and the time interval from onset to EVT were well balanced, the results were opposite. Thus, large-scale RCTs are required to fully elucidate what factors could influence the effects of the two anesthesia methods on the clinical outcomes during endovascular treatment for anterior circulation ischemic stroke.

COMPE TING INTERE S TS
All authors declare no competing interests.

D I SCLOS U R E S
All authors report no disclosures.