Effect of Anesthesia on Functional Outcomes in Acute Ischemic Stroke Patients Undergoing Thrombectomy with the RECO Device

Introduction

Acute ischemic stroke (AIS) due to large-vessel occlusion (LVO) accounts for 38.7% of AIS, contributing to 61.6% of post-stroke dependence or death and 95.6% of mortality [1]. Diagnosis relies on clinical assessment and neuroimaging [2], and management includes intravenous thrombolysis (IVT) and mechanical thrombectomy (MT) [3]. MT is currently recognized as the most effective reperfusion strategy for patients with AIS secondary to LVO [4,5]. This endovascular intervention facilitates the recanalization of occluded intracranial arteries, thereby restoring perfusion to the ischemic penumbra and improving neurological outcomes [6]. Importantly, patient cooperation and intraprocedural stability can influence procedural success and clinical prognosis. Thrombectomy procedures are typically performed under local anesthesia (LA) or general anesthesia (GA). However, the optimal anesthesia approach remains a subject of ongoing investigation, and current evidence does not provide a definitive consensus [7].

The choice between GA and LA in MT reflects a fundamental clinical dilemma: the need to balance procedural safety against treatment speed [8]. GA can enhance procedural control, reduce patient movement, and facilitate optimal device deployment; however, it has been historically associated with concerns about delaying reperfusion [9]. LA, by contrast, can allow quicker treatment initiation but at the potential cost of reduced procedural stability, especially in agitated or medically unstable patients [10]. This trade-off remains a central issue in anesthesia decision-making for acute stroke care [11].

Previous studies have reported inconsistent findings regarding whether LA or GA is more beneficial for patients undergoing endovascular therapy [12–16]. A meta-analysis of 7 randomized controlled trials suggested that the use of GA was associated with higher recanalization rates and improved functional recovery at 3 months after stroke [17]. In contrast, a recently published randomized controlled trial found no significant differences in functional independence or major periprocedural complications between the 2 anesthesia strategies [18].

Reported clinical outcomes of MT have been inconsistent [12–16], which may be partly attributed to differences in procedural variables, particularly the thrombectomy devices used across studies. Device selection can influence both procedural efficiency and patient response under different anesthesia modalities, highlighting the need to evaluate anesthesia strategies within a standardized interventional context. The Solitaire FR was the first stent retriever with proven efficacy in MT [19], and subsequent development of novel stent retrievers – including Tigertriever [20], ThrombX [21], Nimbus [22], and Trevo [23] – has aimed to further improve recanalization rates, reduce complications, and enhance patient outcomes.

The RECO flow restoration device is a stent-based system with adjustable mesh segments that securely capture thrombi, aiming to improve clot retrieval while minimizing vessel injury [24]. Clinical studies support its safety and effectiveness: a multicenter registry reported high rates of successful reperfusion and favorable 90-day outcomes [24], and the REDIRECT trial demonstrated comparable efficacy and safety between RECO and Solitaire FR [25]. These findings establish RECO as a reliable option for first-line thrombectomy.

We previously conducted a prospective study focusing on the RECO stent retriever [24]. This single-arm, prospective, multicenter study consecutively enrolled patients undergoing endovascular therapy with the RECO device as first-line treatment at 9 stroke centers. Based on this comprehensive dataset, which minimizes variability related to device selection, the present analysis specifically aimed to compare the effects of anesthesia modalities – GA versus LA – on clinical outcomes, procedural parameters, and safety profiles in patients with LVO treated with MT in routine clinical practice. This study seeks to provide clearer insights into how anesthesia choice can influence treatment effectiveness and safety in a real-world, multicenter context. Therefore, this study aimed to compare outcomes in 266 patients with AIS from LA and GA following mechanical large-vessel thrombectomy using the RECO device.

Material and Methods

ETHICS STATEMENT:

The RECO study was designed to comply with the Declaration of Helsinki and was approved by the ethics committees of Beijing Tiantan Hospital (approval number: KY 2019-080-03) and all collaborating institutions. Written informed consent was duly obtained from all study participants or their legally authorized representatives prior to enrollment. This study accessed the multicenter RECO data through formal approval from the lead coordinating institution (Beijing Tiantan Hospital, Capital Medical University), with additional authorization from our participating center. Data were processed on our HIPAA/GDPR-compliant secure servers under ethics oversight from our Institutional Review Board (2025-KY-043-01). All collected data were anonymized before analysis, and access to identifiable information was restricted to authorized personnel only.

STUDY PARTICIPANTS:

The RECO study was a single-arm, prospective, multicenter post-market study https://www.Clinicaltrials.gov; unique identifier: NCT04840719) that evaluated the safety and performance of the RECO device in patients with AIS and LVO. Between May 22, 2020, and July 30, 2022, 288 patients were enrolled and underwent first-line thrombectomy with the RECO device at 9 stroke centers. Patients aged 18 or older with confirmed AIS, LVO evidence, Alberta Stroke Program Early CT Score (ASPECTS) or posterior circulation (PC)-ASPECTS ≥6, and MT initiation were included. We excluded patients with (1) isolated cervical internal carotid artery or vertebral artery occlusion, (2) no angiographically confirmed LVO, or (3) contraindications to GA/LA. The present analysis is a secondary investigation derived from this dataset. In contrast to the primary study, which focused on overall device outcomes, the present analysis specifically examined the impact of anesthesia modality (GA vs LA) on clinical outcomes. Among 268 initially screened patients, 2 were excluded due to missing anesthesia modality data, yielding 266 patients for final analysis.

DIAGNOSIS OF AIS:

AIS was diagnosed based on the presence of acute neurological deficits consistent with a focal cerebral ischemic event, which was confirmed by neuroimaging [computed tomography (CT) or magnetic resonance imaging (MRI)] showing evidence of acute infarction or the exclusion of other pathologies such as hemorrhage. Patients were evaluated by stroke neurologists upon admission, and stroke severity was quantified using the National Institutes of Health Stroke Scale (NIHSS) score.

DIAGNOSIS OF LVO:

LVO was confirmed using multimodal neuroimaging. Non-contrast CT was performed to exclude intracranial hemorrhage, followed by CT angiography (CTA), magnetic resonance angiography (MRA), or digital subtraction angiography (DSA) to identify occlusions in the intracranial internal carotid artery (T/L), middle cerebral artery (M1/M2), anterior cerebral artery (A1/A2), basilar artery (BA), vertebral artery (V4), or posterior cerebral artery (P1). Perfusion imaging was used when necessary to assess salvageable tissue.

ENDOVASCULAR PROCEDURE:

All patients underwent MT with the RECO device. The procedure was performed by experienced neurointerventionalists under LA or GA, according to institutional practice. Vascular access was obtained via femoral puncture, and a guiding catheter was positioned in the target artery. The RECO device was deployed across the thrombus to achieve clot engagement, followed by retrieval under aspiration when possible. Multiple passes were permitted at the discretion of the operator until successful reperfusion or procedural termination.

DATA COLLECTION:

The patients were stratified into 2 groups based on anesthesia modality: the LA group and GA group. Data encompassing patient demographics, prior medical history, prestroke functional status (assessed via the modified Rankin scale [mRS], originally developed as a post-stroke outcome scale [26] but widely used in stroke studies to evaluate baseline function to the index event [27]), trial of org 10172 in acute stroke treatment (TOAST) classification, baseline NIHSS score, physiological measurements, laboratory biomarkers, neurovascular imaging findings, time metrics across clinical workflows, and treatment outcomes were prospectively gathered using standardized protocols prior to the index stroke incident.

OUTCOME MEASURES:

The primary outcomes were: (1) Functional independence were assessed at 90±7 days using the mRS (range 0–6, with 0 indicating no symptoms and 6 indicating death), with favorable outcome defined as mRS 0–2. A neurologist (with 5+ years of stroke outcome assessment experience) performed evaluation through standardized telephone interview or outpatient clinic visit. (2) Symptomatic intracranial hemorrhage (sICH) was assessed via standardized non-contrast CT scans within 12–36 h after the intervention. Each scan was independently evaluated by a radiologist and a neurologist, applying the Heidelberg Bleeding Classification [28] while blinded to treatment allocation and clinical outcomes. Discrepancies were resolved through consensus review of the original digital imaging and communications in medicine (DICOM) images by a senior stroke neurologist with ≥10 years’ neuroimaging experience.

Secondary measures included the following: (1) Success reperfusion defined as modified thrombolysis in cerebral infarction (mTICI) grade 2b to 3 at the end of the procedure. Two neurointerventionalists (each with ≥5 years of endovascular experience) jointly assessed the final angiographic runs. In cases of disagreement, a third senior interventional neuroradiologist (≥10 years’ experience) reviewed the unlabeled DICOM images for final adjudication; (2) Ordinal analysis of the full mRS spectrum (0–6) to assess global disability shifts.

STATISTICAL ANALYSIS:

Data analyses were conducted using IBM SPSS Statistics (v21.0, IBM Corp., Armonk, NY). Normality testing was performed using the Shapiro-Wilk test. Numerical variables with non-normal distributions were expressed as medians with interquartile ranges (IQRs), whereas categorical data were summarized as counts and percentages. Between-group comparisons employed nonparametric Mann-Whitney U tests for continuous/ordinal variables and Pearson’s χ² test or Fisher’s exact test for categorical variables, as appropriate for sample size requirements. Subsequently, to improve the balance between groups, we performed 1: 1 propensity score matching (PSM) with nearest-neighbor matching (caliper width=0.1, without replacement). All subsequent comparative analyses were then carried out in this matched cohort. To identify independent predictors of 90-day poor outcomes, variables demonstrating marginal associations (P<0.10) in preliminary univariate screening were subsequently entered into multivariable logistic regression models. To address the potential influence of inter-center variability in this multicenter study, we performed a sensitivity analysis by sequentially excluding each participating center and re-evaluating the outcomes. A two-tailed P value of <0.05 was set a priori as the threshold for statistical significance.

Results

BASELINE CHARACTERISTICS:

Of the 268 patients in the RECO study, 2 lacked available anesthesia modality data; therefore, a total of 266 patients were included in the final analysis, with 119 patients in the LA group and 147 in the GA group. After PSM, 30 patients in the LA group and 30 in the GA group were identified.

Before PSM, no significant differences in sex, TOAST classification, lifestyle factors, or major comorbidities were observed between groups. However, a significantly higher proportion of patients in the LA group were transferred from other hospitals compared to the GA group [56 (47.1%) vs 21 (14.3%), P<0.001]. In addition, the use of IVT was more frequent in the GA group [43 (29.3%)] compared to the LA group [16 (13.4%), P=0.002]. Median baseline NIHSS score was significantly higher in the LA group [18 (IQR: 15–24)] than in the GA group [16 (IQR: 12–22), P=0.040]. Similarly, time metrics were notably longer in the LA group: onset-to-door time [182 (IQR: 99–299) vs 96 (IQR: 54–224) minutes, P<0.001], onset-to-puncture time [319 (IQR: 210–431) vs 205 (IQR: 154–331) minutes, P<0.001], and onset-to-reperfusion time [374 (IQR: 270–515) vs 300 (IQR: 212–421) minutes, P<0.001]. No significant difference was observed in puncture-to-reperfusion time [55 (IQR: 41–100) vs 64 (IQR: 46–90) minutes, P=0.328] or door-to-puncture time [101 (IQR: 69–152) vs 95 (IQR: 76–119) minutes, P=0.286) (Table 1).

After PSM, the clinical characteristics between the 2 groups were well balanced.

CLINICAL OUTCOMES:

Table 2 showed the clinical outcomes of the LA and GA group. In the pre-matched population, final reperfusion, defined as achieving mTICI grade 2b or 3, was similar between the 2 groups (95.7% in LA vs 97.3% in GA, P=0.481). However, the incidence of sICH was significantly higher in the LA group [11/89 (12.4%)] compared to the GA group [5/132 (3.8%), P=0.016)]. At 90 days, favorable functional outcome, defined as a mRS score of 0–2, was significantly more frequent in the GA group [85/147 (57.8%)] than in the LA group [46/119 (38.7%), P=0.002] (Table 2). Distribution of the 3-month mRS scores also showed significant differences (P=0.001), with a higher proportion of patients in the GA group achieving mRS 0 (40.8% vs 14.3%) and a lower proportion reaching mRS 6 (13.6% vs 22.7%) (Figure 1).

Among the post-matched patients, the proportion of favorable functional outcome was significantly higher in the GA group [17/30 (56.7%)] compared to the LA group [8/30 (26.7%), P=0.018]. Both groups demonstrated similar rates of sICH.

UNIVARIATE AND MULTIVARIATE PREDICTORS OF POOR FUNCTIONAL OUTCOME:

Univariate analysis identified several variables associated with poor prognosis at 90 days, including male sex (P=0.020), history of being transferred from another hospital (P=0.032), absence of IVT (P=0.003), higher baseline NIHSS score (P<0.001), and longer onset-to-door time (P=0.049). The use of GA was associated with a significantly better prognosis compared to LA [favorable functional outcome: 85/131 (64.9%) vs 62/135 (45.9%), P=0.002] (Table 3).

Multivariate logistic regression analysis (Table 3) identified higher baseline NIHSS score as an independent predictor of poor outcome [adjusted odds ratio (OR) 1.079; 95% CI: 1.037–1.123; P<0.001]. The use of GA showed a marginally significant association with favorable outcome (adjusted OR 0.557; 95% CI: 0.311–1.000; P=0.050). The adjusted odds ratio of 0.557 indicates that GA was associated with approximately 44.3% lower odds (1–0.557) of poor outcomes compared to the LA group. Although this protective effect was marginally significant (P=0.050) and the confidence interval includes the null value (upper limit 1.000), the point estimate suggested a potentially clinically relevant reduction in adverse outcomes. IVT trended toward an independent protective effect but did not reach statistical significance (adjusted OR 0.548; 95% CI: 0.277–1.082; P=0.083). Other variables, such as sex, diabetes, and inter-hospital transfer, were not independently associated with outcomes in the multivariate model.

SENSITIVITY ANALYSIS:

The sensitivity analysis results showed that no single center exerted a dominant effect on baseline characteristics or primary/secondary endpoints, suggesting that center-level clustering did not significantly bias our results.

Discussion

This study compared the clinical outcomes of LA and GA in patients with AIS undergoing MT. The GA group had higher 90-day functional independence. A notable strength of our study is the uniform use of the same thrombectomy device, which minimizes procedural variability and allows a clearer comparison of anesthesia strategies.

Wu et al reported similar functional outcomes and complication rates between GA and LA, indicating no significant advantage of either approach [16]. In contrast, the GSR-ET registry study by Feil et al found that conscious sedation (CS) was associated with better functional outcomes, lower rates of periprocedural complications, and reduced mortality compared with GA, suggesting potential disadvantages of GA in routine practice [14]. The systematic review and meta-analysis by Goyal et al also highlighted this inconsistency: pooled analyses suggested higher odds of functional independence with non-GA [15]. Finally, Meyer et al demonstrated that GA and LA/CS yielded comparable reperfusion and safety outcomes for distal medium-vessel occlusions, although GA can facilitate complete reperfusion in anterior cerebral artery strokes [12].

Taken together, these studies illustrate the heterogeneity in outcomes associated with anesthesia strategies in MT, which can be influenced by patient selection, procedural factors, and vessel location. Our study adds to the current evidence by focusing on a multicenter cohort treated with a uniform stent retriever (the RECO device), minimizing device-related variability, and showing that GA can confer an advantage in functional outcomes without increasing procedural risk.

Although GA was associated with better clinical outcomes in our study, it is important to acknowledge several baseline differences that may confound this association. As shown in Table 1, patients in the GA group had significantly shorter onset-to-door, onset-to-puncture, and onset-to-recanalization times compared with those in the LA group. These time intervals are known predictors of prognosis in AIS and can contribute to improved outcomes independently of the anesthesia modality. Additionally, IVT was more frequently administered in the GA group, which may have further influenced functional outcomes. Notably, initial NIHSS scores were higher in the LA group, suggesting a greater stroke severity at presentation, potentially biasing results against the LA cohort. While we adjusted for several baseline variables in multivariate analyses, these imbalances raise the possibility of residual confounding and limit the ability to draw causal inferences. Additionally, after matching analysis, the 2 groups showed similar baseline characteristics, supporting balanced comparison. Although the GA group demonstrated better outcomes, the small sample size necessitates cautious interpretation of this finding.

The optimal anesthesia method for MT remains uncertain[18,29,30]. Earlier analyses like MR CLEAN and HERMES favored LA [31,32], but potential bias limits interpretation. Guidelines also reflect this ambiguity, with both European and American stroke associations accepting either approach [33,34].

Given the ongoing debate and variability in clinical practice, our study aimed to contribute further evidence by focusing on a more homogeneous setting. Unlike many previous trials, we ensured the use of a single type of thrombectomy device across all cases, thereby minimizing procedural variability. This approach allowed a more direct comparison between GA and LA while controlling for confounding factors related to device performance. Despite multivariable adjustments, residual confounding may persist due to unmeasured factors such as differences in anesthesia protocols and clinical practices across participating centers, variability in operator experience, and patient comorbidities not captured in the dataset. For instance, centers with more experienced teams or standardized anesthesia procedures might achieve better outcomes independent of anesthesia type. Similarly, variations in perioperative management, blood pressure control, or patient selection criteria could contribute to observed differences. These uncontrolled factors limit the ability to definitively attribute outcome differences to anesthesia modality alone.

While no significant difference was observed in final mTICI scores between the 2 groups (P=0.654), our initial analysis suggested a higher rate of postprocedural hemorrhage in the LA group. However, after performing PSM to account for baseline imbalances, this difference in sICH rates was no longer statistically significant. The discrepancy between our initial and matched analyses likely reflects the impact of confounding factors in the unmatched cohorts, where the LA group presented with higher baseline NIHSS scores indicating greater stroke severity which is a known risk factor for hemorrhagic complications. Previous studies have suggested that patient agitation under LA might increase procedural risks [35], but our matched analysis indicates that when accounting for baseline characteristics, the choice of anesthesia type did not independently affect bleeding risk.

In terms of functional outcomes, a significantly higher proportion of patients in the GA group achieved a favorable prognosis. This finding is consistent with previous reports [13]. One possible explanation is the ability of GA to provide better control over blood pressure fluctuations and stress responses during the procedure, thereby maintaining a more stable physiological environment. Conversely, some studies have suggested that LA also confers advantages by avoiding the hemodynamic effects of induction and intubation, potentially leading to more stable blood pressure in selected patients. Such stability may help minimize secondary brain injury and contribute to improved postoperative recovery.

Although retrospective studies have suggested that intubation-related delays under GA can negatively impact outcomes [36], our study did not observe such delays. The door-to-puncture time was slightly shorter in the GA group compared to the LA group (95 min vs 101 min), while puncture-to-recanalization time was marginally longer (64 min vs 55 min), with no statistically significant differences. These findings align with previous studies [13], and others have even reported shorter procedural times under GA, likely due to improved surgical conditions and patient immobility [37]. These observations suggest that, when well-coordinated workflows are in place, GA does not delay treatment and may even enhance procedural efficiency, potentially contributing to the better outcomes observed in the GA group.

Although our findings suggest a potential advantage of GA in improving outcomes, LA may still be appropriate in selected patients – particularly those unable to tolerate GA or with specific contraindications. Moreover, LA may be a more suitable choice in clinical scenarios involving patients with cardiorespiratory instability, a known difficult airway, or other contraindications to GA. It may also be preferable in emergency situations requiring rapid intervention when anesthesiology support is unavailable or delayed. These considerations highlight the importance of tailoring anesthesia strategies to individual patient profiles and institutional capacities, rather than applying a uniform approach. Further studies are needed to explore the differential impact of anesthesia strategies across various subgroups of stroke patients to optimize individualized treatment approaches.

This study has several limitations. First, although it was conducted across multiple centers, variations in local protocols and operator experience may have introduced unmeasured confounding factors. Second, while the sample size was moderate, larger-scale studies are warranted to validate our findings and enhance statistical power. Third, despite the consistent use of a single thrombectomy device, other procedural variables – such as differences in anesthetic management or operator techniques – may still have influenced the outcomes. Furthermore, all procedures used the RECO device, reducing variability but limiting generalizability. The findings may not apply to other devices or populations, warranting further research. Additionally, while PSM was performed to address baseline imbalances, the resultant matched cohort had a relatively small sample size, potentially limiting the statistical power to detect clinically meaningful differences. Moreover, the decision to use GA or LA was based on individual clinician judgment and institutional practice rather than a standardized protocol, which may have introduced selection bias and affected outcome comparisons. This limitation highlights the need for prospective studies with uniform anesthesia selection criteria to better elucidate the impact of anesthesia modality on MT outcomes.

Conclusions

GA may be associated with better 90-day functional independence compared to LA in patients with AIS following mechanical large-vessel thrombectomy using the RECO device. This finding needs to be further validated in prospective or randomized controlled studies.

Data Availability Statement

The datasets used and analyzed during the current study are available upon reasonable request.