Analysis of current evidence

The literature search did not identify any RCTs specifically addressing this PICO question, which focused on the comparison between IVT and no IVT. Although BAO was not an exclusion criterion in the pivotal IVT trials,17–19 it is very likely that the number of patients with BAO included in these trials was very small. This is primarily because the majority of patients enrolled in these trials did not undergo vascular imaging. Additionally, BAO accounts for only approximately 1–2% of all AISs and is often associated with a very severe neurological deficit, which was an exclusion criterion in the ECASS trials.19 20 Therefore, the results of the available IVT trials cannot be directly applied to patients with acute BAO.

Our literature search identified three observational studies (all with critical bias, as shown in figure 1) comparing IVT vs no IVT. These studies were included in a meta-analysis. The Basilar Artery International Cooperation Study (BASICS) international prospective registry recruited 592 consecutive patients with acute symptomatic BAO (mean age: 63, median National Institutes of Health Stroke Scale (NIHSS) score: 22) between 2002 and 2007.2 The treatment, which was left to the discretion of each investigator, was heterogeneous and divided into three groups for the main analysis: “antithrombotic therapy only” (antiplatelets or anticoagulation mostly by heparin; n=183), “primary IVT” (n=121) which included subsequent intra-arterial thrombolysis in 41 (33.9%) patients, and “intra-arterial therapy only” (n=179). Functional outcome was assessed at 1 month and the presentation of the results was stratified by clinical severity (severe deficit: coma, locked-in state, tetraplegia; mild-to moderate severity: any other situation). Compared with “antithrombotic therapy only,” patients in the “primary IVT” group tended to have a lower probability of mRS≥4 at 1 month in case of severe deficit (adjusted RR 0.88, 95% CI 0.76 to 1.01) but not in case of mild-to-moderate deficit (adjusted RR 0.94, 95% CI 0.60 to 1.45; p for interaction not provided).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

PICO 1 – Bias evaluation for the observational studies.

The other two identified studies were small, retrospective, and focused on outcome prediction rather than comparison of treatments, which were heterogeneous and left at the discretion of each physician.21 22 In each study, only a minority of patients did not receive endovascular therapy.

All three studies were deemed to have a serious-to-critical level of bias (figure 1), including selection bias (possibly including contraindication to IVT as a reason why IVT was not administered in the control group) and a major risk of confounding (notably confounding by indication).

No formal meta-analysis was conducted due to not only serious but critical limitations of the available studies. The MWG concludes that there is insufficient evidence to provide evidence-based recommendations on this PICO question.

Additional information

In this situation, where the bias of the three included observational studies is mostly critical (figure 1), and results of available RCTs comparing IVT with alteplase to placebo do not directly apply to patients with acute BAO, it must be pointed out that the catastrophic prognosis of untreated BAO was the most important reason for the lack of randomized data for IVT. Consequently, many centers have considered IVT as the standard treatment for this condition for over two decades2 23 24 and it has been considered unethical to randomize patients to a trial comparing IVT with no IVT. In fact, single-arm observational data of consecutive angiography-verified BAO patients (median admission NIHSS 17) showed that up to 50% of patients achieved mRS scores of 0–3 at 3 months regardless of the time window (up to 48 hours) if they presented negligible early ischemic changes in the posterior circulation on non-contrast computed tomography (CT) imaging (posterior circulation Alberta Stroke Program Early CT Score (pc-ASPECTS) ≥8).24 Another analysis of 245 patients (median NIHSS 18) treated with IVT alone (50% <6 hours, 19% 6–12 hours, and 31% >12 hours from last-seen well) reported favorable outcome (mRS 0–3) in 47%,25 which is identical to the EVT arms of recent RCTs. Symptomatic intracranial hemorrhage (sICH) in that study ranged from 7% to 11%, which is in line with the data from the only RCT that used the same sICH criteria.9

In the BASICS registry,2 mRS scores of 0–2 were more frequent in the IVT group compared with the group receiving conventional treatment, with an unadjusted OR 1.83 (95% CI 1.10 to 3.06). The recent ESO guidelines on IVT for AIS recommend IVT with alteplase even in AIS patients with clinically severe symptoms (NIHSS score ≥25) lasting <4.5 hours (strong recommendation, moderate quality of evidence).3 This recommendation highlights that IVT should not be withheld from AIS patients with severe symptoms. Finally, PICO 7 addressed the role of IVT before EVT.

Analysis of current evidence

The literature search identified three RCTs addressing this PICO question. Only one trial recruited patients within 6 hours of estimated symptom onset, while the other two recruited patients within 8 and 12 hours.

Endovascular Therapy for Stroke due to Basilar-Artery Occlusion (BASICS) was a multicenter, international, open-label with blinded outcome assessment RCT of EVT for BAO conducted at 23 centers in seven countries.7 Patients were randomized in a 1:1 ratio within 6 hours of the estimated time of onset to receive EVT (intervention) or BMT (control), which was IVT in 80% of patients.7 At the beginning of recruitment, patients were eligible if they were younger than 85 years of age and had an NIHSS score of 10 or more. After the inclusion of 91 patients, inclusion criteria were expanded to allow recruitment of patients who were 85 years of age or older, those who had an NIHSS score of less than 10, and those who had contraindications to IVT. The primary outcome was a favorable functional outcome, defined as an mRS score of 0 to 3. A total of 300 patients were enrolled (154 in the EVT group and 146 in the BMT group). There was no difference in the proportion of patients with a good outcome (mRS 0–3 at 3 months: 44% EVT vs 38% BMT, RR 1.18, 95% CI 0.92 to 1.50), favorable outcome (mRS 0–2) or distribution of mRS scores. sICH occurred in 4.5% of patients after EVT and in 0.7% of those after BMT (RR, 6.9; 95% CI 0.9 to 53.0).

Endovascular Treatment vs Standard Medical Treatment for Vertebrobasilar Artery Occlusion (BEST) was a multicenter, prospective, open label with blinded outcome assessment RCT of EVT for vertebrobasilar occlusion at 28 centers in China (NCT02441556).⁶ Patients were randomized in a 1:1 ratio within 8 hours of the angiography-confirmed BAO to receive EVT (intervention group) or BMT (control group), which included IVT in only 30% of patients. Patients were eligible if they were 18 years of age or older, had an occlusion of the basilar artery or the distal intracranial vertebral artery with no flow to the basilar artery. The primary outcome was favorable functional outcome defined as an mRS score of 0 to 3 at 3 months. The trial was terminated early after enrolling 131 patients (66 in the EVT group and 65 in the BMT group) because of excessive crossovers and a progressive drop in the rate of recruitment. The median NIHSS at baseline was very high, 32 in the EVT and 26 in the standard arm. There was a substantial rate of crossovers (22.5% from the BMT arm into EVT), and no difference in the proportion of patients with a good outcome (mRS 0–3 at 3 months: 42% EVT vs 32% control, adjusted RR, 1.74, 95% CI 0.81 to 3.74).

Endovascular Treatment for Acute Basilar-Artery Occlusion (ATTENTION) was a multicenter, prospective, open-label RCT of EVT for BAO at 36 centers in China.8 Patients were randomized in a 2:1 ratio within 12 hours (median time from onset to randomization was 5 hours (3.5–7.0)) after the estimated time of onset to receive EVT (intervention) or BMT (control), which was IVT in only every third patient. Patients were eligible if they were at least 18 years of age and had an NIHSS score ≥10. Furthermore, for patients <80 years of age, a pc-ASPECTS of at least 6 was required, whereas for those older than 80, it was at least 8. The estimated time of occlusion occurrence was defined as a sudden onset of BAO symptoms, with no consideration of any preceding minor prodromal symptoms. For patients with unknown time of stroke onset, a 12-hour time window was calculated from the last time the patient was seen well. The primary outcome was good functional outcome defined as an mRS score of 0 to 3 at 3 months. A total of 340 patients were included in the intention-to-treat analysis: 216 and 124 patients were randomized within and beyond 6 hours from symptom onset, respectively. EVT was associated with a higher proportion of patients with good outcomes (mRS 0–3 at 3 months) compared with BMT (46% vs 23%, adjusted rate ratio 2.06 and 95% CI 1.46 to 2.91; P<0.001).

All three trials presented performance bias, as the randomized participants and the treating physicians were aware of the allocated intervention (figure 2A). Furthermore, minor deviations from the intended interventions were noted in two RCTs. In addition, the ATTENTION trial did not clearly report the use of a minimization process to balance the two treatment groups with appropriate stratification, leading to some concerns about randomization bias. In the BEST trial, a high rate of crossover occurred, and the final sample size was only 38% of the planned target of 344 patients, resulting in an underpowered analysis. Furthermore, there may have been a selection bias, as one-third of patients declined trial participation. Regarding indirectness, the BEST trial included patients with very severe symptoms (median NIHSS 32), while the ATTENTION trial included patients with at least 10 NIHSS points. In contrast, the BASICS trial started with patients having NIHSS ≥10, but the inclusion criteria were later modified to include the whole range of NIHSS scores. Furthermore, controls are not directly comparable between the three trials, because the proportion of IVT in BMT and timing of IVT administration differed significantly among the trials. Only the BASICS trial included patients with a time window of 6 hours, whereas in the BEST and ATTENTION trials the time window was 8 and 12 hours, respectively. However, there are remarkable differences in the definition of time windows among the trials.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

A. PICO 2 – Risk of bias for RCTs included in PICO 2. B. PICO 2 – Meta-analysis of randomized-controlled clinical trials: favorable functional outcome (mRS scores of 0–2 at 3 months) in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled adjusted RR, random-effects meta-analysis, P=0.12). C. PICO 2 – Meta-analysis of randomized-controlled clinical trials: favorable functional outcome (mRS scores of 0–2 at 3 months) in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone, and stratified by high vs low proportion of IVT-treated patients in the BMT arm (pooled adjusted RR, random-effects meta-analysis, P=0.003 for interaction). The BEST trial was excluded from this interaction analysis due to its extremely high rate of crossovers (22.5%) from EVT into BMT arm. D. PICO 2 – Meta-analysis of randomized-controlled clinical trials: Good functional outcome (mRS scores of 0–3 at 3 moths) in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled adjusted RR, random-effects meta-analysis, P=0.04). E. PICO 2 – Meta-analysis of randomized-controlled clinical trials: Good functional outcome (mRS scores of 0–3 at 3 months) in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone, and stratified by high vs low proportion of IVT-treated patients in the BMT arm (pooled adjusted RR, random-effects meta-analysis, P=0.02 for interaction). The BEST trial was excluded from this interaction analysis due to its extremely high rate of crossovers (22.5%) from EVT into BMT arm. F. PICO 2 – Meta-analysis of randomized-controlled clinical trials: Distribution of mRS scores at 3 months (shift analysis) in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled adjusted RR, random-effects meta-analysis, P=0.03). G. PICO 2 – Meta-analysis of randomized-controlled clinical trials: Mortality at 90 days in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled adjusted RR, random-effects meta-analysis, P=0.01). H. PICO 2 – Meta-analysis of randomized-controlled clinical trials: Symptomatic ICH in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled adjusted RR, random-effects meta-analysis, P=0.003). I. PICO 2 – Risk of bias for registry studies. J. PICO 2 – Meta-analysis of registry studies: Good functional outcome (mRS scores 0–3 at 3 months) in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone, and stratified by high vs low proportion of IVT-treated patients in the BMT arm (pooled adjusted RR, random-effects meta-analysis, P=0.0001 for interaction). K. PICO 2 – Forest plot showing differential effect of reperfusion therapy stratified by high vs low proportion of IVT-treated patients in the BMT arm (P=0.03 for interaction), including data from randomized-controlled clinical trials (RCTs) and one registry study. Distribution of mRS scores at 3 months (shift analysis) in patients with acute ischemic stroke presenting within 6 hours from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (Cochran’s Q-test for interaction testing).

We conducted a random-effects meta-analysis of studies that reported outcomes deemed critical and important. Furthermore, for functional outcomes, we performed additional analyses to test for interactions among RCTs with high vs low percentages of IVT in the BMT arm of a study (figure 2B-H). The BEST trial was excluded from this interaction analysis due to its extremely high rate of crossovers (22.5%) from the EVT into the BMT arm.6 The ATTENTION investigators listed in the limitation section that initially, patients had to pay for the thrombolytic drug, which may have contributed to the low use of thrombolytics.8 9 We identified several significant interactions (see table 1), further supported by the fact that no difference between EVT and BMT was observed in the BASICS trial,7 while in the ATTENTION trial,8 no superiority of EVT was observed in the analysis when BMT included 100% IVT (adjusted rate ratios 1.57 (95% CI: 0.97 to 2.54)). Frequencies of sICH were significantly higher in the EVT arms.

Additional information

The literature search identified three registry-based non-randomized studies addressing this PICO question, the bias of which is described in figure 2I and in PICO 3.

The Endovascular Treatment for Acute Basilar Artery Occlusion Study (BASILAR) registry was a nationwide prospective registry of consecutive patients presenting with an acute, symptomatic, radiologically confirmed BAO at 47 comprehensive stroke centers across 15 provinces in China between January 2014 and May 2019.26 Patients with BAO within 24 hours of estimated symptom onset were divided into groups receiving BMT plus EVT (n=647) or BMT alone (n=182), of whom 463 and 127 were treated within 6 hours from symptom onset, respectively. The rate of IVT in the whole cohort was 20%. The primary clinical outcome was the improvement in mRS scores at 3 months across the two treatment groups assessed as a common OR using ordinal logistic regression shift analysis, adjusted for prespecified prognostic factors. The secondary efficacy clinical outcome was good functional status, defined as mRS scores of 0 to 3 at 3 months. However, the only reported outcome for the 6 hour time window is distribution of mRS at 3 months (common OR).

The BASICS registry2 was a prospective, international (Europe, South America, North America, and Australia), observational registry of consecutive patients who presented with an acute symptomatic and radiologically confirmed BAO between November 1, 2002, and October 1, 2007. The primary clinical outcome was assessed at 1 month and defined as mRS scores of 4 to 6. Patients presenting within 24 hours from symptom onset were divided into three groups according to the treatment they received: antithrombotic treatment only (AT), which comprised antiplatelet drugs or systemic anticoagulation; primary intravenous thrombolysis (IVT), including subsequent intra-arterial thrombolysis; or intra-arterial therapy (IAT), which comprised intra-arterial thrombolysis, mechanical thrombectomy, stenting, or a combination of these approaches. Of the 592 patients who were analysed, 183 were treated with only AT, 121 with IVT, and 288 with IAT. A total of 84, 99, and 186 within 6 hours, respectively. The patient-level outcome data (unadjusted mRS 0–3) for the 6 hour time window are available only for IVT and IAT subgroups.

The ATTENTION registry27 is an ongoing prospective, multicenter registry in China. The sample comprised 2134 patients within 24 hours of estimated time of acute BAO recruited at 48 comprehensive stroke centers between March 2017 and February 2021. 462 patients received BMT (less than 20% IVT) and 1672 underwent EVT plus BMT. The median time from estimated time of BAO to treatment was 419 minutes (IQR: 273–682), but the number of patients treated with BMT as well as the combination of EVT with BMT within 6 hours from symptom onset was unavailable in the relevant publication. BMT consisted of IVT, antiplatelets, anticoagulants or combinations. Endovascular approach consisted of mechanical thrombectomy, thromboaspiration, stenting, IA thrombolysis or combination. The primary clinical outcome was a favorable functional outcome, defined as mRS scores of 0 to 3 at 3 months. The outcome data were reported as RR, and the number of the patients in the subgroups was not reported. All other studies reported either raw data or ORs.

The registry study by Räty et al., compared 122 of IVT-only vs EVT±IVT treated BAO patients.25 The primary outcome was mRS 0–3 and the data were analysed with conventional and doubly robust inverse probability-weighted regression analysis. The primary outcome was more frequent in the IVT only group compared with EVT±IVT. In that study, about 60% of patients had delays of less than 6 hours.

Differential effect of reperfusion therapy stratified by high vs low proportion of IVT-treated patients in the BMT arm is outlined in figure 2J and K.

Table 1 provides details regarding the assessment of the quality of evidence for all outcomes evaluated in PICO 2. To better understand the differential effect of reperfusion therapy stratified by the composition of BMT, please see also PICO 3 and the discussion.

Analysis of current evidence

The literature search identified two published RCTs8 9 addressing this PICO question.

The ATTENTION trial was otherwise described in PICO 2, however, we want to point out that only one patient received IVT in the time window of more than 6 hours from estimated time of BAO to imaging. The Basilar Artery Occlusion Chinese Endovascular (BAOCHE) trial, a multicenter Chinese prospective RCT, aimed to assess the effect and safety of EVT in conjunction with BMT compared with BMT alone. The trial enrolled patients with AIS due to BAO and an absence of large baseline infarct on neuroimaging who underwent randomization in 6 to 24 hours after symptom onset.9 Symptoms onset was defined as a time point when symptoms started or, if unknown, as time when patients were last seen well. Isolated vertigo was not considered onset of symptoms. Treatment start was defined as time of groin puncture. The original primary outcome, an mRS score of 0 to 4 at 3 months, was subsequently changed to a good functional status (mRS-scores of 0 to 3).

Assessment of the risk of bias is presented in figure 3A

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

A PICO 3 – Risk of bias in randomized-controlled clinical trials. B. PICO 3 – Meta-analysis of randomized-controlled clinical trials (RCTs): Good functional outcome (mRS scores 0–3 at 3 months) in patients with acute ischemic stroke presenting within 6–24 hours from time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled adjusted RR, random-effects meta-analysis). C. PICO 3 – Risk of bias in observational studies. D. PICO 3 – Meta-analysis of observational studies: Good functional outcome (mRS scores 0–3 at 3 months, except for the BASICS registry: 1 month) in patients with acute ischemic stroke presenting within 6–24 hours from time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled RR, random-effects meta-analysis). E. PICO 3 – Sensitivity analysis of observational studies after inclusion of the studies that presented raw data regarding good functional outcome (mRS scores 0–3 at 3 months, except for the BASICS registry: 1 month) in patients with acute ischemic stroke presenting within 6–24 hours from time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (RR, random-effects meta-analysis). F. PICO 3 – Forest plot showing differential effect of reperfusion therapy stratified by geographical regions including RCTs and observational studies: Good functional outcome (mRS scores 0–3 at 3 months, except for the BASICS registry: 1 month) in patients with acute ischemic stroke presenting within 6–24 hours from time last known well treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (P-value for interaction <0.0001, Cochran’s Q-test for interaction testing).

Both trials8 9 presented performance bias, as randomized participants and treating physicians were aware of the allocated intervention. Furthermore, minor deviations from the intended interventions were noted in both RCTs. In addition, the ATTENTION trial8 did not clearly report the use of a minimization process to balance the two treatment groups with appropriate stratification, leading to some concerns about randomization bias. Finally, the BAOCHE trial9 presented minor concerns due to missing outcome data. The overall risk of bias was high for both the ATTENTION8 and BAOCHE9 trials.

Data regarding patients presenting within 6–24 hours from time last known well were available in one of the trials only as adjusted RRs with corresponding 95% CIs, without presenting the raw data. For that reason, we used a generic inverse variance meta-analysis to provide a pooled overall effect (figure 3B). Compared with patients randomized to BMT, the pooled adjusted RR for a good functional outcome in patients randomized to EVT was 1.90 (95% CI: 1.41 to 2.57; P<0.01; I²: 0%; figure 3B).

Additional information

The literature search identified four registry-based observational studies addressing this PICO question.2 26–28 The ATTENTION registry27 was described in PICO 2. Qualifying patients had to present within 24 hours of estimated symptom onset. The number of patients treated with BMT as well as the combination of EVT with BMT beyond 6 hours from symptom onset was unavailable in the relevant publication. The BASILAR registry,26 a nationwide prospective registry, was described in PICO 2. A total of 184 and 55 patients were treated with BMT plus EVT and BMT alone beyond 6 hours from symptom onset, respectively. The BASICS registry2 was described in PICO 2. A total of 99, 21 and 102 patients received AT, IVT and IAT beyond 6 hours, respectively.

A registry presented by Gruber et al., 28 was a mandatory prospective stroke inpatient quality assurance registry covering the entire federal state of Hessen in Germany. Gruber et al., analysed the clinical course and short-term outcomes of patients with radiologically confirmed acute BAO dichotomized by BMT plus EVT (n=270) or BMT alone (n=133). This registry also included patients presenting beyond 24 hours from symptom onset (n=26) and with unknown time from symptom onset (n=58). The primary clinical outcome was good functional status, defined as mRS score of 0 to 3 at 3 months. A total of 46 and 30 patients were treated with BMT plus EVT and BMT alone between 6 and 24 hours from symptom onset, respectively.

The registry study by Räty et al., 25 was described in PICO 2. It compared 122 of IVT-only vs EVT±IVT treated BAO patients and included about 40% of patients with delays of more than 6 hours.

The MWG assessment of the risk of bias in the included observational studies for PICO 3 was performed according to the Cochrane ROBINS-I tool10 and is presented in figure 3C.

All four studies presented moderate confounding bias, since there were several significant baseline differences between the different treatment groups. The ATTENTION registry,27 the BASILAR registry,26 the registry presented by Gruber et al.,28 and by Räty et al.,25 were based on data derived from centers of specific countries (ie, China in the first two studies, Germany in the third, and Finland in the last), thus moderate selection bias may occur. No significant misclassification, deviation from intervention, or missing bias occurred in any of the included observational studies. Assessment by blinded, certified investigators was reported to have been performed only in the BASILAR registry, while in the other three studies no clear description of the assessment was presented. The BASICS registry² did not predefine sICH as an outcome measure, and the follow-up period was restricted to only 1 month, rendering the study vulnerable to serious reporting bias. Finally, the study of Gruber et al., 28 presents moderate reporting bias since sICH was not assessed or reported as a safety outcome.

We conducted a study-level, random-effects meta-analysis of the four observational studies included in PICO 3 for the outcome mRS score of 0–3 at 3 months. However, it should be noted that the ATTENTION registry reported only the adjusted RR for the patients presenting within 6–24 hours from time last -known well and achieving mRS 0–3 at 3 months, without providing raw data. Therefore, we were not able to calculate the unadjusted RR for this study. We used the generic inverse variance meta-analysis to provide a pooled overall effect, but we also presented two subgroups stratifying by the adjusted vs unadjusted RR. Patients treated with EVT had a similar likelihood of achieving mRS 0–3 at 3 months compared with patients treated with BMT (figure 3D).

A sensitivity analysis was performed by including only the four studies that presented raw data, and similar results were obtained (figure 3E).

Further, we present forest plot showing differential effect of reperfusion therapy stratified by geographical region, in which the patients were randomized (Asian vs European/International) (figure 3F). In line with the findings presented in PICO 2, we found a significant interaction (P<0.00001) between the two regions. In the Asian studies, EVT led to better outcomes compared with BMT, whereas the opposite trend was observed in the European/International studies. There are several plausible explanations for this heterogeneity, including differences in systems of care and ethnicity-related issues.

The BAOCHE and ATTENTION investigators listed in the limitation section that initially, patients had to pay for the thrombolytic drug, which may have contributed to the low use of thrombolytics.8 9 Notably, in both the ATTENTION and BAOCHE trials, no superiority of EVT was observed in analysis when BMT included 100% IVT (adjusted rate ratios 1.57 (95% CI: 0.97 to 2.54) and 1.74 (95% CI: 0.36 to 8.4), respectively).8 9

It is not known how standard treatment differs among various centers worldwide for patients who underwent EVT compared with those who have not received any reperfusion therapy at all (as was the case in most of the patients in Asian trials, who received merely secondary prevention). It is possible that the latter group was not admitted to intensive or intermediate care units. Regarding ethnicity-related issues, the high prevalence of ICAD in the Asian population was mentioned as a reason why the results of the BAOCHE and ATTENTION trials may not be generalizable to Western countries.8 9 Finally, the ATTENTION investigators acknowledged that their results are not generalizable to patients with an NIHSS score of less than 10.8 9

Table 2 provides details regarding the assessment of the quality of evidence for all outcomes evaluated in PICO 3 both using randomized and observational data.

Analysis of current evidence

The aim of this PICO question was to investigate the presence or absence of a difference in treatment effect (interaction/effect modification) based on a specific presentation (ie, severity of neurological symptoms and/or occlusion location) at baseline. To address this question, we focused on reperfusion therapy studies that provide subgroup analyses stratified by a specific baseline situation. For the comparison of EVT (±IVT) vs no EVT, the literature search identified four RCTs and three registries that reported outcomes at 3 months.2 6 8 9 26 27 29

One observational study, which reported outcomes only at 1 month2, is described in additional information section.

EVT vs no EVT depending on initial stroke severity

The four identified RCTs, BEST, BASICS, ATTENTION, and BAOCHE, have all been described in PICO questions 2 and 3. All trials reported subgroup analyses stratified by baseline NIHSS score, but the stratification cut-off differed substantially across the trials. Some of the NIHSS cut-off values are of lesser clinical relevance (29 in BEST and 20 in BAOCHE and in ATTENTION). In the BEST trial, there was no evidence of a differential effect (p for interaction=0.79) of EVT vs no EVT on mRS 0–3 at 90 days in patients with NIHSS score ≤29 (OR 1.56; 95% CI 0.60 to 4.10) and >29 (OR 1.91; 95% CI 0.61 to 6.00). In the ATTENTION trial, the adjusted RR for the association between EVT and mRS 0–3 at 3 months were 1.51 (1.05–2.18) and 3.53 (1.71–7.29) in patients with NIHSS score 10–19 and ≥20, respectively. No P-value for interaction was reported. No data exist for less than 10 NIHSS points, because inclusion criteria in ATTENTION was 10 or higher. In the BASICS trial, the RR for the association between EVT and mRS 0–3 at 3 months in patients with NIHSS score <10, 10–19 and ≥20 were 0.85 (0.62–1.16), 1.55 (1.06–2.27), and 1.28 (0.67–2.46), respectively. No P-value for interaction was reported in the original publication, however, it was presented by Dr W. Schonewille during ESO conferences in 2020 and 2023 (ESOC 2020 and ESOC 2023): P-value for interaction was 0.02 and the conclusion was that EVT is not better than BMT in patients with BAO and less than 10 NIHSS points. We also performed a post-hoc interaction test, based on the data from the original publication of the BASICS trial and found very similar P-values for the interaction. Of note, BASICS was the only trial with a high proportion (~80%) of IVT in the BMT arm. In the BAOCHE trial, the magnitude of the treatment effect on mRS 0–3 seemed similar in patients with NIHSS score 6–20 (adjusted RR 1.80 (1.21–2.67)) and >20 (adjusted RR 1.83 (0.73–4.58)). No P-value for interaction was reported in the original publication. However, very recent meta-analysis of the BASICS and BAOCHE trials30 reported outcomes of patients with BAO and NIHSS <10. In this subgroup analysis of 78 patients, frequencies of favorable (mRS 0–3) or excellent (mRS 0–2) clinical outcome between the EVT and the BMT groups were comparable. favorable functional outcome (mRS 0–3) at 3 months was achieved in 26 of 37 patients (70.3%) in the EVT group and in 30 of 41 patients (73.2%) in the BMT group. Excellent clinical outcome (mRS 0–2) occurred in 22 of 37 patients (59.5%) in the EVT group, and 24 of 41 patients (58.5%) in the BMT group. The rate of sICH in patients with NIHSS <10 was 8.1% in the EVT group, whereas no sICH occurred in the BMT group. The mortality rate in the EVT group was 18.9% (7 of 37 patients) and 17.1% (7 of 41) in the BMT group. P-value for the interaction for the primary outcome (mRS 0–3) was 0.04. Hence, in BAO patients with less than 10 NIHSS points, EVT is not superior to BMT and is less safe. The interaction (P-value) in subgroup analysis stratified by 10 NIHSS points was slightly different between the aforementioned meta-analysis BASICS and BAOCHE (P-value for interaction 0.04) compared with data from the BASICS trial alone (P-value for interaction 0.02). This difference may be explained by different proportion of IVT in the BMT arm of BASICS compared with BAOCHE (80% vs 22%).

The BASILAR registry study was described in PICO 2 and 3. Only 20% of the patients received IVT (with alteplase or urokinase). Otherwise, BMT included antiplatelet drugs, systematic anticoagulation, or a combination of these treatments, at the discretion of the treating physician. Subgroup analyses according to a NIHSS cut-off of 26 points did not suggest a modification of treatment effect by baseline NIHSS score (adjusted common ORs for lower mRS scores at 90 days: 2.2, 95% CI: 1.3 to 3.6 in the NIHSS 0–26 subgroup; 3.3, 95% CI: 1.7 to 6.5 in the NIHSS >26 subgroup; P for interaction=0.52). Again, the selection of the cut-off value (NIHSS 26) is of lesser clinical relevance.

Between 2014 and 2016, 167 patients (median age: 75 (66-82); median NIHSS score: 24 (10-30)) were enrolled in the prospective multicenter RESCUE Japan Registry 2 study within 24 hours of symptomatic BAO.29 The treatment applied was decided by the attending physician (EVT group, n=129, 77.2% or BMT group, n=38, 22.8%), and the analysis was stratified by baseline NIHSS score cut-off of 10 points. Proportion of patients who achieved mRS ≤3 score at 3 months (primary outcome) after EVT compared with BMT (including IVT in about 24%) was 54% vs 12% (P<0.01) in the severe subgroup (NIHSS score 10–40), and 72% vs 86% (P=0.43) in the mild subgroup (NIHSS score 0–9). No P-value for interaction or adjusted analyses were provided in the original publication, however, we have computed P-value of 0.004 for this interaction.

The ATTENTION registry27 was described in PICO 2. The proportion of patients who achieved an mRS ≤3 score at 3 months (primary outcome) after EVT compared with BMT (including IVT in about 20%) was 36.8% vs 23.4% (adjusted relative risk 1.58 (95% CI: 1.30 to 1.91)) in the severe subgroup (NIHSS score at least 10), and 58.7% vs 51.4% (adjusted relative risk 1.05 (95% CI: 0.80 to 1.38)) in the mild subgroup (NIHSS score 0–9). Significant interaction was observed (P<0.001).

Evaluation of bias for the four RCTs is visualized in PICO 2 and 3, whereas bias for the three observational studies is in figure 4A.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

A. PICO 4 – Risk of bias in observational studies. B. PICO 4 – Meta-analysis of randomized-controlled clinical trials (RCTs) stratified by clinical severity at baseline (P-value for interaction 0.03): Good functional outcome (mRS scores of 0–3 at 3 months) in patients with acute ischemic stroke presenting within 6 hours (BASICS), within 12 hours (ATTENTION), or within 6 to 24 hours (BAOCHE) from time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled RR, random-effects meta-analysis, Cochran’s Q-test for interaction testing). Footnote: Only a minor proportion of patients randomized to ATTENTION and BAOCHE received IVT as part of the BMT. C. PICO 4 – Meta-analysis of randomized-controlled clinical trials (RCTs): Symptomatic intracranial hemorrhage in patients with acute ischemic stroke presenting with<10 NIHSS, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (RR, random-effects meta-analysis). D. PICO 4 – Forest plot showing differential effect of reperfusion therapy stratified by NIHSS cut-off 10, including data from randomized-controlled clinical trials (RCTs) and registry studies. Good functional outcome (mRS scores of 0–3 at 3 months) in patients with acute ischemic stroke presenting within 6 hours (BASICS), within 12 hours (ATTENTION), within 6 to 24 hours (BAOCHE), or 24 hours (RESCUE Japan Registry 2, ATTENTION registry) from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (P-value for interaction 0.0004, Cochran’s Q-test for interaction testing). Footnote: Only a minor proportion of patients randomized to ATTENTION and BAOCHE received IVT as part of the BMT. E. PICO 4 – Meta-analysis of randomized-controlled clinical trials (RCTs) stratified by occlusion location at baseline (p-value for interaction 0.01): Good functional outcome (mRS scores of 0–3 at 3 months) in patients with acute ischemic stroke presenting within 6 hours (BASICS), within 12 hours (ATTENTION), or within 6 to 24 hours (BAOCHE) from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (pooled RR, random-effects meta-analysis, Cochran’s Q-test for interaction testing). Footnote: Only a minor proportion of patients randomized to ATTENTION and BAOCHE received IVT as part of the BMT. F. PICO 4 – Forest plot showing differential effect of reperfusion therapy stratified by NIHSS cut-off 10, including data from randomized-controlled clinical trials (RCTs) and registry studies. Good functional outcome (mRS scores of 0–3 at 3 months in all except BASICS prospective registry, where it was assessed at 1 month) in patients with acute ischemic stroke presenting within 6 hours (BASICS), within 12 hours (ATTENTION), within 6 to 24 hours (BAOCHE), or 24 hours (RESCUE Japan Registry 2), or no time limit (BASICS prospective registry) from the time last known well, treated with endovascular treatment plus best medical treatment (BMT) vs BMT alone (p-value for interaction<0.00001, Cochran’s Q-test for interaction testing). Footnote: Only a minor proportion of patients randomized to ATTENTION and BAOCHE received IVT as part of the BMT. G. PICO 4 – Interaction testing for treatment effect between EVT±IVT and no EVT (100% IVT) in patients with GCS 3–7 and 8–15.

In line with the above-mentioned recent meta-analysis from RCTs using NIHSS cut-off 1030, we performed a random-effects meta-analysis of randomized data stratified by the same baseline NIHSS cut-off value (figure 4B,C). Of note, all patients randomized into the ATTENTION trial had baseline NIHSS ≥10, whereas the BEST trial (median NIHSS of randomized patients of 32 and 26 for EVT+BMT vs BMT arms, respectively) did not provide results for this NIHSS cut-off. This analysis demonstrated a differential effect of reperfusion therapy stratified by NIHSS cut-off 10 (P=0.03 for interaction). Similar interactions were detected also in non-randomized registry studies: RESCUE JAPAN LIMIT (P=0.01) and ATTENTION (P=0.02). For the purpose of visual demonstration, we created forest plots showing differential effect of reperfusion therapy stratified by NIHSS cut-off 10 including both randomized and non-randomized data (figure 4D). Because clinical severity in patients with BAO is strongly related to the location of the occlusion, we also analyzed whether there is a differential effect between EVT and BMT as stratified by occlusion location (proximal, middle, distal) (figure 4E).

Table 3 provides details regarding the assessment of the quality of evidence for mRS score of 0–3 at 3 months in PICO 4.

Additional information

Schonewille et al., 2 reported data from a prospective BAO registry stratified by stroke severity on admission (mild-to-moderate vs severe). Severe symptoms were described as coma, locked-in state, or tetraplegia, whereas all other symptoms were considered mild-to-moderate. The registry had three arms (antithrombotics, primary IVT, and IAT. The IAT group comprised intra-arterial thrombolysis, mechanical thrombectomy, stenting, or a combination of these approaches. The outcome was assessed only at 1 month and not at 3 months as in all other studies. In addition, another major difference compared with other studies is that the primary IVT group included also subsequent IAT. For these two reasons, we only considered IAT vs no IAT (conventional, antithrombotics) comparison. For the purpose of these guidelines, we considered that “mild-to-moderate” stroke severity corresponded to patients with an NIHSS <10, whereas “severe” symptoms corresponded to patients with NIHSS ≥10. We created forest plots showing the differential effect of reperfusion therapy stratified by NIHSS cut-off 10 including both randomized and non-randomized data (figure 4F). The P-value for interaction was<0.00001.

Ritvonen et al., 31 reported similar frequencies of outcomes based on the severity of the initial Glasgow Coma Scale (GCS): the 3 month mRS 0–3 in comatose (GCS <8) and non-comatose (GCS 8–15) patients treated with EVT (±IVT) vs BMT (100% IVT) was 16.7% vs 22.2%, respectively, and the P-value for interaction was 0.70 (figure 4G).

A very large US study analyzed data from the National Inpatient Sample (2018–2020), which included 5795 patients with less than 10 NIHSS points at baseline. Of those, 880 (15.4%) underwent EVT. The effect of EVT was compared with BMT. The primary outcome was discharge to home or self-care, adjusted for robust outcome predictors. A secondary analysis was performed with the same adjustments and evaluated the length of stay. After adjustments, in multivariable regression, EVT was reported to be associated with increased odds of discharge to home (OR 1.95 (95% CI 1.31 to 2.90); P=0.001) and a decreased length of hospital stay (B, −0.74 (95% CI −1.36 to −0.11); P=0.02) compared with BMT. However, on January 9, 2024, an eLetter was published by the Stroke Editorial office32 stating that after publication, an error was discovered. Specifically, the variables for EVT and IVT were switched, and the article was retracted.

Finally, in addition to the above-mentioned interactions for the treatment effect of EVT vs no EVT stratified by baseline stroke severity, we have noticed that the direction of the forest plots comparing EVT vs BMT largely depends on the composition of the BMT group. In case it comprises mostly conventional therapy (aspirin, anticoagulation), the forest plot favored EVT, however, when BMT was IVT in the majority of the patients, there was no difference between the two arms.

IVT vs no IVT depending on initial stroke severity

We did not identify any RCTs or subgroup data within such studies addressing the relationship between initial stroke severity and the effect of IVT on outcomes at 3 months in BAO-patients. However, given the effectiveness of IVT regardless of initial stroke severity shown in RCTs on IVT in disabling stroke,18 as well as evidence of its benefit in both the anterior and posterior circulation,33 34 it is highly likely that IVT has a beneficial effect on patients with BAO, regardless of their initial stroke severity. This is further supported by the findings of Ritvonen et al.,31 where no significant difference was found between IVT alone and EVT±IVT in patients stratified by a GCS score of 8 (figure 4G).

Imaging of acute tissue ischemia in BAO

The extent of ischemia in BAO is most typically described by early ischemic changes (EIC) on neuroimaging using the pc-ASPECTS score. This score was originally based on CT-angiography source images35 but is also applicable to non-contrast CT or magnetic resonance imaging (MRI) based diffusion-weighted imaging (DWI).36 Lower pc-ASPECTS scores indicate more extensive EIC. Interpretation of pc-ASPECTS on CT can be difficult due to beam hardening artifacts at the level of the temporal bones/skull base. Other less commonly used scores include the Pons-Midbrain Index (PMI) on non-contrast-CT,37 Pons-Midbrain and Thalamus (PMT) score on DWI-MRI,38 and the Critical Area Perfusion Score (CAPS) on CT-perfusion.39 These studies indicate that the extent of ischemic changes seen on acute neuroimaging remains a strong prognostic factor even after successful reperfusion with EVT.

Analysis of current evidence

This PICO question focuses on the treatment effect of acute recanalization therapy in patients with high vs low pc-ASPECTS points. Patients with low scores may have less or no viable tissue that could benefit from such therapy. PICO questions 2 to 4 describe the evidence of the effect of recanalization treatments for BAO based on time and stroke severity. For the current PICO question, we investigated whether there is an interaction between reperfusion treatment effects in patients with high vs low pc-ASPECTS in RCTs.

While there are no randomized data regarding solely the effect of IVT based on pc-ASPECTS, but the literature search identified three potentially relevant RCTs (EVT plus BMT vs BMT) that have already been described in detail in PICO questions 2 and 3. The subgroup (interaction) analyses in these three trials used different cut-offs of pc-ASPECTS, being nine in the BAOCHE9 and 8 in the BASICS7 and ATTENTION8 trials (all showing no difference). Very importantly, the median pc-ASPECTS scores of the randomized patients were rather high. In the BASICS trial, only 17% of the patients had pc-ASPECTS lower than 8 at baseline, whereas median pc-ASPECTS at 24 hours based on angiography source imaging was 9 (8-10) in the EVT+BMT group and 9 (7-10) in the BMT group. Similarly, in the ATTENTION trial, only 20% of the patients had pc-ASPECTS lower than 8 at baseline (median 9 (8-10) in the EVT+BMT group and 10 (8-10) in the BMT group). In the BAOCHE trial, patients had baseline pc-ASPECTS median of 8 (7-10) in both arms.

Hence, the proportion of patients with low pc-ASPECTS scores was insufficient to perform a formal meta-analysis and draw conclusions about the interaction of the treatment effect in patients with high vs low pc-ASPECTS. Furthermore, for two of three critical outcomes (mRS 0–2 and mortality at 3 months) data from only one trial (BAOCHE) were available, and for mRS 0–3 only from two trials (BASICS and ATTENTION).

Additional information

Numerous studies have shown a strong association between poor outcomes and lower pc-ASPECTS in BAO patients, regardless of recanalization treatment.24 35 40–44 In one of these studies, patients receiving recanalization therapy (IVT or EVT) had 1 year mortality of 38% in those with pc-ASPECTS 8–10, whereas it was 66% for pc-ASPECTS <8. In another study, patients receiving recanalization therapy (IVT or EVT), 3 month mortality was 31% in those with pc-ASPECTS 8–10, whereas it was 64% for pc-ASPECTS <8. In the same study, mRS 4–6 was observed in 46% and 88%, respectively. A very recent Korean study suggested some potential benefit of EVT in patients with low pc-ASPECTS45 based on the inverse probability of treatment weighting model for mRS score of 0–3 (33% vs 24%, P=0.03), but not based on propensity-score matching for the same outcome. For mRS score of 0–2, no difference was observed in any of the models.

Analysis of current evidence

The literature search did not identify any published RCTs addressing this PICO question, but identified one post-hoc analysis46 derived from a Chinese registry-based observational study.

The BASILAR registry has been described in PICO questions 2 and 3. Patients with evaluated Basilar Artery on CT Angiography (BATMAN) score were included in the analysis (n=828).46 The primary efficacy outcome was good functional status, defined as mRS scores of 0 to 3 at 3 months. The secondary efficacy outcomes included functional independence defined as mRS score of 0 to 2 at 3 months, and successful reperfusion.

In all three categories of the BATMAN score (0–3, 4–6, and 7–10), EVT+BMT was associated with higher odds in achieving better outcomes and lower mortality compared with BMT (approximately 80% conventional treatment with antiaggregatory or anticoagulation). P-value for interaction was 0.52.

The study presented moderate confounding bias (figure 5), since there were several significant baseline differences between the different treatment groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

PICO 6 - Risk of bias in an observational study.

Thus, the only study relevant to this PICO question evaluated the effect of collateral flow. No other advanced imaging criteria were found to be tested.

Analysis of current evidence

The literature search identified no randomized trials and three prospective cohort studies as relevant for this PICO.

Nie et al., 47 reported post-EVT outcomes in patients with and without concurrent IVT in a prospective multicenter RESCUE-RE cohort study accompanied by a meta-analysis of the existing literature. The RESCUE-RE study enrolled patients with AIS due to vertebrobasilar occlusion that were 18 years or older, had a pre-stroke mRS score of 0 to 2 and were followed up for 3 months. IVT, if indicated, was administered within 4.5 hours from symptom onset (0.9 mg alteplase/ kg). Between July 2018 to October 2020, 1701 patients were enrolled in the registry, of which 321 patients were included in the study.

Singer et al., 48 reported post-EVT outcomes in a retrospective multicenter cohort study, ENDOSTROKE. This study enrolled both prospectively and retrospectively patients with any large vessel occlusion in the anterior or posterior circulation, who were 18 years of age or older and in whom EVT was attempted. The study included a subgroup of 148 patients with attempted EVT for BAO in whom 3 months follow-up data were available. Concurrent IVT was permitted in their study (not stated to how many it was administered), however, patients experiencing thrombolysis-related recanalization before EVT were excluded. The primary outcome was mRS score of 0–2 at 3 months. The main angiographic outcome was recanalization defined as a final thrombolysis in cerebral infarction (TICI) score of 2b or greater.

Siow et al., 49 reported results from a retrospective multicenter cohort study. Patients were included if they underwent EVT for acute BAO and had a pre-stroke mRS score of 0–2. Between January 2015 and December 2019, 322 patients who met the inclusion criteria were included in the study. Patients received IVT (0.9 mg/kg alteplase) if they had no contraindications and could be treated within 4.5 hours of symptom onset. The primary outcome was mRS score of 0–3 at 3 months.

Nappini et al., 50 reported results of a secondary analysis from a national prospective registry of EVT. Patients were included if they underwent EVT for BAO, either with or without IVT with tissue plasminogen activator (time window of 4.5 hours from symptom onset). The outcomes were recanalization status, and different dichotomizations of the 90-day mRS. Between 2011 and 2017, 464 who underwent EVT for BAO were included in the registry. Overall, patients treated with EVT alone had less favorable baseline characteristics, including higher NIHSS scores and higher prevalence of baseline co-morbidities and anticoagulant treatment. Clinical outcomes were better in patients receiving bridging IVT in the unadjusted analysis, but this did not hold true after adjusting for confounding variables. In a post-hoc subgroup analysis in patients treated with EVT within 6 hours from symptom onset, patients receiving bridging IVT had a reduced risk of death and a shift towards better 90-day mRS in the adjusted analysis.

Singh Kohli et al., 51 report a small single-center series of 31 BAO patients undergoing EVT, 22 of which underwent direct EVT while nine received bridging IVT. Baseline characteristics and time to treatment were generally more favorable in the patients who received bridging IVT (time window of 4.5 hours from symptom onset). Unadjusted clinical and technical outcomes were more favorable in the bridging IVT group; however, the small group size did not permit adjusted analysis.

Risk of bias assessment for the included non-randomized studies (figure 6A) showed serious risk of bias for all included studies.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

A. PICO 7 - Risk of bias for the non-randomized trials included in PICO 7. B. PICO 7 - Meta-analysis of observational studies: Good functional outcome (mRS scores 0–3 at 90 days) in adults with acute ischemic stroke due to BAO, treated with direct endovascular thrombectomy vs intravenous thrombolysis and endovascular thrombectomy (pooled OR, random-effects meta-analysis). C. PICO 7 - Meta-analysis of observational studies: Good functional outcome (mRS scores of 0–2 at 3 months) in adults with acute ischemic stroke due to BAO, treated with direct endovascular thrombectomy vs intravenous thrombolysis and endovascular thrombectomy (pooled OR, random-effects meta-analysis). D. PICO 7 - Meta-analysis of observational studies: Good functional outcome (shift mRS scores of at 3 months) in adults with acute ischemic stroke due to BAO, treated with direct endovascular thrombectomy vs intravenous thrombolysis and endovascular thrombectomy (pooled adjusted OR, random-effects meta-analysis). E. PICO 7 - Meta-analysis of observational studies: Symptomatic intracranial hemorrhage post treatment in adults with acute ischemic stroke due to BAO, treated with direct endovascular thrombectomy vs intravenous thrombolysis and endovascular thrombectomy (pooled adjusted OR, random-effects meta-analysis). F. PICO 7 - Meta-analysis of observational studies: favorable recanalization (mTICI 2b/3 post treatment) in adults with acute ischemic stroke due to BAO, treated with intravenous thrombolysis and endovascular thrombectomy vs direct endovascular thrombectomy (pooled adjusted OR, random-effects meta-analysis).

We conducted several meta-analyses to provide a quantitative synthesis of the results (figures B-F), and we state in the figure if the available estimates were adjusted for potential confounders. Briefly, point estimates of critical outcomes (all mRS-related outcomes) were in favor of combined IVT and EVT treatment. Statistically significant differences were found for shift mRS and adjusted mRS score of 0–2 at 3 months. For sICH and modified Thrombolysis in Cerebral Ischemia (mTICI), no difference was found. For mortality at 90 days, only data from one study were available, hence, no meta-analysis was conducted. The adjusted ORs for this outcome with combined treatment compared with direct EVT was 1.79 (0.87–3.70).

Table 4 provides details regarding the assessment of the quality of evidence in PICO 7.

Additional information

In the anterior circulation, non-inferiority of direct EVT could not be proven in a patient-level meta-analysis of all anterior circulation randomized direct-to-EVT trials.52 Of note, an RCT of tenecteplase before EVT compared with EVT alone is ongoing in patients with BAO (POST-ETERNAL).

Analysis of current evidence

Stent retriever thrombectomy was the preferred technique in pivotal trials demonstrating benefits of mechanical thrombectomy plus BMT over BMT alone in the acute anterior circulation strokes.4 Based on the expert opinion in the latest ESO-ESMINT guideline for Mechanical Thrombectomy in Acute ischemic Stroke,5 A Direct Aspiration First Pass Technique (ADAPT) may be used as a standard first-line treatment, followed by stent retriever thrombectomy as a rescue therapy if needed.

The literature search did not identify any completed RCTs comparing the different first-line treatment techniques in patients with BAO. For the comparison of the first-line contact aspiration and stent-retriever thrombectomy, the literature search identified one post-hoc analysis of an RCT,53 seven registry-based observational studies,54–60 and four single-center retrospective observational studies.61–64

In the post-hoc analysis of the BASICS trial by Knapen et al.,53 127 patients with BAO who underwent EVT with either direct aspiration (n=60) or stent retriever thrombectomy (n=67) as the first-line approach were included. The primary outcome was mRS score of 0–3 at 3 months. Secondary outcomes included mRS score at 3 months, procedure duration, mortality at 3 months, and sICH.

The retrospective analysis of two stroke registries by Abdelrady et al., 57 investigated the influence of the frontline endovascular technique in 128 patients with BAO between January 2015 and December 2019. Of those 128, 33 were treated with contact aspiration, 35 with stent-retriever thrombectomy, 35 underwent combined technique (contact aspiration+stent-retriever), and in 25 patients the technique was switched. The outcomes included first pass mTICI three reperfusion, mTICI 2b-3, and mTICI 2 c-3, as well as favorable clinical outcome (mRS score 0–2 at 3 months). The authors also reported frequency of sICH.

The STAR registry58 was a prospective, multicenter registry in the United States and Germany, recruiting patients between June 2014 and December 2018. Of 3045 patients, 345 presenting with posterior circulation stroke and treated with mechanical thrombectomy using modern devices were included in the analysis comparing different techniques (contact aspiration, stent-retriever, combined approach). Of the 345 patients, 121 were treated with contact aspirations, 90 patients with stent-retriever thrombectomy, and the rest with combined approach. The outcome measures included successful recanalization mTICI 2b-3, clinical outcome (mRS score 0–2 at 3 months) and frequency of sICH.

In the study by Baik et al.,59 161 patients from two university hospital stroke registries with acute BAO referred for mechanical thrombectomy between March 2013 and December 2019 were enrolled, out of which 43 underwent contact aspiration and 118 stent-retriever thrombectomy. The authors reported mTICI 2b-3, mTICI 3, clinical outcome mRS score of 0–2 at 3 months, mortality at 3 months, and frequency of sICH, all outcomes stratified according to the angiographic characteristics of the occlusion.

The MR CLEAN Registry60 was a nationwide prospective registry of consecutive patients who underwent EVT in the Netherlands between March 2014 and December 2018. Some 205 patients with intracranial proximal occlusion in the posterior circulation (basilar artery, intracranial part of the vertebral artery, and posterior cerebral artery), who underwent EVT with contact aspiration (n=71) or stent retriever thrombectomy (n=134) as the first-line approach were analyzed. Outcome measures included mRS score (0–2 and 0–3 at 3 months) and final eTICI reperfusion grade. Mortality and frequency of sICH was also reported.

A post-hoc analysis from the Endovascular Treatment in ischemic Stroke (ETIS) registry by Gory et al., 54 included 100 patients presenting with BAO between March 2010 and October 2016 at three comprehensive stroke centers. Forty-six patients underwent first-line contact aspiration and 54 first-line stent-retriever thrombectomy. The reported outcome measures included mTICI 2b-3, mTICI 3, mRS score of 0–2 at 3 months, 3 month mortality, and sICH.

The Tama-REgistry of Acute Thrombectomy (TREAT) was a regionwide, multicenter, retrospective observational registry in Japan. The post-hoc analysis by Kaneko et al., 55 comprised of 48 patients with acute BAO who underwent EVT between January 2015 and December 2017, out of which 12 patients underwent first-line contact aspiration and 33 first-line stent-retriever thrombectomy. The primary outcomes were functional outcomes (mRS scores of 0–2 and 0–3) and all-cause mortality at 3 months.

The Endovascular thrombectomy for acute basilar artery occlusion (ENTHUSE) trial was a retrospective, multicenter, observational study, conducted at three high-volume stroke centers in South Korea.56 The post-hoc analysis comprised of 212 patients with acute BAO who underwent EVT between January 2011 and August 2017, out of which 67 underwent first line contact aspiration and 145 first-line stent-retriever thrombectomy. The reported outcome measures included mTICI 2b-3, mTICI 3, mRS score 0–2 at 3 months, and 3 month mortality.

A single center retrospective study by Choi et al., 63 included 50 patients with acute BAO treated with contact aspiration (n=34) or stent-retriever thrombectomy (n=16) between March 2016 to December 2019. The reported outcome measures included successful reperfusion mTICI 2b-3, mRS score of 0–2 at 3 months, 3 month mortality, and sICH.

A single-center retrospective study by Lee et al., 62 included 38 patients with 40 vertebrobasilar occlusions, that were treated with contact aspiration (n=11) or stent-retriever thrombectomy (n=29) between March 2010 to December 2017. The reported outcome measures included mTICI 2b-3 and mRS score of 0–2 at 3 months.

A single-center study by Sangpetngam et al., 64 retrospectively analyzed 66 patients with vertebrobasilar occlusions treated with EVT (the authors reported 9 patients with vertebral artery occlusion among 61 patients with successful reperfusion). Thirty-two patients were treated with first-line contact aspiration and 34 patients with first-line stent-retriever thrombectomy. The reported outcomes included mTICI 2b-3, and mRS score of 0–2.

A single-center study by Son et al., 61 retrospectively analyzed 31 patients with acute BAO treated with EVT between March 2010 to December 2013. Eighteen patients were treated with first-line contact aspiration and 13 patients with first-line stent-retriever thrombectomy. The reported outcomes included mTICI 2b-3, mTICI 3, and mRS score of 0–2.

The Posterior Circulation Ischemic Stroke Evaluation: Analysing Radiographic and Intraprocedural Predictors for Mechanical Thrombectomy (PC-SEARCH Thrombectomy) registry65 was a multicenter retrospective collaboration from eight high-volume centers in the United States consisting of consecutive patients with BAO treated with EVT between January 2015 and December 2021. Out of 383 patients included in the retrospective analysis, 219 underwent first-line contact aspiration and 164 received first-line stent-retriever thrombectomy. The reported outcome measures included mTICI 2b-3, mRS scores of 0–2 and 0–3 at 3 months, and rate of sICH.

Risk of bias for the included studies is presented in figure 7A.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

PICO 8- Risk of bias of the studies. B. PICO 8 - Meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Good functional outcome (mRS scores of 0–3 at 3 months) in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). C. PICO 8 - Meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): favorable functional outcome (mRS scores of 0–2 at 3 months) in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). D. PICO 8 - Meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Successful recanalization (mTICI 2B-3) in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). E. PICO 8 - Meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Symptomatic ICH in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). F. PICO 8 - Meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Mortality at 90 days in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). G. PICO 8 - Sensitivity meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Good functional outcome (mRS scores of 0–3 at 3 months) in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). H. PICO 8 - Sensitivity meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): favorable functional outcome (mRS scores of 0–2 at 3 months) in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). I. PICO 8 - Sensitivity meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Successful recanalization (mTICI 2B-3) in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). J. PICO 8 - Sensitivity meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Symptomatic intracranial hemorrhage in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis). K. PICO 8 - Sensitivity meta-analysis of observational studies (except for *post-hoc analysis of the BASICS RCT): Mortality at 90 days in adults with acute ischemic stroke due to acute BAO, treated with EVT using direct aspiration vs stent retriever as the first-line strategy (pooled OR, random-effects meta-analysis).

We performed several random-effects meta-analyses comparing the two techniques of interest (figure 7B–F).

Sensitivity analyses (after excluding studies comprising all posterior-circulation strokes) of critical and important outcomes are depicted in figure 7H–K.

Table 5 provides details regarding the assessment of the quality of evidence for critical and important outcomes evaluated in PICO 8.

Additional information

We also identified four observational studies66–69 that reported data on the endovascular technique used in the posterior circulation stroke thrombectomy. However, the authors of the above-mentioned studies reported results for stent-retriever thrombectomy alone and combined (simultaneous) contact aspiration plus stent-retriever thrombectomy. Based on the consensus of the MWG, these studies were excluded from the meta-analysis as the combined approach was considered as a separate endovascular technique. Data from these four studies listed below favor direct aspiration as the first-line strategy.

The Registro Endovascolare Lombardo Occlusione Basilar Artery (RELOBA) study group included 102 patients with acute BAO treated endovascularly in 12 centers in the region of Lombardy (Italy) between January 2010 and December 2015.66 Successful reperfusion TICI 2b-3 was achieved in 20/27 (74.1%) patients treated with contact aspiration and in 47/65 (72.3%) patients with stent-retriever thrombectomy (alone or combined).

A study by Li et al., 67 was a single-center retrospective study of 68 patients with acute BAO who underwent EVT between January 2014 and December 2016. The primary outcome, mRS score of 0–2 at 3 months, was achieved in 5/7 (71.4%) patients treated with contact aspiration and in 20/50 (40.0%) patients treated with stent-retriever thrombectomy (including 47 patients treated with stent-retriever alone and 3 patients treated with combined technique).

A retrospective analysis of prospectively collected data by Monteiro et al., 68 comprised of 83 patients with acute BAO between January 2013 to December 2020. Twenty-three patients were treated with contact aspiration, 20 patients with stent-retriever alone, and 40 patients with combined technique. The reported outcomes included successful reperfusion TICI 2b-3, first pass TICI 2 c-3 and mRS score of 0–2.

The CICAT was a prospective registry including all stroke patients in Catalonia from January 2016 to January 2020. The post-hoc analysis by Terceno et al., 69 included 298 patients with posterior circulation stroke (out of which 216 patients had BAO). The data on endovascular technique were available in 261/298 patients. The mRS score of 0–2 in 3 months was achieved in 27/62 (43.5%) patients treated with contact aspiration, in 32/108 (29.6%) treated with stent-retriever alone, and in 33/91 (36.3%) with a combined technique.

A study by Gerber et al., 70 reported recanalization according to the Arterial Occlusive Lesion (AOL) scale instead of mTICI. AOL 2–3 was achieved in 9/13 (69%) stent retriever patients, whereas it was 17/20 (85%) in the aspiration arm. In order to maintain consistency in the reported outcome (mTICI vs AOL), this study was excluded from the meta-analysis for reperfusion outcomes.

Analysis of current evidence

The literature search identified no RCTs addressing this PICO question. As ICAD is often diagnosed after EVT rather than before, RCTs are unlikely to be performed. We identified one observational study conducted in China that addressed this PICO in a subgroup analysis of patients with ICAD.71 The proportion of mRS score of 0–3 was 33% in EVT alone (40% in successfully recanalized, 15.9% in non-recanalized), compared with 26.8% in EVT plus rescue treatment (P=0.004). The 90-day mortality differed little between the groups; 46.4% in EVT alone (34.9% in successfully recanalized, 79.5% in non-recanalized), compared with 47.7% in EVT plus rescue treatment. Hence, among patients in whom EVT was not successful, those who underwent rescue percutaneous transluminal angioplasty (PTA) and/or stenting had better clinical outcomes, lower mortality, and lower sICH, although non-significant rates than those in whom no rescue PTA and/or stenting was performed. In the EVT arms of recent BAOCHE and ATTENTION RCTs, angioplasty/stenting was performed in 39.8–54.5%. Both trials recruited Chinese patients having a high prevalence of ICAD, and EVT alone vs EVT plus rescue treatment in ICAD patients was not addressed in either study.8 9 Furthermore, in a subgroup analysis of the ATTENTION trial, patients with underlying ICAD as the cause of stroke, did not show a clear benefit from EVT compared with BMT (OR 1.59, 95% CI 0.91 to 2.68).8

Bias of the aforementioned observational study is illustrated in figure 8. No meta-analysis was performed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

PICO 9 - Bias evaluation of the observational studies.

Additional information

ICAD is a disease of major intracranial arteries with different manifestations, ranging from subtle arterial wall thickening to severe stenosis with vulnerable atherosclerotic plaques.72 Depending on the study, the basilar artery is the most common or second most common affected intracranial vessel.72 73 ICAD prevalence shows marked racial/ethnical differences. In the Northern Manhattan Stroke Study, a prospective registry study of 714 patients, ICAD was the presumed cause of stroke in 9% of Caucasian patients, 15% of Hispanic, and 17% of African-American patients.74 ICAD is responsible for 10–48% of all large-vessel occlusion (LVO) strokes; it is particularly common in Asia but even in Europe, up to 1 in 10 LVO strokes are caused by ICAD.75 76 In the Trevo endovascular registry, which included mainly European patients, ICAD accounted for only 10% of all EVT cases of BAO,77 while in the Chinese ATTENTION and BEST trials, atherosclerosis was the underlying stroke etiology in 44%–56% of cases.6 8

Studies comparing EVT in patients with BAO due to ICAD vs other stroke mechanisms found nominally higher numbers of rescue PTA and/or stenting in patients with underlying ICAD,77–79 although proportions differed significantly only in one study.80 Despite these rescue treatments, EVT in BAO due to underlying ICAD was in most studies associated with poorer outcomes, longer procedure times and in some studies, less successful reperfusion compared with other stroke mechanisms,77 79 81 whereas one study found no difference in outcomes between BAO caused by ICAD compared with non-ICAD.80 ICAD-related occlusions are prone to re-occlude, occurring in up to 40% of patients.82 While the apposition thrombus that has formed adjacent to the atherosclerotic plaque can be removed by EVT alone, new thrombus may form at the thrombogenic plaque surface, thereby leading to re-occlusion This risk may be even higher after an endovascular attempt, as the traumatic fibrous cap disruption and vessel wall trauma caused by endovascular devices increase thrombogenicity even further. PTA with or without stenting can eliminate or reduce the stenosis caused by the atherosclerotic lesion, and in theory, stenting may reduce the risk of re-occlusion by covering the thrombogenic lesion. On the other hand, PTA/stenting may cause perforator occlusions by pushing plaque fragments into small perforator orifices, requiring dual antiplatelet therapy, which increases the risk of hemorrhage, particularly in cases with concomitant IVT.83

Two studies specifically assessed rescue therapy in failed EVT for BAO, but were not confided to patients with underlying ICAD, although ICAD patients accounted for the majority that underwent rescue treatment (77.3%–88.5%), with the comparator being all patients with successful or failed EVT in one study,84 and only failed EVT in the other.85 If we put aside successful recanalization in non-ICAD patients after EVT alone, those who achieved recanalization after rescue therapy had better prognosis than those not recanalized at all. Of note, compared with failed EVT without rescue therapy, the rate of sICH was lower in the EVT plus rescue therapy group in one study (14.2% compared with 4.2%, P=0.002),84 while the other study reported small numbers of events (one case of sICH in each group) without significant difference.85

Another approach in case of severe underlying basilar artery stenosis after EVT is use of antithrombotic agents such as Glycoprotein (GP) IIb/IIIa inhibitors. One study compared this treatment to angioplasty with or without stenting in 55 patients and found no difference in sICH, mortality, or functional outcome between the two strategies.56

Analysis of current evidence

The literature search did not identify any published RCTs addressing the PICO question, but eight non-randomized studies were identified: six observational registry-based studies,86–91 one non-randomized trial,92 and one study combining data from a prospective registry and an open-label, single-arm trial.93 Seven studies86–92 compared add-on tirofiban, whereas one study eptifibatide93 to no add-on antithrombotic medication for patients undergoing EVT±IVT. Studies that included solely BΑO or dominant vertebral artery occlusion patients will be described in this section, whereas reports from studies with a subgroup of BAO patients or secondary analysis from posterior circulation studies (with uncertain proportion of BAO patients) will be presented in additional information below.

The study by Chen et al., 88 compared patients treated with EVT for BAO based on whether they did (n=363) or did not (n=282) receive add-on tirofiban. IVT was administered for 17.1% and 20.2%, whereas IAT for 8.0% and 18.8% of the patients, respectively. The cohort was drawn from the Chinese, nationwide, prospective BASILAR registry comprising consecutive adult patients with BAO within 24 hours of symptom onset between January 2014 and May 2019. Patients with pre-stroke mRS≥3 were excluded. Tirofiban was administered intravenously 0.4 µg/kg/min for 30 minutes followed by 0.1 µg/kg/minutes for up to 24 hours. The choice of tirofiban use was left at the discretion of the treating physician but was recommended under conditions with an increased risk of re-occlusion or distal embolization, such as stenting, angioplasty, a high number of passes, or atherosclerotic etiology. The primary efficacy outcome was the mRS score at 90 days. Safety events according to IVT-treatment status are not reported. However, the authors speculated that the higher mortality and sICH in patients not receiving tirofiban were due to higher frequency of previous anticoagulation, IVT and IAT (even though the last two were included as covariates in the adjusted analyses).

The study by Sun et al., 86 was a single-center, retrospective, observational study from China on consecutive 18‒80-year-old patients with atherosclerotic BAO who underwent EVT within 24 hours of symptom onset between January 2012 and July 2018. Patients with pre-stroke mRS>1, NIHSS <10 or > 35 (or 0 in the item 1A), significant cerebellar mass effect, bilateral extended brain stem ischemia, or embolic occlusion were excluded. The treatment groups received either tirofiban (0.3–0.4 mg within 6–8 minutes intra-arterial (IA) and 0.15 µg/kg/minutes intravenous (IV) for 24 hours) followed by dual antiplatelet therapy (n=74) or immediate dual antiplatelet therapy (n=31). Tirofiban was used based on the treating physician’s decision in cases with emergency stenting or balloon angioplasty, local new thrombosis or vascular dissection, and severe atherosclerotic lesions with a high risk of re-occlusion. In the tirofiban group, 24.3% received IVT and 20.3% IAT, whereas the rates were 6.5% and 32.3% in the no-tirofiban group. The primary outcomes were 90-day functional independence (mRS 0–2) and favorable functional outcome (mRS 0–3).

Yang et al., 90 included consecutive adult acute stroke patients with major large artery occlusion undergoing EVT between June 2015 and December 2017 from the Chinese, multicenter, prospective ANGEL registry. The posterior circulation occlusion subgroup (n=158/662) consisted of basilar and dominant vertebral occlusions treated within 24 hours of symptom onset, excluding patients with NIHSS <6 and pre-stroke mRS>1. Add-on tirofiban (0.25–1 mg IA, followed by 0.1 µg/kg/min IV for 24 hours) was considered for patients with emergency stenting or angioplasty, presumed endothelial damage, instant re-occlusion, or severe in situ atherosclerosis with a high risk of early re-occlusion (n=74), whereas the rest did not have add-on tirofiban (n=84). Bridging IVT was used in 23.9% of the tirofiban group and 35.2% of the no-tirofiban group in the whole cohort but the numbers could not be extracted solely for the posterior circulation occlusion subgroup. The primary efficacy endpoints were functional independence (mRS 0–2) and mortality at 90 days, and the primary safety endpoint was sICH at 24-hour imaging control.

Additional information

A study by Pan et al., 87 was a prospective registry study from two Chinese centers comparing tirofiban (n=64) vs no tirofiban (n=66) as an adjunctive therapy of EVT for patients with vertebral or BAO between October 2016 and July 2021. Tirofiban was administered 0.25–1 mg IA, followed by 0.1–0.15 µg/kg/minute IV for 16–24 hours at the discretion of the treating physician for patients with severe residual stenosis (≥ 50%) after thrombectomy, rescue treatment with stenting or angioplasty, ≥3 passes, or severe atherosclerosis with a high risk of re-occlusion. IVT was received by 25.0% in the tirofiban and 39.4% in the no-tirofiban group. The outcomes were 90-day mRS score of 0–2, NIHSS at discharge, in-hospital and 90-day mortality, frequency of sICH, and successful recanalization (TICI ≥2 b).

A study by Kellert et al.,89 was a prospective registry study from Germany on consecutive AIS patients treated with EVT between 2006 and 2011. In the posterior circulation occlusion subgroup, 20 patients received tirofiban IV for at least 12 hours according to weight and kidney function (recommended if stenting was performed or endothelial injury was feared) and 14 did not. The IVT rate was 65.0% in the former and 78.5% in the latter group. Outcomes included excellent (mRS 0–1) and good (mRS 0–2) functional outcome at 90 days, sICH rate, mortality, and successful recanalization (TICI ≥2 b).

Zhao et al., 91 compared patients undergoing EVT who did (n=37 with posterior circulation occlusions) or did not (n=25 with posterior circulation occlusions) receive add-on tirofiban between January 2013 and February 2017 from a Chinese, single-center, prospective registry. Only patients for whom second-generation stent retrievers were used were included. Tirofiban dosing was 0.25–0.5 mg IA, followed by 0.2–0.25 mg/h for 12–24 hours. Typical indications for tirofiban at the interventionists’ discretion were emergency stenting or angioplasty, successful recanalization by three or more passes, and severe atherosclerosis lesions with high possibility of re-occlusion. In the tirofiban group, 11% received IVT and 24% IAT, whereas the respective numbers were 4% and 19% in the no-tirofiban group. The primary outcome was sICH, and the secondary outcomes included 90-day and long-term functional outcome, mortality, early re-occlusion, and successful recanalization.

Wu et al., 92 reported results from a Chinese, non-randomized, single-arm trial with an original plan to give tirofiban to all adult EVT patients within 2 years. However, the trial was stopped after 1 year due to safety concerns (ICH), so during the second year no patients received tirofiban. Thus, the patients treated within the first (n=23/94 with posterior circulation occlusions) and the second year (n=17/124 with posterior circulation occlusions) were compared. The patients with EVT after 24 hours from symptom onset or ICH were excluded. Contrary to other studies, tirofiban was administered only as IA boluses with doses depending on the bleeding risk (maximum dose 10 μ/kg). The IVT and IAT rates were not reported for the posterior circulation stroke patients separately but were 16.0% and 4.4% in the whole cohort of tirofiban-treated patients and 30.1% and 4.2% among the patients who did not receive tirofiban. The presence of sICH was the primary outcome complemented by other hemorrhagic outcomes, 90-day functional outcomes, and mortality.

Finally, the study by Ma et al., 93 was the only one to investigate add-on eptifibatide vs no eptifibatide in patients treated with endovascular approach within 24 hours of onset for large-vessel occlusion. The study derived the intervention arm from the Chinese, multicenter, open-label, single-arm EPOCH trial (April 2019 to March 2020) and the control arm from the Chinese, multicenter, prospective ANGEL-ACT registry (November 2017 to March 2019). The former included only patients with mechanical thrombectomy, whereas the latter allowed patients with any EVT including sole IAT. The posterior circulation subgroup comprised 46/162 patients in the propensity score matched cohort, 23 in each treatment arm. Eptifibatide was delivered as 135–180 µg/kg in 5 minute IV/IA, followed by 0.75–2 µg/kg/minute IV for 24 hours. The IVT rate was 25.9% in each treatment arm of the propensity score matched cohort but was not reported for posterior circulation occlusion patients separately. The primary efficacy outcome was 90-day good outcome, defined as mRS score of 0–2, and propensity score matching was used for analyses.

We excluded one retrospective registry study on tirofiban vs no tirofiban for patients with vertebrobasilar occlusion (86% BAO) treated with endovascular approach within 24 hours of onset94 due to inconsistent reporting of the results. The authors were contacted several times for clarification, but they did not respond to the request.

The risk of bias is outlined in figure 9A Severity of the risk of confounding bias ranged from moderate to critical. The most common concern appearing in all observational studies was that the add-on antithrombotic agent was chosen based on periprocedural factors that differed systematically between the treatment groups, such as the number of passes or instant re-occlusion, in-situ thrombosis, or residual stenosis requiring emergency angioplasty or stenting.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9

A. PICO 10 - Risk of bias of the studies included. B. PICO 10 - Metanalysis of observational studies comparing add-on antithrombotic treatment vs no add-on antithrombotic medication stratified by studies with only basilar or dominant vertebral artery occlusion vs studies, where basilar artery occlusion was a subgroup of patients: mRS score of 0–2 at 3 months (pooled OR, random-effects meta-analysis, Cochran’s Q-test for interaction testing).* unadjusted studies. C. PICO 10 - Metanalysis of observational studies comparing add-on antithrombotic treatment vs no add-on antithrombotic medication stratified by studies with only basilar or dominant vertebral artery occlusion vs studies, where basilar artery occlusion was a subgroup of patients: Mortality (pooled OR, random-effects meta-analysis, Cochran’s Q-test for interaction testing).* unadjusted studies. D. PICO 10 - Metanalysis of observational studies comparing add-on antithrombotic treatment vs no add-on antithrombotic medication stratified by studies with only basilar or dominant vertebral artery occlusion vs studies, where basilar artery occlusion was a subgroup of patients: sICH (pooled OR, random-effects meta-analysis, Cochran’s Q-test for interaction testing).* unadjusted studies. E. PICO 10 - Metanalysis of observational studies comparing add-on antithrombotic treatment vs no add-on antithrombotic medication stratified by studies with only basilar or dominant vertebral artery occlusion vs studies, where basilar artery occlusion was a subgroup of patients: recanalization TICI 2B-3 (pooled OR, random-effects meta-analysis, Cochran’s Q-test for interaction testing).* unadjusted studies.

We performed a meta-analysis stratified by the proportion of BAO patients within the studies: a) studies including solely patients with BAO or BAO plus dominant vertebral artery occlusion; and b) studies with a subgroup of BAO patients or uncertain proportion among other posterior circulation strokes (figure 9B-E).

For both critical outcomes (mortality and sICH) and one important outcome (mTICI 2B/3), the analyses favored add-on antithrombotic treatment in studies including solely patients with BAO or BAO plus dominant vertebral artery occlusion, whereas no difference was noticed if we included studies, where BAO patients were only a part of posterior circulation strokes. However, it should be noted that the significant findings are mainly based on the study by Chen et al., in which no-tirofiban group had a very poor outcome (mortality 52%, sICH 10%). The authors discussed the reliability of their findings and speculated if this was due to higher frequency of previous anticoagulation, IVT, and IAT (even though the last two and cardioembolic etiology were included as covariates in the adjusted analyses).

Table 6 provides details regarding the assessment of the quality of evidence for PICO 10.