Endovascular thrombectomy (EVT) has become the first-line treatment for patients with acute ischemic stroke caused by large vessel occlusion (AIS-LVO) (1), offering significant improvements in clinical outcomes. While EVT's efficacy is well-established (2), optimizing procedural variables, including anesthesia type, remains a critical area of investigation. Current evidence on the impact of anesthesia type—general anesthesia (GA) (3) vs. local anesthesia (LA) or conscious sedation (CS)—on functional outcomes is conflicting. Individual-level meta-analysis of the HERMES collaboration demonstrated worse outcomes with GA (3), whereas data from randomized trials, including the GOLIATH, SIESTA, and ANSTROKE studies, suggested improved or comparable outcomes with GA (4, 5). Most recently, the AMETIS trial reported similar results, further complicating the narrative.

Robust pial collaterals have been identified as a critical imaging biomarker for improved EVT outcomes (6), contributing to slower infarct progression and enhanced hemodynamic stability in ischemic areas. Patients with good collateral status may benefit from preserved cerebral blood flow, even in the setting of ischemia. Anesthesia choice can modulate these dynamics: LA/CS, being less invasive, is less likely to cause procedural hypotension, which could impair collateral circulation (7, 8). Conversely, GA provides airway protection and facilitates procedural control but may introduce hemodynamic instability. Thus, understanding how collateral status interacts with anesthesia type to influence outcomes is crucial for tailoring treatment strategies.

We hypothesize that the effect of anesthesia type on EVT outcomes is modulated by collateral status, as reflected by the hypoperfusion intensity ratio (HIR). This ratio, derived from perfusion imaging, reflects the extent of ischemic compromise and could serve as a surrogate marker to guide anesthesia selection. In this study, we aimed to investigate the relationship between HIR and functional outcomes after EVT and determine whether HIR could help identify patients who would benefit from a specific anesthesia approach.

A total of 948 patients with anterior circulation occlusion underwent endovascular thrombectomy (EVT) during the study period, of whom 702 met the inclusion criteria. Detailed reasons for exclusion are presented in Figure 1. Among the included patients, 327 exhibited good collateral circulation (HIR <0.4), while 375 had poor collateral circulation (HIR ≥ 0.4). In the good collateral cohort, 208 patients received general anesthesia (GA), and 122 underwent local anesthesia (LA) or conscious sedation (CS). In the poor collateral cohort, 218 and 157 patients were treated under GA and LA/CS, respectively.

Before propensity score matching, significant differences in baseline characteristics, including sex, stroke etiology, history of hypertension and atrial fibrillation, mismatch volume, tirofiban use, and time from symptom onset to reperfusion, were observed in the good collateral cohort (p < 0.05, Table 1). After propensity score matching, these variables were balanced in both the good and poor collateral cohorts (Supplementary Table 1).

Restricted cubic spline (RCS) analysis demonstrated that higher HIR was associated with a lower probability of functional independence (mRS 0–2) at 90 days (unadjusted p < 0.01; adjusted p = 0.073; Supplementary Figure 1). This trend was consistent across GA and LA cohorts but did not reach statistical significance (Figure 2). There was no significant interaction between anesthesia type and functional independence (p = 0.649).

In the good collateral subgroup (HIR <0.4), GA was associated with worse functional outcomes compared to LA/CS. Specifically, the probability of achieving mRS 0–2 was significantly lower (49% vs. 70%; OR 2.88, 95% CI: 1.29–6.43). GA was also associated with worse outcomes in terms of mRS 0–1 (38% vs. 60%), mRS 0–3 (62% vs. 82%), and higher NIHSS scores at 7 days (median: 6 vs. 2; Table 2). In contrast, among patients with poor collateral circulation (HIR ≥ 0.4), outcomes were similar between GA and LA/CS, with comparable rates of mRS 0–2 (50% vs. 59%; OR 1.73, 95% CI: 0.92–3.27; Table 3).

Sensitivity analyses using five inferential models—multivariate logistic regression adjusted for all covariates, propensity score-based IPTW, and doubly robust estimations with and without adjustment for unbalanced covariates—produced consistent results. Across all models, GA was associated with worse functional outcomes in the good collateral subgroup, while no significant differences were observed in the poor collateral subgroup.

This study demonstrates that the hypoperfusion intensity ratio (HIR) is strongly associated with functional outcomes following endovascular thrombectomy (EVT) in anterior circulation occlusions. Notably, we observed a differential effect of anesthesia type based on collateral status: general anesthesia (GA) was associated with worse outcomes in patients with good collateral circulation (HIR <0.4), whereas no significant differences were found between GA and local anesthesia (LA) in patients with poor collateral circulation (HIR ≥ 0.4). These findings highlight the potential of HIR as a surrogate marker to guide anesthesia selection in EVT.

HIR, a perfusion imaging-derived biomarker, has proven to be a reliable indicator of microvascular collateral flow and tissue viability in ischemic stroke (15). Robust collateral circulation, reflected by low HIR values, is critical for preserving microvascular integrity, reducing infarct growth, and enhancing tissue salvage (16). This study underscores the value of HIR not only as a prognostic marker but also as a tool for tailoring procedural strategies, such as anesthesia choice (17).

Most drugs involved in general anesthesia can affect hemodynamics to varying degrees. In patients with well-developed collateral circulation, vasodilation in non-ischemic regions may shunt blood away from the ischemic penumbra after general anesthesia (18). Additionally, certain anesthetic drugs and perioperative factors (such as hypotension, hypercapnia, and inflammatory cascade reactions) may alter the permeability of the blood-brain barrier. Even in the presence of good collateral circulation, this can increase susceptibility to reperfusion injury (19).

The finding that GA is associated with worse outcomes in patients with good collateral circulation aligns with the hypothesis that GA-induced hemodynamic instability, such as hypotension, may disproportionately affect ischemic regions with higher perfusion reserves. In contrast, for patients with poor collateral circulation, the limited perfusion reserve may render the protective effects of LA less pronounced, resulting in similar outcomes across both anesthesia types (8, 20–22).

Our results are consistent with prior studies demonstrating worse functional outcomes with GA, including a pooled analysis of randomized trials and prospective cohorts where GA was independently associated with lower odds of good outcomes (adjusted OR: 0.64, p = 0.021) and higher rates of neurological deterioration (adjusted OR: 2.10, p = 0.045). However, these studies did not fully account for the role of collateral status. Interestingly, while prior analyses suggested no difference in outcomes based on collateral status, our study found a robust and statistically significant association between GA and worse outcomes in patients with good collateral, supported across all statistical models.

This distinction may reflect differences in study design, imaging methodologies, or statistical approaches. Importantly, no prior randomized trials have specifically evaluated the interaction between imaging biomarkers such as HIR and anesthesia type, further emphasizing the novelty and clinical relevance of our findings.

From a clinical perspective, our findings suggest that anesthesia strategies for EVT should be tailored based on pre-procedural imaging markers, such as HIR. Patients with good collateral status may benefit from avoiding GA when feasible, as LA appears less likely to disrupt hemodynamics. Integrating HIR into routine EVT workflows could enhance patient selection and optimize outcomes.

This study has several limitations. As a single-center, retrospective analysis, it is subject to potential selection bias and residual confounding, even with the use of propensity score matching and doubly robust statistical methods. The non-randomized design limits causal inference, and unmeasured confounders, such as the indication for GA in patients with more severe conditions, cannot be excluded. Intraoperative hemodynamic parameters were not recorded in this study. Blood pressure decrease caused by anesthetic drugs, the duration of intraoperative hypotension, and fluctuations in MAP can all lead to poor neurological prognosis. Therefore, we cannot determine whether the observed association is caused by the anesthetic method or hemodynamic fluctuations. Moreover, in our study, vasoactive drugs were administered only when the MAP dropped by 30%, which may result in insufficient cerebral blood perfusion. This approach is likely to lead to a poorer prognosis in patients undergoing GA. Furthermore, the skewed distribution of HIR may introduce bias in regression analyses, and HIR alone may not fully capture the complexity of collateral dynamics. Future research should focus on validating these findings in multicenter, prospective studies and randomized controlled trials. Specifically, studies should explore the interaction between imaging biomarkers, anesthesia type, and patient outcomes to establish evidence-based criteria for anesthesia selection. Mechanistic studies investigating how GA affects cerebral hemodynamics in different collateral states may further refine procedural protocols and improve EVT outcomes.

This study identifies HIR as a key biomarker for predicting functional outcomes and guiding anesthesia strategies in EVT. Our findings highlight the need for personalized approaches to anesthesia selection, particularly for patients with good collateral circulation. Further validation in larger, randomized studies is warranted to confirm the role of HIR in optimizing EVT outcomes.