Patients with acute BAO who underwent MT with failed recanalization were identified from the Endovascular Treatment Key Technique and Emergency Work Flow Improvement of Acute Ischemic Stroke (ANGEL-ACT) database (ClinicalTrials.gov Identifier: NCT 03370939). The ANGEL-ACT study prospectively recruited consecutive patients with intracranial large vessel occlusion who underwent endovascular treatment (EVT) at 111 comprehensive stroke centers in China over 2 years, and the main article was published previously (18). The enrollment criteria in the present study were as follows: (1) age ≥ 18 years; (2) initial National Institutes of Health Stroke Scale (NIHSS) score ≥ 10; (3) onset-to-puncture time ≤ 24 h; (4) acute posterior circulation stroke; and (5) BAO documented during EVT.

Data of each patient, including the baseline characteristics, risk factor profile, procedural details, and time metrics, were collected using a pre-defined electronic case report form (eCRF) in the electronic data capture (EDC) system. Some original records of the patients, including the laboratory findings and electrocardiograms, were uploaded to the EDC as photographs. Before enrollment, all sites were uniformly trained regarding the use of the EDC system for data entry and electronic signatures. The EDC system data underwent central quality checks by blinded statisticians to control for consistency, accuracy, and completeness. Data queries were sent in cases of inconsistencies or missing data. All imaging data were copied to optical discs and sent to the central core laboratory, including pre-procedure non-enhanced computed tomography (NECT) with computed tomography angiogram (CTA) and/or diffusion-weighted magnetic resonance imaging (MRI) with magnetic resonance angiography (MRA) and digital subtraction angiography (DSA) during EVT, and follow-up computed tomography (CT) or MRI at 24 ± 6 h post-procedure. CT was performed if a patient's neurological status deteriorated abruptly.

Baseline CT/MRI and CTA/MRA, DSA, and post-procedural imaging were evaluated by an imaging core laboratory blinded to the clinical data and outcomes. The pre-procedure posterior circulation Alberta stroke program early CT score (pc-ASPECTS) was evaluated using a baseline non-contrast CT. Successful recanalization was defined as an mTICI score of 2b or 3 on the final catheter angiogram. All neuroimaging data were assessed independently by two neuroradiologists who were blinded to the clinical data; in cases of disagreements, decisions were made by a third experienced neuroradiologist.

EVT was performed under general anesthesia or conscious sedation at all the participating centers. The type of MT device was at the operator's discretion in each center for the study was a retrospective analysis of the data collected from each stroke center. In cases of failed MT with persistent high-grade stenosis (>70% stenosis according to the Warfarin Aspirin Symptomatic Intracranial Disease criteria) (19) and a tendency toward immediate reocclusion, RS was performed at the discretion of the operator. The antiplatelet and anticoagulant strategies during and/or after the procedure were also dependent on the consensus between the operator and the stroke neurologist according to each hospital's protocol. The protocols mainly followed the following principles. A glycoprotein IIb/IIIa inhibitor (tirofiban) was administered according to the recommendations from the manufacturer; it was initiated with a weight-adapted intravenous loading dose of 0.4 μg/kg/min over 30 min followed by a continuous infusion of 0.1 μg/kg/min over 24 h. Dual antiplatelet therapy with 100 mg aspirin and 75 mg clopidogrel was started either orally or via a nasogastric tube 4 h before stopping the infusion of tirofiban.

CT or MRI was performed in all patients within 2 days of EVT as per each center's follow-up protocol. Post-procedural clinical assessment was performed at 3 months via telephone from the study's central laboratory, which included assessments of the modified Rankin Scale (mRS) score, Barthel Index (BI) score, and death due to any cause. Radiographic examinations (DSA/MRA/CTA) were repeated 24 ± 6 h after stenting to evaluate the patency of the stent or target vessel.

The rate of successful recanalization was evaluated in both the RS and non-RS groups. The baseline characteristics, laboratory findings, use of an intravenous tissue plasminogen activator, procedural details, target vessel and stent patency rates at 24 ± 6 h after EVT, rate of good clinical outcomes, the incidence of symptomatic intracranial hemorrhage (sICH), and mortality were compared between the two groups. A favorable outcome was defined as an mRS score of 0–3 at 3 months. A good outcome was defined as an mRS of 0–2 at 3 months. sICH was diagnosed in cases of newly observed intracranial hemorrhage with an increase in NIHSS score by at least 4 points with no other evident causes for this change.

Statistical analyses were performed using SAS v9.0 (SAS Institute, Cary, NC, USA). Categorical variables are expressed as number (frequency), and quantitative variables are presented as median ± interquartile range or mean ± standard deviation. The baseline demographic characteristics, risk factors, imaging characteristics, tool used for MT, and clinical outcomes were compared between the groups using the χ² test or Fisher exact test for categorical and binary data, respectively, and the Mann–Whitney U test or Student's t-test for continuous data. Statistical significance was set at P < 0.05.

Overall, 1,793 patients were enrolled from comprehensive stroke centers across China. Of them, 397 had posterior circulation large vessel occlusion. MT failure was identified in 93 patients with BAO who were eligible for inclusion in the current study, and a flowchart depicting the inclusion of patients is presented in Figure 1.

Of 93 patients with BAO with failed MT (23.4%), 81 were treated with permanent RS. Recanalization with an mTICI score of 2b-3 was achieved in 92.6% (75/81) of patients in the RS group, while 12 patients did not undergo recanalization with stenting. The baseline demographic data, characteristics, procedures, and clinical outcomes are summarized in Table 1. Overall, the mean age was 61.4 ± 12.0 years and 80 (86%) patients were male. A majority of the patients (64/84, 76.2%) had atherosclerosis as the Trial of Org 10172 in Acute Stroke Treatment (TOAST) subtype, while nine patients had missing data regarding TOAST classification. The mean NIHSS and Glasgow Coma Scale (GCS) scores at admission were 23 (13–25) and 7 (5–11), respectively. Sixteen patients (17.2%) received intravenous thrombolysis in conjunction with MT.

Most patients (85/93, 90.4%) were treated using the primary stent retriever technique, while three patients (3.2%) were treated with primary aspiration, and the remaining five patients (5.4%) were treated with both MT devices. The mean number of MT attempts was 2 ± 2, and 82 patients (88.2%) were treated with a glycoprotein IIb/IIIa inhibitor. The median time between the onset of symptoms and final recanalization was 360 min (range, 297–475 min). Complete successful reperfusion (mTICI score: 2b-3) was achieved in 80.6% of patients. sICH was observed in three patients (3.2%).

After 3 months, 44/93 patients (47.3%) had a favorable clinical outcome (mRS score: 0–3). The functional independence rate was 40.9% (BI score ≥ 60). The all-cause mortality rate at follow-up within 90 days was 23.7% (22/93).

There was no significant difference in the risk factors of stroke between the groups at baseline. The two groups also did not demonstrate differences in the initial NIHSS score, GCS, blood pressure, blood glucose level, pc-ASPECT score, or causes of stroke categorized according to the TOAST classification. Patients in the RS group had a significantly higher platelet count on admission than those in the non-RS group (229 vs. 184 g/L, respectively; P = 0.025). The duration between the onset of symptoms and admission (150 vs. 120 min, respectively; P = 0.632) and that between the onset of symptoms and puncture (286 vs. 260 min, respectively; P = 0.697) were longer in the non-RS group than those in the RS group; however, the differences were not statistically significant. The procedure time was significantly longer in the non-RS group than that in the RS group (203 vs. 104 min, respectively; P < 0.001), and the mean number of attempts of MT was higher in the non-RS group than that in the RS group (3 ± 2 vs. 2 ± 1, respectively; P = 0.003). However, there were no differences in the overall first-line modality used for MT between the groups. The rate of use of a glycoprotein IIb/IIIa inhibitor was higher in the RS group than that in the non-RS group (91.4 vs. 66.7%, respectively; P = 0.033).

In the RS group, recanalization with an mTICI score of 2b-3 was achieved in 92.6% (75/81) of patients, and 51.9% (42/81) of patients had a favorable clinical outcome (mRS ≤ 3). Furthermore, of the patients in the RS group with successful recanalization, 52% (39/75) of them had a favorable clinical outcome, which was comparable to the rate of favorable clinical outcomes (52.3%, 90/172) in patients with successful recanalization with MT alone (52 vs. 52.3%, P = 0.962) (Figures 1, 2). In contrast, the proportions of favorable outcomes (mRS ≤ 3) and functional independence (BI score ≥ 60) at 90 days after the procedure were lower in the non-RS group than those in the RS group (16.7 vs. 51.9%, P = 0.023, and 8.3 vs. 45.7%, P = 0.014, respectively). sICH at 24 h was observed in 1.2% of patients in the RS group, which was not significantly different from that (8.3%) in the non-RS group. The non-RS group had a significantly higher rate of all-cause mortality within 90 days of follow-up than the RS group (58.3 vs. 18.5%, respectively; P = 0.006).

Follow-up imaging at 24 h after stenting revealed that target vessel patency rate was higher in the RS group than that in the non-RS group [71.6% (58/81) vs. 41.7% (5/12), respectively; P = 0.038], which was associated with the use of a glycoprotein IIb/IIIa inhibitor (95.2 vs. 73.3%, respectively; P = 0.004) during or after the procedure but not with the use of anticoagulants (64.0 vs. 46.0%, respectively; P = 0.584) (Table 2). The stent patency rate was higher with the use of glycoprotein IIb/IIIa inhibitors than that without the use of glycoprotein IIb/IIIa inhibitors (94.8 vs. 82.6%, respectively; P = 0.096) (Table 3). Neither glycoprotein IIb/IIIa inhibitors nor anticoagulants increased the rate of sICH. Furthermore, compared with patients with occluded target vessels or stents on imaging at 24 h, those with patent vessels or stents had a higher proportion of good clinical outcomes [49.2% (31/63) or 50.0% (29/58)] and a lower rate of mortality [12.7% (8/63) or 13.8% (8/58)], respectively (Figure 3).

Our study has all limitations that come along with a observational study design. Major limitation is the small number of patients without rescue stenting This may be due to each participating hospital had its own protocols regarding referral of patients for RS following the failure of MT for BAO, and the decision regarding performing RS depended on the interventionists and patient conditions. Therefore, selection bias may have affected the results. Second, the present study only enrolled the patients with admission NIHSS ≥ 10, whether the patients with mild stroke (NIHSS <10) on admission could benefit from the RS following MT failure was needed further investigation. Furthermore, our findings should be interpreted cautiously due to regional differences in the prevalence of ICAS, which is more frequent in Asian populations than the European and North American populations.