The data analyzed in this study were obtained from the BASILAR registry (7). The BASILAR registry was a prospective multicenter cohort study that enrolled patients with ABAO from 47 stroke centers across China between January 2014 and May 2019. This study followed the ethical principles of the Helsinki Declaration and was approved by the ethics committees of all participating centers. All patients or their legal representatives provided written informed consent before enrollment. The inclusion criteria of enrolled centers have been published in the BASILAR main study.

The inclusion criteria for this study were as follows: (1) age ≥ 18 years; (2) ABAO confirmed by computed tomographic angiography (CTA), magnetic resonance angiography (MRA), or digital subtraction angiography (DSA); (3) time from symptom onset or last known well within 24 h; (4) ability to provide informed consent. The exclusion criteria were: (1) pre-stroke modified Rankin scale (mRS) score > 2; (2) neuroimaging evidence of intracranial hemorrhage on initial presentation; (3) absence of 90-day follow-up data; (4) current pregnancy or lactation; (5) severe, advanced or terminal illnesses.

The following data were collected for all patients: demographics characteristics, medical history, prodromal symptoms, systolic blood pressure (SBP), stroke severity at admission, pre-stroke modified Rankin Scale score (mRS) score, neuroimaging findings, stroke etiology, intravenous thrombolysis (IVT) administration, workflow (time from symptom onset to imaging and treatment), EVT characteristics, anesthesia type, reperfusion status, and functional outcomes at 90 days and 1 year. Stroke severity of ABAO was dichotomized into extremely severe symptoms (i.e., NIHSS score > 25), moderate to severe symptoms (ie, NIHSS score 10–25) and mild to moderate symptoms (i.e., NIHSS score <10) groups using baseline NIHSS score from the BASILAR study. Stroke etiology was classified according to the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification system. Collateral circulation was evaluated using the American Society of Interventional and Therapeutic Neuroradiology/Society of Interventional Radiology (ASITN/SIR) grading system, with collateral status categorized as poor (grades 0–1), moderate (grade 2), or good (grades 3–4). The extent of ischemic injury was assessed using the posterior circulation-Alberta Stroke Program Early Computed Tomography Score (pc-ASPECTS). Reperfusion status was evaluated on the final angiogram using the modified Treatment in Cerebral Ischemia (mTICI) score, with successful reperfusion defined as mTICI 2b-3.

The primary effectiveness outcome was the distribution of mRS scores at 90 days (ranging from 0 [no symptoms] to 6 [death]). Secondary effectiveness outcomes included the proportion of patients achieving mRS 0–3, mRS 0–2, and mRS 0–1 at 90 days. Safety outcomes included 90-day mortality and symptomatic intracerebral hemorrhage (sICH) within 48 h confirmed by computed tomography (CT) or magnetic resonance imaging (MRI). sICH was defined according to the Heidelberg Bleeding Classification (11).

Normality of continuous variables was assessed using the Kolmogorov-Smirnov test. Normally distributed continuous variables were presented as mean ± standard deviation (SD), while non-normally distributed variables were presented as median [interquartile range (IQR)]. Categorical variables were expressed as frequencies and percentages. Between-group comparisons were performed using Student's t-test for normally distributed continuous variables or Mann–Whitney U test for non-normally distributed variables. Categorical variables were compared using the chi-square test or Fisher's exact test when appropriate.

Logistic regression analyses were performed to assess the association between EVT and clinical outcomes. The shift in mRS score distribution was analyzed using ordinal logistic regression. Binary logistic regression was performed for dichotomous outcomes including mRS 0–3, mRS 0–2, mRS 0–1, sICH, and mortality. Results were reported as odds ratios (OR) with 95% confidence intervals (CIs). Multivariable regression analyses were performed adjusting for covariates selected based on clinical relevance and baseline characteristics. For analyses of patients with extremely severe symptoms (NIHSS score > 25) and moderate to severe symptoms (NIHSS score 10–25), covariates included age, sex, systolic blood pressure (SBP), baseline NIHSS, baseline pc-ASPECTS, smoking history, ASITN/SIR grade, stroke etiology, occlusion site, and IVT. For analyses of patients receiving EVT, we adjusted for age, sex, baseline pc-ASPECTS, atrial fibrillation (AF), coronary heart disease (CHD), ASITN/SIR grade, stroke etiology, anesthesia type, and reperfusion status.

Two separate propensity score matching (PSM) analyses were conducted to minimize selection bias and balance baseline characteristics between treatment groups. For patients with extremely severe symptoms, propensity scores were estimated using multivariable logistic regression with age, SBP, smoking history, baseline pc-ASPECTS, occlusion site, and IVT as covariates. For patients with moderate to severe symptoms, propensity scores incorporated SBP, baseline NIHSS, baseline pc-ASPECTS, and occlusion site as covariates. For both analyses, 1:2 nearest neighbor matching algorithm without replacement was employed to compare outcomes between SMT and EVT groups. Matching was conducted using a caliper width of 0.2 standard deviations of the logit of the propensity score. Detailed information regarding both PSM analyses was provided in Methods S1 and S2, respectively.

Marginal effects plots were constructed to visualize the association between key variables (onset to treatment time and puncture to reperfusion time) and clinical outcomes (mRS 0–3 and mortality), adjusting for the same covariates used in the multivariable logistic regression models. Subgroup analyses were performed to assess the effects of EVT on mRS distribution in specific subgroups. Interaction terms between treatment groups and subgroup indicators were incorporated into the models to evaluate treatment effect heterogeneity across subgroups. Given the minimal missing data for key variables, no imputation methods were employed. A two-tailed P-value <0.05 was considered statistically significant. Statistical analyses were conducted using IBM SPSS Statistics version 27.0 (IBM Corp., Armonk, NY, USA) and R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).

Among 829 patients with ABAO, 436 patients with extremely severe symptoms were included in the analysis (Figure S1). Of these, 342 (78.4%) underwent EVT. The median age was 65 (IQR, 57–74) years, and 317 (72.7%) patients were male. Compared with the SMT group, patients treated with EVT had higher proportions of AF (25.7% vs. 16.0%, P = 0.048) and smoking history (35.7% vs. 14.9%, P < 0.001), lower SBP (150 vs. 160 mmHg, P < 0.001), and higher baseline pc-ASPECTS (8 vs. 7, P = 0.003). EVT patients were less frequently to receive IVT (19.3% vs. 34.0%, P = 0.002). Additionally, significant differences were observed between groups in stroke etiology (P = 0.003) and occlusion sites (P = 0.002). No significant differences were observed in other baseline characteristics between treatment groups. Among EVT patients, the median onset to puncture time was 320 (IQR, 220–487) min, and the median puncture to reperfusion time was 102 (IQR, 70–142) min. General anesthesia was administered in 152 (45.0%) patients, and successful reperfusion was achieved in 267 (78.1%) patients. After PSM, baseline characteristics were well-balanced between the two groups. Other details were presented in Table 1.

The median mRS score was 6 (IQR, 6-6) in the SMT group and 6 (IQR, 4–6) in the EVT group. Patients receiving EVT demonstrated significantly greater odds of favorable shift in mRS distribution compared to those receiving SMT [adjusted OR (aOR) 3.98, 95% CI 2.02–7.84, P < 0.001; Figure 1A and Table 2]. Compared with the SMT group, the EVT group showed significantly higher odds of achieving mRS 0–3 (aOR 4.18, 95% CI 1.28–13.67, P = 0.018) and mRS 0–2 (aOR 6.58, 95% CI 1.39–31.22, P = 0.018). The incidence of sICH within 48 hours was higher in the EVT group (aOR 8.80, 95% CI 1.14–67.70, P = 0.037). However, mortality was significantly lower in the EVT group than in the SMT group (aOR 0.26, 95% CI 0.13–0.53, P < 0.001). Following PSM, this difference in sICH was no longer statistically significant. Other results remained consistent before and after PSM (Figure 1B and Table 2). Long-term follow-up at 1 year demonstrated comparable clinical outcomes between the two groups (Figure S2 and Table S1). Additionally, we compared the effectiveness and safety of EVT vs. SMT among patients with NIHSS score of 10–25, with detailed results presented in Tables S2–S4.

To evaluate the impact of baseline stroke severity on treatment outcomes, we stratified the 647 ABAO patients receiving EVT into three groups based on NIHSS scores: 0–9 (mild to moderate), 10–25 (moderate to severe), and >25 (extremely severe), and compared treatment effectiveness. Baseline characteristics of patients were shown in Table S5. Higher stroke severity was significantly associated with increased rates of AF (P = 0.008) and CHD (P = 0.014), lower baseline pc-ASPECTS (P < 0.001), and higher rates of general anesthesia (P < 0.001). Additionally, significant differences in ASITN/SIR grade distribution were observed across the three severity groups (P < 0.001). No significant differences were observed among the three groups for other baseline characteristics. Compared with the NIHSS 0–9 group, both NIHSS 10–25 (aOR 0.29, 95% CI 0.17–0.50, P < 0.001) and NIHSS > 25 (aOR 0.14, 95% CI 0.08–0.25, P < 0.001) were significantly associated with lower odds of achieving a favorable shift in mRS distribution (Figure 2A and Table S6). Similar trends were observed for outcomes of mRS 0–3, mRS 0–2, and mRS 0–1. Additionally, NIHSS 10–25 (aOR 2.42, 95% CI 1.06–5.49, P = 0.035) and NIHSS > 25 (aOR 5.16, 95% CI 2.30–11.61, P < 0.001) were significantly associated with higher odds of mortality compared with NIHSS 0–9. No significant differences in the odds of sICH within 48 h among the three severity groups. These findings persisted at 1-year follow-up, with similar long-term clinical outcomes observed between groups (Figure 2B and Table S7).

The predicted probabilities of achieving mRS 0–3 and mortality among ABAO patients with extremely severe symptoms (NIHSS score > 25) were estimated based on onset to treatment time, as illustrated in Figure 3. The predicted probability of achieving mRS 0–3 progressively decreased with increasing onset to treatment time, while the predicted probability of mortality correspondingly increased. The EVT group consistently demonstrated a higher predicted probability of achieving mRS 0–3 and a lower predicted probability of mortality compared with the SMT group. However, no interaction was observed between the onset to treatment time and groups for either outcome (P for interaction = 0.169 and 0.487, respectively).

The predicted probabilities of achieving mRS 0–3 and mortality among ABAO patients stratified by NIHSS score in the EVT group (NIHSS score 0–9 vs. 10–25 vs. >25) were estimated based on puncture to reperfusion time, as depicted in Figure S3. The predicted probability of achieving mRS 0–3 progressively decreased with longer puncture to reperfusion time, while the predicted probability of mortality correspondingly increased. Lower baseline stroke severity was associated with higher predicted probability of achieving mRS 0–3 and lower predicted probability of mortality. Additionally, no interaction was found between the puncture to reperfusion time and baseline stroke severity for either outcome (P for interaction = 0.322 and 0.869, respectively).

Subgroup analyses were conducted to explore potential heterogeneity in EVT treatment effects on mRS distribution across different patient populations. No significant heterogeneity was observed across all subgroups stratified by age, sex, baseline NIHSS score, baseline pc-ASPECTS, AF, ASITN/SIR grade, stroke etiology, occlusion site, IVT administration, and onset to treatment time (Figure S4).

However, this study has several limitations. First, the observational study design inherently carries the risk of selection bias, and our findings require validation through RCTs. Second, as the enrolled population was Chinese, the generalizability of our results to other ethnic populations requires further investigation. Finally, our findings necessitate confirmation in larger-scale studies.