Comparison of Risk Factors, Safety, and Efficacy Outcomes of Mechanical Thrombectomy in Posterior vs. Anterior Circulation Large Vessel Occlusion

Background and Purpose: It is believed that stroke occurring due to posterior circulation large vessel occlusion (PCLVO) and that occurring due to anterior circulation large vessel occlusion (ACLVO) differ in terms of their pathophysiology and the outcome of their acute management in relation to endovascular mechanical thrombectomy (MT). Limited sample size and few randomized controlled trials (RCTs) with respect to PCLVO make the safety and efficacy of MT, which has been confirmed in ACLVO, difficult to assess in the posterior circulation. We therefore conducted a meta-analysis to study to which extent MT in PCLVO differs from ACLVO. Materials and Methods: We searched the databases PubMed, Cochrane, and EMBASE for studies published between 2010 and January 2021, with information on risk factors, safety, and efficacy outcomes of MT in PCLVO vs. ACLVO and conducted a systematic review and meta-analysis; we compared baseline characteristics, reperfusion treatment profiles [including rates of intravenous thrombolysis (IVT) and onset-to-IVT and onset-to-groin puncture times], recanalization success [Thrombolysis In Cerebral Infarction scale (TICI) 2b/3], symptomatic intracranial hemorrhage (sICH), and favorable functional outcome [modified Rankin Score (mRS) 0–2] and mortality at 90 days. Results: Sixteen studies with MT PCLVO (1,172 patients) and ACLVO (7,726 patients) were obtained from the search. The pooled estimates showed higher baseline National Institutes of Health Stroke Scale (NIHSS) score (SMD 0.32, 95% CI 0.15–0.48) in the PCLVO group. PCLVO patients received less often IVT (OR 0.65, 95% CI 0.53–0.79). Onset-to-IVT time (SMD 0.86, 95% CI 0.45–1.26) and onset-to-groin puncture time (SMD 0.59, 95% CI 0.33–0.85) were longer in the PCLVO group. The likelihood of obtaining successful recanalization and favorable functional outcome at 90 days was comparable between the two groups. PCLVO was, however, associated with less sICH (OR 0.56, 95% CI 0.37–0.85) but higher mortality (OR 1.92, 95% CI 1.46–2.53). Conclusions: This meta-analysis indicates that MT in PCLVO may be comparably efficient in obtaining successful recanalization and 90 day favorable functional outcome just as in ACLVO. Less sICH in MT-treated PCLVO patients might be the result of the lower IVT rate in this group. Higher baseline NIHSS and longer onset-to-IVT and onset-to-groin puncture times may have contributed to a higher 90 day mortality in PCLVO patients.


INTRODUCTION
Intravenous thrombolysis (IVT) has become the mainstay of acute intervention in ischemic stroke presenting within 4.5 h of symptom onset when other contraindications have been excluded (1). However, IVT has been shown to be less effective in proximal large vessel occlusion (LVO), mainly in the terminal internal carotid artery, proximal middle cerebral artery, and basilar artery, than in more distal occlusion (2). Therefore, clinical worsening is to be expected in many cases of LVO unless endovascular mechanical thrombectomy (MT) is initiated (2).
The second-generation MT devices that were introduced in the last decade have shown superiority to first-generation MT devices and, hence, have been widely used in MT since then (3)(4)(5). Consequently, it could be argued that most studies on MT conducted before the surge of second-generation MT devices could have been compromised by the inferiority of first-generation MT devices. Nowadays, MT in LVO may be conducted up to 24 h without waiting for IVT outcome (6)(7)(8).
To date, many randomized controlled trials (RCTs) have reported the safety and efficacy of MT in acute ischemic stroke due to anterior circulation LVO (ACLVO). However, there is lack of substantial data on the safety and efficacy of MT in posterior circulation LVO (PCLVO) (5,6,9).
Posterior circulation stroke is defined as the development of ischemic lesions occurring in the vascular territories supplied by branches of the vertebrobasilar arterial system (10). It occurs in about 20-25% of all ischemic strokes (11,12), and neurological deficits caused by PCLVO have been described as catastrophic with severe disability and death occurring in about 68% of patients (13,14). The rarity of PCLVO poses the challenge of obtaining a significant sample size for conducting observational and controlled trials in comparison to anterior circulation stroke.
MT in ACLVO has been accepted in most clinical settings as the best way for obtaining recanalization, and therefore, the randomization of patients with PCLVO into groups including no-MT is considered mostly unethical. Among the very few RCTs that focused on posterior circulation stroke, the Basilar Artery Occlusion Endovascular Intervention vs. Standard medical Treatment (BEST) RCT was terminated due to loss of equipoise, which resulted from a high crossover rate and was topped by a small sample size. This trial, however, reported no difference in favorable outcomes of MT patients and those receiving only standard medical treatment including IVT (15). On the other hand, a larger non-randomized cohort study, the Endovascular Treatment for Acute Basilar Artery Occlusion (BASILAR) study, reported that MT within 24 h of estimated occlusion time in basilar artery occlusion patients is associated with better functional outcomes and reduced mortality (16). Data on the safety and efficacy of MT in PCLVO from the randomized controlled Basilar Artery International Cooperation Study (BASICS) and Basilar Artery Occlusion Chines Endovascular Trial (BAOCHE) trials are pending (17,18). Available data show a strong probability of differences in MT in PCLVO and ACLVO, which may contribute to their safety and efficacy outcomes (19,20). An improved functional outcome and reduced mortality in moderate-to-severe ACLVO stroke patients have been shown to be dependent on a small infarct core, moderateto-good collateral circulation, and rapid MT (21). Some PCLVO studies have associated MT with a poor outcome despite having a high recanalization rate, and this has raised interest in possible predicting factors of outcome in PCLVO such as initial stroke symptom severity, collateral status, age, infarct volume, stroke etiology, respiratory insufficiency, and other comorbidities (22)(23)(24)(25)(26). Bad outcome could also be a consequence of a delayed treatment since symptoms of posterior circulation stroke are known to be often fluctuating with about 55-63% cases of prodromal transient ischemic attack in spite of a persistent vessel occlusion (14).
Due to the conflicting nature of available studies, we conducted a systematic review and meta-analysis on studies published from 2010 to January 2021 with data comparing MT in PCLVO vs. ACLVO in order to assess the differences of risk factors, as well as safety and efficacy outcomes between both circulations.

MATERIALS AND METHODS
This study was conducted in accordance with the guidelines of the Preferred Reporting Item for Systematic Reviews and Meta-Analysis (PRISMA) (27).

Data Source and Searching
We conducted a database search in PubMed, Cochrane library, and EMBASE before January 23, 2021, for literature from 2010 to 2021 using the following medical search heading (MeSH) and keywords: "acute stroke, " "mechanical thrombectomy, " "endovascular treatment, " "posterior circulation, " "vertebrobasilar occlusion, " "anterior circulation, " and "large vessel occlusion" (see Supplementary Methods). The PubMed search strategy was adapted for use in Cochrane library and EMBASE search databases. No restrictions were made in relation to literature type and text availability. Literature was however screened for study suitability based on title and abstract. Only subject-relevant studies were therefore assessed for eligibility.

Study Selection and Data Extraction
Studies were included if they met the following criteria: (1) retrospective or prospective observational studies with a combined sample size for PCLVO and ACLVO of at least 40, (2) comparison of baseline characteristics and at least two reperfusion treatment profile parameters (i.e., rate of IVT, onset-to-IVT time, onset-to-groin puncture time, onset-torecanalization time, and number of passages) in PCLVO and ACLVO as main and/or subgroup analysis, and (3) outcome defined by at least two of the following: Thrombolysis In Cerebral Infarction scale (TICI) 2b/3 (28), symptomatic intracranial hemorrhage (sICH), modified Rankin Scale score (mRS) 0-2 at 90 days, and mortality at 90 days. Exclusion criteria included the following: (1) non-English literature, (2) no MT conducted, (3) duplicate literature, (4) insufficient data for comparison purposes, and (5) same datasets used by multiple studies. Duplicates were identified and eliminated using EndNote X9 citation manager software (Clarivate, Philadelphia, PA, USA).

Quality Assessment
Quality assessment by means of Risk of Bias Assessment tool for Non-randomized Studies (RoBANS) was performed to assess the methodological quality of all included studies under which studies were rated as having either a high, a low, or an uncertain risk of bias (29).
Data search, eligibility assessment, selection, and extraction as well as quality assessment were conducted and crosschecked by two independent investigators and contentions were resolved through a consensus between the two. Publication bias was assessed by means of a funnel plot asymmetry.

Statistical Analysis
Statistical analyses were performed using Review Manager (RevMan) [Computer program], version 5.4.1 (30). We used the Mantel-Haenszel statistical method and a random or fixed effects analysis model for studies with moderate/high and low heterogeneity, respectively, to estimate the pooled effect size. Cochrane I 2 statistics was used to assess heterogeneity. We defined high heterogeneity as I 2 > 75%, moderate heterogeneity as 25% < I 2 < 75%, and low heterogeneity as I 2 < 25% (31). PCLVO and ACLVO were the comparison groups, and the corresponding meta-analysis was performed for each outcome of interest.
Sensitivity analysis was performed for baseline characteristics and recanalization treatment profiles in which retrospective studies were excluded due to higher risk of selection bias (29). We furthermore conducted subgroup analyses in which we excluded studies with <20 patients with PCLVO due to a possible risk of lack of representation of PCLVO with low sample sizes in the real world and studies that primarily recruited patients who received MT until 2012, with the presumption that results could have been compromised by the use of first-generation MT devices (32).
Odds ratios (ORs) and standardized mean difference (SMD) were calculated with 95% confidence intervals (CIs) and p < 0.05 was considered significant. For continuous data, the corresponding estimated mean and estimated standard deviation were calculated (33).

Risk of Bias Assessment
Data on risk of bias are shown in Supplementary Figure 1; generally, there was a high risk of bias with respect to patient selection, confounding variables, and outcome reporting and a lower risk of incomplete data across all studies.
The funnel plots showed no asymmetry with respect to sICH, indicating a low probability of publication bias across the included studies for sICH. Asymmetry was, however, observed for recanalization success, favorable functional outcome, and mortality (Supplementary Figure 2).

Baseline Characteristics
Results of age, sex, baseline NIHSS, and site of LVO for the included studies are reported in Table 1 Figures 7-9, respectively).

Outcomes of Study
Results of TICI 2b/3, sICH, mRS 0-2 at 90 days, and mortality for the included studies are reported in Table 2.

Successful Recanalization
Studies that reported successful recanalization (i.e., TICI 2b/3) showed moderate heterogeneity (I 2 = 50%, p = 0.01). The pooled estimates showed no difference in outcomes in both PCLVO and ACLVO [OR = 1.07 (95% CI 0.81-1.42), p = 0.44]. In a subgroup analysis in which three studies were excluded on the basis of <20 PCLVO patients and patient recruitment primarily until 2012, the remaining studies showed similar results [OR FIGURE 7 | Forest plot comparing "onset-to-groin puncture time" in patients with large vessel occlusion in the posterior circulation (PC) vs. anterior circulation (AC) who were treated with endovascular mechanical thrombectomy. Chi 2 , chi-square statistic; CI, confidence interval; df, degrees of freedom; I 2 , I-square heterogeneity statistic; IV, weighted mean difference; P, p-value; SD, standard deviation; Std., standardized; Tau 2 , estimated variance of underlying effects across studies; Z, Z statistic.

DISCUSSION
To the best of our knowledge, until the conduction of this metaanalysis, there had been two prior meta-analyses comparing MT in anterior and posterior circulation stroke with both studies, however, focusing on MT safety and efficacy outcomes (19,20). This study, conducted independently from previous studies, included more recent literature on MT in PCLVO and ACLVO and, in a detailed meta-analysis, further sought to compare demographics and baseline characteristics, risk factors, as well as recanalization treatment profiles between the two brain circulations. Hence, this study presents at the time of publication the most current data on MT in PCLVO vs. ACLVO. Results on etiology of LVO in our meta-analysis show large artery atherosclerosis (36.7% and 23.1%) and cardiac embolism (34.8 and 47.0%) to be the most common causes of PCLVO and ACLVO, respectively (Supplementary Table 2), with large artery atherosclerosis and cardiac embolism being an equally likely etiology in PCLVO. This was consistent with reports from previous literature that reported 26-36% for large artery atherosclerosis and 30-35% for cardiac embolism in PCLVO, although this was based only on basilar artery occlusion (14). The difference in stroke etiologies for the other classification groups such as other determined LVO etiologies (dissection, thrombophilia, paraneoplastic, etc.) and unknown causes, might, however, be inconclusive due to a possible lack of standardized classification of etiology across studies (49).
It is thought that NIHSS gives more weight to neurological deficits in anterior circulation stroke due to factors such as aphasia, facial palsy, and hemiparesis as opposed to limb ataxia, oculomotor disorders, and hemianopia in posterior circulation stroke. However, depending on the level of occlusion, some cases of PCLVO are accompanied by hemiparesis, facial palsy, and dysarthria. It could therefore be argued that a substantial overlap in clinical characteristics exists in both anterior and posterior circulation strokes (14,50). Contrary to the presumption that NIHSS gives more weight to neurological deficits in anterior circulation strokes, our study detected a higher admission NIHSS in PCLVO than in ACLVO (Figure 4). This could be due to more PCLVO patients with reduced consciousness on admission although we do not have data to support this presumption.
Our meta-analysis showed that fewer PCLVO patients are likely to receive IVT in comparison to ACLVO (Figure 5). Previous literature have reported prodromal symptoms in up to 60% cases of PCLVO, which, in most cases, is a reason for misdiagnosis and wrong specialty consultation (17,51). As a result, PCLVO patients may not succeed presenting within the widely accepted 4.5-h time window to receive IVT (52).
Furthermore, the delay in neurological intervention in patients with posterior circulation stroke was reflected in the longer onset-to-IVT and onset-to-groin puncture times in PCLVO (Figures 6, 7). This association is supported by previous studies (53). In spite of the longer onset-to-IVT and onset-to-groin puncture times in PCLVO, a favorable 90 day functional outcome in PCLVO is, however, equally possible just as in ACLVO (Figure 10). This could support the hypothesis that salvageable brain tissue in posterior circulation stroke persists for a longer time as compared to anterior circulation stroke possibly due to a better collateralization in the brainstem (54). Shorter onset-to-IVT and onset-to-groin puncture times could therefore influence a better functional MT outcome in PCLVO. Although onset-to-recanalization time and number of passages were comparable between PCLVO and ACLVO, they tended to be increased in PCLVO patients (Supplementary Figures 7, 8). MT in PCLVO was, however, shown to be associated with longer onset-to-recanalization times in the sensitivity analysis, which was not surprising due to the known delays in hospital admission and intervention of PCLVO patients (14,53).
In our study, we found lower likelihood of sICH in PCLVO (Figure 9). Previous literature has attributed the scarcity of sICH in posterior circulation stroke to relatively smaller infarct volumes and the anatomically smaller nature of vessels that supply the brainstem and cerebellum (55,56). However, this could also be attributed to the lower number of PCLVO patients who receive IVT (Figure 5). Several other studies have shown IVT in posterior circulation stroke to be associated with lower occurrence of sICH than IVT in anterior circulation stroke (57,58). Similar results have been demonstrated by a more recent meta-analysis that indicated a lower likelihood of sICH after IVT in posterior circulation stroke (59). On the other hand, a randomized clinical trial that enrolled 656 patients showed no significant difference in sICH in IVT and non-IVT patients although this study included patients with either ACLVO or PCLVO (60). This raises the question as to whether sICH after MT in PCLVO could therefore be independent of IVT administration. In the anterior circulation, however, MT in addition to IVT has been identified as a significant independent predictor of ICH (56,61).
Studies by previous meta-analyses showed no statistical difference between PCLVO and ACLVO in both recanalization success and 90 day functional outcome (19,20). A comparable likelihood of obtaining successful recanalization (Figure 8) and 90 day favorable functional outcome (Figure 10) was also found in our meta-analysis that included more recent studies. A successful recanalization could therefore influence a good functional outcome irrespective of the circulation involved. Although this study makes a comparison between MT in PCLVO vs. ACLVO, we believe that RCTs are warranted to study if MT in PCLVO is generally efficient.
The incidence of higher mortality in PCLVO has been discussed in numerous previous studies (19,20). Our study likewise provided data to support the claim that MT in PCLVO is associated with a higher mortality as compared to ACLVO ( Figure 11). Although it is believed that younger patients tend to have a better stroke outcome in comparison to older patients (55,56), this study shows that PCLVO patients are younger but yet are still associated with a higher mortality than ACLVO patients.
Due to the higher probability of futile recanalization, physicians may be more conservative and may not attempt MT in older PCLVO patients, hence creating a selection bias with higher numbers of younger patients being considered for MT. It is, however, worth mentioning that this study does not include data on futile recanalization and physicians' patient management.
The higher rate of mortality in PCLVO compared to ACLVO could be partly due to the relatively higher NIHSS on admission and, hence, stroke severity in posterior circulation stroke (Figure 4). This is in line with previous suggestions that stroke severity on admission is an important predictor of stroke outcome, especially in the posterior circulation, and that higher baseline NIHSS in PCLVO is associated with a poor outcome (24,62).
In addition, stroke due to basilar artery occlusion has been described as severe in relation to other occlusion sites in PCLVO (63). We reported basilar artery occlusion as the most frequent site of PCLVO (33.5%) ( Table 1), which may also have contributed to the higher mortality. This study therefore suggests onset-to-IVT and onset-to-groin puncture times, NIHSS, and basilar artery occlusion as factors that could influence outcome in PCLVO. As a reason for high mortality in PCLVO, we propose a subtle progressive-over-time damage or a non-life supporting damage in the posterior circulation, possibly due to the gravity and irreversible nature of the damage to support life despite neurorehabilitation. Such damages have been described as a comatose state or locked-in syndrome, dysphagia, tracheostomy, hypostatic pneumonia, and complications as a result of being long-term bedridden (42). However, this hypothesis cannot be supported by the present study and we therefore encourage further studies to detect causes of higher mortality of MT in PCLVO.

Limitations
Although we implemented measures to limit setbacks in this study, we were nonetheless posed with a couple of challenges. Firstly, such a meta-analysis with several studies over such long duration faces the problem of high heterogeneity with respect to stroke management across studies. Secondly, although efforts were made to exclude the possible effects of the use of firstgeneration MT devices, there was no 100% guarantee that all remaining studies included in our subgroup analyses exclusively used second-generation MT devices. Thirdly, there was a huge disparity in number of PCLVO and ACLVO patients. Finally, the lack of RCT in both groups introduces selection bias.

CONCLUSION
Although MT in PCLVO differs characteristically and also in terms of outcome from ACLVO, our meta-analysis indicates that MT in PCLVO may be equally efficient just as in ACLVO in achieving successful recanalization and a favorable 90 day functional outcome. Although MT in PCLVO is associated with lower likelihood of sICH, possibly due to fewer PCLVO patients receiving IVT because of late recognition and presentation, PCLVO is associated with a higher occurrence of mortality. This higher mortality could be explained through the high baseline NIHSS, longer onset-to-IVT and onset-to-groin puncture times, and basilar artery occlusion being the most predominant site of PCLVO.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
JM and SP conceived and designed the study, undertook data extraction, analyzed the data, and drafted the manuscript. KP and JT independently rechecked all extracted data and analysis. AG-E, YW, KF, and AM helped with data analysis. UZ helped interpret the data. SP had full access to all the data in the study and had final responsibility for the decision to submit for publication. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

FUNDING
We acknowledge support by Open Access Publishing Fund of University of Tübingen.