PDF
Citation
Print
INTRODUCTION
Endovascular mechanical thrombectomy (EMT) represents the cornerstone of treatment for acute ischemic stroke (AIS) and can be performed with general anesthesia (GA) or non-GA techniques. Age, comorbidities, pre-stroke functional status, procedural factors, site of the clot, institutional experience, and resource availability influence the choice of anesthesiological management, requiring a patient-tailored approach [1].
GA and non-GA techniques both have pros and cons. On the one hand, GA offers a still patient, a secure airway, and improved patient and neurointerventionalist comfort. However, at the same time, GA can delay the beginning of the procedure, a critical factor in AIS management where “time is brain.” Furthermore, GA increases the risk of intraoperative hypotension with reduced cerebral perfusion and does not allow prompt neurological examination [2,3]. On the other hand, non-GA techniques can be initiated sooner, are associated with a reduced risk of intraoperative hypotension and postoperative pulmonary complications, and allow real-time neurologic examination. However, ineffective pain management, spontaneous patient movements, and failure to completely protect the airway, especially in the case of prolongation of the procedure, may require urgent conversion to GA.
Despite several meta-analyses of whether GA or non-GA techniques are superior in the context of EMT, findings regarding the main outcomes are discordant [4-7]. The decision to use GA or non-GA should ideally be based on the patient’s characteristics, considering clinical features, stroke localization, and institutional experience based on the current literature [8,9]. One of the main reasons for this confusion is the extensive and indiscriminate use of the term "non-GA" techniques, which can include very different levels of sedation, from simple local anesthesia (LA) to deep sedation. Second, to the best of our knowledge, sedation levels in non-GA interventions are not routinely assessed or reported using tools such as the Richmond Agitation-Sedation Scale (RASS) [10]. The lack of procedure standardization makes comparison among studies difficult.
The aim of this propensity score-matched (PSM) retrospective single-center study was to analyze the relationship between clinical outcomes and GA and non-GA techniques in patients who underwent EMT for acute anterior ischemic stroke.
MATERIALS AND METHODS
This retrospective study was approved by local ethical committee "Campania 2", (protocol ID 2025/5133). Patient privacy was rigorously protected in accordance with current national legislation (General Data Protection Regulation of the European Union number 2016/679 and the Italian Legislative Decrees number 196/2003 and 101/2018). The study was conducted following the International Conference on Harmonization Good Clinical Practice guidelines and the 2008 Declaration of Helsinki provisions. Written informed consent for procedures and data collection was obtained from all patients or, when appropriate, their relatives. The present study is reported according to STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidelines.
Patients
For this analysis, we used our local database, which included all patients admitted to AOU San Giovanni di Dio e Ruggi d’Aragona who required EMT for AIS from January 1, 2018 to December 31, 2021. All data were collected in an anonymous electronic database.
Objectives
The main goal of this study was to evaluate the relationship between anesthesiological management and the following outcomes: in-hospital death, 90-day functional independence (defined as modified Rankin Scale [mRS] ≤2), 90-day death rate, full recanalization (defined as thrombolysis in cerebral infarction [TICI] score 2b-3), procedural complications, and intracerebral hemorrhage (ICH).
Data Collection
We collected patient demographics, past medical history, home therapies, baseline mRS, details of in-hospital presentation and EMT procedure (including site location, techniques, and timing), anesthesiological management (GA or non-GA), and anesthesia and procedure-related complications. Anesthesia-related complications were defined as hypotension (mean arterial pressure [MAP] <65 mm Hg) unresponsive to drugs, hypoxemia (oxygen saturation <92%), aspiration, or the need for emergent conversion from non-GA to GA due to respiratory failure or excessive patient movement. Procedure-related complications were defined as subarachnoid hemorrhage/arterial dissection, distal clot migration, and other (local hematoma/pseudoaneurysm). Length of stay was also recorded.
Intervention Definitions
According to our local protocol, the attending anesthesiologist made the choice regarding anesthetic management (GA or non-GA) after carefully evaluating the patient's clinical conditions and the procedure to be carried out. Standard intraoperative monitoring included SpO2, heart rate, electrocardiogram, non-invasive blood pressure, end-tidal CO2 (EtCO2), and diuresis. EtCO2 was monitored to compensate for oxygen saturation.
For GA, rapid sequence intubation was performed with fentanyl (1.0–1.5 μg/kg), propofol (1.0–1.5 mg/kg) or midazolam (0.05–0.1 mg/kg), and rocuronium (1.2 mg/kg). Endotracheal intubation was followed by mechanical ventilation with attempted normoventilation. Anesthesia was maintained with propofol and remifentanil with simple (respectively dosage 2.0–10.0 mg/kg/hr and 0.05–0.15 μg/kg/min) or targeted controlled infusion. At the end of the procedure, in stable clinical conditions, sedation infusions were stopped, sugammadex (2.0–4.0 mg/kg) was administered, and patients were extubated in the neurointerventional suite immediately after the procedure.
LA plus sedation was the primary non-GA technique used in this study (lidocaine 2.0%). The attending anesthesiologist administered drugs until an adequate sedation level (RASS 0–3) was reached, ensuring both patient and neuroradiologist comfort. The anesthesiologist could choose between drugs administered as single or repeated boluses (fentanyl 50–100 μg or midazolam 1.0–5.0 mg) or as propofol (1.0–2.0 mg/kg/hr) or remifentanil (0.05–0.1 μg/kg/min) or dexmedetomidine (0.6–1.2 μg/kg/hr) continuous infusions. During non-GA, patients received supplemental oxygen by face mask and oxygen flow rate was titrated with the goal of a stable SpO2 target >92%, with EtCO2 monitoring. The anesthesiologist continuously evaluated the occurrence of airway obstruction and adopted the appropriate maneuvers (e.g., Guedel cannula placement, sedation infusion flow rate reduction). Fluids and drugs were titrated to reach a MAP value ≥65 mm Hg. In cases with MAP <65 mm Hg, a single bolus of ephedrine (5.0 mg) or noradrenaline (10.0 μg) was administered. Noradrenaline infusion was started if clinically needed.
Explorative Analysis and Group Definitions
Before proceeding to the statistical computations, we performed an explorative analysis. In the identification phase, we evaluated the presence of missing data (MD) in the following variables: intubation on admission, time from event to in-hub admission, anesthesiological management, National Institutes of Health Stroke Scale (NIHSS) on admission, basal Alberta Stroke Program Early Computed Tomography Score (ASPECT), basal, discharge, and 90-day mRS, occlusion site, and procedure performed. MD were considered to be missing completely at random, and observations were deleted. In the screening phase, we applied the following exclusion criteria: age <18 years, pre-stroke mRS >2, patients intubated on admission, posterior circulation site occlusion, performing not only EMT, and procedure performed only with LA. Then, we divided our sample into GA and non-GA groups. In the non-GA group, we excluded patients requiring urgent GA for a deteriorating clinical condition during EMT.
Statistics
Descriptive statistics were computed. Categorical variables are reported as absolute numbers and percentages (%). Continuous data were tested for a normal distribution with the Shapiro-Wilk test and are reported as mean±standard deviation or, otherwise, as median and interquartile range (IQR). For MD, statistics were computed based on the available data. For non-GA and GA unmatched groups, comparisons were performed with the chi-square test for categorical variables and Student t-test or the Mann-Whitney U-test for continuous variables based on the distribution.
PSM analysis was performed with the nearest neighbor matching method to limit differences in baseline characteristics between patients receiving different types of anesthesiological management. We choose covariates that influence clinical decisions about anesthesiological management for PSM: age, smoking status, symptoms upon awakening, occlusion side, vessel occlusion, time from event to in-hub admission, pre-stroke mRS, NIHSS, and basal ASPECT. PSM balance was assessed by checking standardized mean differences between covariates, with a value <0.2 indicating negligible imbalance between the groups [11]. We performed logistic regression to determine the relationship between anesthesiological management and clinical outcomes. Odds ratio with 95% CI were computed. All P-values were two-sided, and a P-value <0.05 was considered significant. In the case of MD in outcome variables, we computed the statistics using the available data. All analyses were performed using R-studio (R Core Team 2024. R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing).
RESULTS
Patient Characteristics
After deleting MD, we obtained 351 observations from 486 observations. Application of exclusion criteria decreased the total number of observations to 196: 78 in the GA group and 118 in the non-GA group. In the non-GA group, two patients (1.7%) required urgent conversion to GA for worsening gas exchange during EMT. No other anesthesia-related complications were reported. The present analysis was therefore performed on 194 observations (78 in the GA group and 116 in the non-GA group). The flowchart for records selection is shown in Figure 1, and the main characteristics of the overall sample are described in Table 1. Median age was 71.2 years (IQR, 62.6–81.3 years), and 91 patients (46.9%) were male. The three most common comorbidities were hypertension (160 patients, 82.5%), hyperlipidemia (68 patients, 35.0%), and atrial fibrillation (67 patients, 34.5%). One hundred forty-five patients (74.7%) were directly admitted to hub, with a median time from the event of 130.0 minutes (IQR, 85.0–236.2 minutes). Thirty-four patients (17.5%) displayed symptoms upon awakening, and the most common occlusion side was left (106 patients, 54.6%), mainly large vessel occlusion (LVO; 166 patients, 85.6%). Frequencies of pre-stroke mRS equal to 0, 1, and 2 were 125 (64.4%), 58 (29.9%), and 11 patients (5.7%), respectively. NIHSS and ASPECT median scores were 14.0 (IQR, 10.0–18.7) and 9.0 (IQR, 8.0–10.0), respectively.
Median onset to groin time was 4.4 hours (IQR, 3.1–6.4 hours), and A ADAPT (Direct Aspiration first Pass Technique) was the most common technique (93 patients, 48.2%). Median number of attempts was 2.0 (IQR, 1.0–3.0), and median procedural time was 40.0 minutes (IQR, 20.0–64.2 minutes). Full recanalization (TICI 2b-3) was obtained in 167 patients (86.1%). Procedural complication rate was 36.1% (70 patients), and the most common procedural complication was distal clot migration (62 patients, 31.9%). Median NIHSS after 24 hours was 7.0 (IQR, 4.0–11.0), with a median percentage variation of –47.1% (IQR, –66.7% to –27.8%). Symptomatic and asymptomatic ICH rates were 4.5% (8 patients) and 19.8% (35 patients). In-hospital mortality rate was 5.1% (10 patients), and median mRS at discharge was 1.0 (IQR, 0.0–3.0). The 90-day mRS median value was 1.0 (IQR, 0.0–3.0), and 144 patients (74.2%) were functionally independent (mRS 0–2). The 90-day mortality rate was 5.7% (11 patients).
PSM Groups
No MD were noted in covariates used for PSM. After PSM, 70 pairs of observations were obtained. Fifty-six observations in the non-GA group and eight in the GA group were excluded (Figure 3). Results of statistical comparisons between the PSM groups are presented in Table 2. After PSM, no differences were noted in patients, characteristics, home therapies, in-hospital presentation, procedures, complications, or main outcomes.
Logistic Regression Analysis
Results of logistic regression analyses in the unmatched and PSM groups are presented in Table 3. Performing EMT with GA was unrelated to in-hospital and 90-day death, 90-day functional independence, full recanalization rate, procedural complications, or ICH (all P>0.05).
DISCUSSION
Our study's main finding was that anesthesiological management was not related to our clinical outcomes of interest in either the unmatched or PSM groups. The latest meta-analyses have emphasized the role of GA in improving the rate of recanalization of the occluded vessel. Campbell et al. [5] reported that GA was associated with higher rates of successful recanalization and functional independence in LVO patients treated with EMT when compared with non-GA techniques, providing high-quality evidence that GA should be the first choice in patients treated with EMT in those centers able to provide expert anesthesiology services. They advised that updated guidelines should incorporate a level 1A recommendation for improved recanalization with GA and a level 1B recommendation for functional recovery. However, other meta-analyses did not demonstrate a relationship between full recanalization rate and functional outcomes. Santos et al. [6] showed that among patients who underwent EMT, despite the increase in recanalization success rates in the GA group, GA and conscious sedation (CS) groups showed similar rates of good functional recovery, 3-month mortality, incidence of pneumonia, and ICH.
Moreover, other meta-analyses have highlighted the risks arising from GA. Al-Salihi et al. [4] reported that despite GA having superior recanalization rates, functional outcomes, mortality, and NIHSS scores were similar to those of patients who underwent EMT using non-GA techniques, with a higher risk of hypotension. Jia et al. [7] demonstrated that the choice of anesthetic modality did not influence 3-month neurological outcomes despite GA being superior to CS in terms of successful reperfusion rate. However, patients in the GA group were at a higher risk of developing hypotension and pneumonia. In our analysis, GA was not associated with an increased successful recanalization rate, as defined by a TICI score 2b-3. The discrepancy between our results and the literature evidence is likely related to two factors: first, some meta-analyses [5,7] included studies of patients with posterior circulation occlusion, which is a more serious clinical condition than occlusion of the anterior circulation; second, the term non-GA technique in an umbrella term that covers different approach, from simple LA to CS or moderate-deep sedation.
Regarding the first explanation for the discrepant results, recent meta-analyses included two studies reporting data on posterior circulation stroke [12,13]. Hu et al. [12], in a single-center, randomized, prospective, blinded end-point cohort study, enrolled 139 patients who underwent EMT for posterior occlusion stroke. The primary outcome of mRS at 90 days was not significantly different between the GA and non-GA groups (defined as Monitored Anesthesia Care [MAC]). Moreover, they found comparable reperfusion rates between the two groups. Nevertheless, the infarct volume growth and final infarct volume were higher in the GA group as a possible consequence of procedural delays and longer door-to-groin puncture times. Liang et al. [13] performed a randomized parallel-group exploratory trial with blinded end-point evaluation (Choice of Anesthesia for Endovascular Treatment of Acute Ischemic Stroke [CANVAS II]). They enrolled 87 patients with posterior occlusion stroke. They found that CS was not better than GA with regard to the primary outcome of functional recovery and was perhaps worse for the secondary outcome of successful reperfusion. When these studies were included in meta-analyses, differences in sample size may have influenced the results regarding recanalization rate outcomes. However, posterior occlusion stroke is a more severe event, also demonstrated by the higher conversion rate than in studies of patients with anterior circulation stroke.
Furthermore, the term “non-GA” is an umbrella term that includes all techniques that are not GA. Its extensive use generates confusion by making different sedation strategies equivalent while not taking into account the fact that sedation is a continuum and it is not always possible to predict how a patient will respond. In our study, we defined non-GA as a drug-induced depression of consciousness according to a standardized targeted goal (RASS score of 0 to –3). Moreover, our local policy always involves the presence of an anesthesiologist during EMT able not only to provide adequate sedation but also to treat complications related to over-sedation or worsening neurological status. McCusker et al. [14], in a retrospective analysis, demonstrated that routine involvement of an anesthesia team during EMT was not associated with improved outcomes but was associated with improved efficiency and greater adherence to guidelines-based physiological parameters. Padmanaban et al. [15] performed a PSM analysis comparing the safety and efficacy of EMT performed with nursing-administered CS supervised by a trained interventionalist and MAC supervised by an anesthesiologist. They found that CS managed by a nurse was safe and effective, but at the same time, patients undergoing MAC received significantly more vasoactive medications and had a lower intraoperative minimum systolic blood pressure without any differences in procedural efficacy, safety, intubation rates, or postoperative complications.
This evidence suggests that, regardless of anesthesiological management, hemodynamic and other physiologic parameters (e.g., ventilation, temperature, blood glucose) may be more important determinants of outcomes [16]. Ordies et al. [17], in a retrospective analysis, demonstrated that hypotensive patients (MAP <65 mm Hg) at any given time during EMT under GA had worse neurological outcomes than non-hypotensive patients. Whalin et al. [18], in a retrospective analysis, demonstrated that a ≥10% MAP drop from baseline was a strong risk factor for poor outcomes in a homogeneous population of patients with stroke undergoing EMT under non-GA methods.
Although a MAP threshold is not well defined in this setting, we titrated fluids and drugs to a MAP value ≥65 mm Hg according to our hospital’s protocol. Proper hemodynamic management could explain the lack of statistically significant differences in the outcomes analyzed in the present study, suggesting that hemodynamics may play a more critical role than anesthesiological management in determining outcomes of EMT. Further studies are needed to elucidate the influence of hemodynamic-targeted protocols on EMT outcomes.
Our study had several limitations. First, PSM can only balance differences in observed covariates between groups. If unobserved variables influence both the probability of receiving treatment and outcomes, these may introduce a bias that PSM cannot correct for. Moreover, and as in our case, PSM results in lost observations, as only units that can be matched are included in the analysis, reducing the number of observations available and potentially limiting the generalizability of the study findings. This effect was noted for the age variable. In fact, in the unmatched groups, the age difference was statistically significant; patients who received GA were younger than those in the non-GA group. Because we considered age as a covariate influencing clinical decisions about anesthesiological management, elderly patients were not matched. Consequently, it was difficult to exclude the possibility that anesthesiological management could affect outcomes in very elderly patients.
Second, we defined non-GA techniques as LA plus targeted sedation based on RASS. We avoided the use of unclear terms such as “CS,” “moderate/deep sedation,” or “MAC,” but were unable to evaluate the level of sedation achieved for each patient, as well as the drugs used. Consequently, we could not perform sub-analyses to explore the effects of different sedation levels and drugs on outcomes. Third, we excluded patients requiring conversion from non-GA to GA. In our population, the conversion rate to urgent GA was low (1.7%), and we were therefore unable to perform a subgroup analysis to find risk factors related to non-GA failure. Fourth, most evidence supports an increased risk of pulmonary complications when EMT is performed with GA. However, we did not collect data on this topic and did not explore it.
In our single center, PSM, retrospective study, anesthesiological management (GA vs. non-GA plus LA) did not affect primary clinical outcomes in patients undergoing EMT for acute anterior ischemic stroke. Despite previous data suggesting that GA might positively influence recanalization, we found no significant difference in outcomes between GA and non-GA groups. This lack of outcome differences between the groups could reflect the importance of optimal physiologic management during EMT rather than the choice between GA and non-GA. Our results support the notion that physiological stability during EMT may impact outcomes more significantly than the type of anesthetic management used. Further studies on this topic are needed.
KEY MESSAGES
▪ Endovascular mechanical thrombectomy is the cornerstone treatment for acute ischemic stroke and can be performed with general anesthesia or non-general anesthesia techniques.
▪ Despite several meta-analyses, it is unclear if anesthesiological management influences the main outcomes of patients who undergo endovascular mechanical thrombectomy for acute anterior ischemic stroke.
▪ In this single-center, retrospective, propensity score-matched study, clinical outcomes of patients who underwent endovascular mechanical thrombectomy for acute anterior ischemic stroke were not influenced by anesthesiological management.
XML Download