General Anesthesia Versus Conscious Sedation for Mechanical Thrombectomy in Patients With Acute Ischemic Stroke in the MOST Trial

Abstract

Background

Several observational studies have supported the use of conscious sedation (CS) for endovascular thrombectomy in patients with acute ischemic stroke and associated general anesthesia (GA) with poor functional outcomes. Recently, few randomized controlled trials have shown no difference in the functional outcomes between GA and CS. The aim of this paper is to compare the outcomes of GA versus CS in patients with acute ischemic stroke undergoing mechanical thrombectomy in the MOST (Multi‐arm Optimization of Stroke Thrombolysis) trial.

Methods

Patients who underwent mechanical thrombectomy for acute ischemic stroke under GA or CS in the MOST trial were included. The primary outcome of interest was the utility weighted modified Rankin Scale (mRS) score at 90 days after stroke. The secondary outcomes were mRS score ≤1, ≤2 at 90 days, 90‐day mRS score, thrombolysis in cerebral infarction 2B or better, 24‐hour National Institutes of Health Stroke Scale score, 24‐hour change in National Institutes of Health Stroke Scale score, and 90‐day mortality.

Results

A total of 219 patients underwent mechanical thrombectomy with 101 patients receiving GA and 118 patients receiving CS. Our analysis showed that GA was associated with a lower average utility weighted mRS score at 90 days compared with CS (P≤0.02). Similarly, GA was associated with higher odds of worse outcomes on 90‐day mRS 0 to 2 (P<0.001), 90‐day mRS median score (P<0.001), a higher 24‐hour National Institutes of Health Stroke Scale score (P<0.001), and a lower change in baseline 24‐hour National Institutes of Health Stroke Scale score (P<0.001). A sensitivity analysis of patients with anterior stroke alone also favored better neurologic outcomes in the CS group compared with the GA group.

Conclusions

In patients undergoing endovascular thrombectomy in the MOST trial, GA was associated with poorer functional outcomes compared with CS.

Nonstandard Abbreviations and Acronyms

AIS

acute ischemic stroke

CS

conscious sedation

EVT

endovascular thrombectomy

GA

general anesthesia

MOST

Multi‐arm Optimization of Stroke Thrombolysis trial

mRS

modified Rankin Scale

NIHSS

National Institutes of Health Stroke Scale

TICI

thrombolysis in cerebral infarction

uwmRS

utility weighted modified Rankin Scale

Clinical Perspective.

What Is New?

• The ideal anesthetic technique that could improve neurologic outcomes in patients with acute ischemic stroke undergoing endovascular thrombectomy is still not known.

• This study examined the impact of general anesthesia versus conscious sedation on the outcomes of patients with acute ischemic stroke undergoing endovascular thrombectomy in the MOST (Multi‐arm Optimization of Stroke Thrombolysis) trial.

What Are the Clinical Implications?

• Our study shows that general anesthesia was associated with poor functional outcomes compared with conscious sedation during an endovascular thrombectomy procedure for patients with acute ischemic stroke.

Endovascular thrombectomy (EVT) is an important component of modern treatment for patients with acute ischemic stroke (AIS). Anesthetic management of patients with AIS undergoing thrombectomy can be performed under general anesthesia (GA) or conscious sedation (CS). Several observational studies have supported the use of procedural sedation for these patients and associated GA with poor functional outcomes, but these results were confounded by baseline patient conditions. ¹ Recently, several randomized controlled trials focused mostly on anterior circulation strokes have shown no difference between GA versus CS in functional outcomes at 3 months as measured by modified Rankin Scale (mRS) score. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ However, a recent meta‐analysis suggests that GA is associated with higher rates of successful recanalization and better functional independence when compared with other non‐GA techniques. ¹¹ Overall, the optimal sedative approach for EVT remains unclear. This article examines the impact of GA versus CS on the outcomes of patients with AIS undergoing mechanical thrombectomy in the MOST (Multi‐arm Optimization of Stroke Thrombolysis) trial. It is important to note that the primary objective of the MOST trial was to examine the effects of anticoagulants/antiplatelets as adjuncts to thrombolytics in improving functional outcomes in patients with AIS. The current report is a secondary analysis of the MOST trial to assess the impact of 2 different anesthetic techniques (GA versus CS) on the functional outcomes of patients with AIS undergoing mechanical thrombectomy.

METHODS

MOST was a phase 3 single‐blinded multicenter randomized controlled trial conducted across 57 sites in the United States. It was designed to examine the impact of argatroban or eptifibatide as adjunctive therapies to intravenous thrombolysis on improving functional outcomes after AIS. ¹² This post hoc analysis is specifically focused on evaluating the impact of GA versus CS on patients with AIS who underwent EVT in the MOST trial. All data and materials have been made publicly available at the National Institute of Neurological Disorders and Stroke database and can be accessed at https://www.ninds.nih.gov/current‐research/research‐funded‐ninds/clinical‐research/archived‐clinical‐research‐datasets. The protocol was approved by the University of Cincinnati Central Institutional Review Board (2018‐1464) and the Food and Drug Administration (IND 63550). All patients or their legally authorized representative provided written informed consent. Enrollment in the trial was open from October 10, 2019, to July 5, 2023. This study was registered on ClinicalTrials.gov (registration number: NCT03735979) on November 8, 2018.

Inclusion Criteria

Patients with the following criteria were included in the MOST trial: (1) >18 years old with stroke symptoms confirmed by emergency physicians or stroke neurologists, (2) received intravenous thrombolysis within 3 hours of stroke onset, (3) National Institutes of Health Stroke Scale (NIHSS) score ≥6, and (4) eligible to receive the investigational drugs within 75 minutes of thrombolysis. The comprehensive list of eligibility criteria was provided in the published protocol. ¹³ For the current analysis, we included only patients who underwent an EVT procedure.

Interventions

Patients in the study were randomly allocated to receive 1 of the following investigational drugs: argatroban, eptifibatide, or placebo. Patients received either a bolus (100 μg/kg) followed by a 12‐hour infusion (3 μg/kg per minute) of argatroban, or a bolus (135 μg/kg) followed by a 2‐hour infusion (0.75 μg/kg per minute) of eptifibatide and a 10‐hour saline infusion, or a 12‐hour saline infusion. Patients and their legal representatives were blinded to the treatment conditions during the entire trial. Patients who underwent mechanical thrombectomy had either GA or CS per local practice. The site‐level distributions of GA versus CS are included in Figure S1. Although the definitions of GA and CS have not been explicitly mentioned in the trial, GA is commonly defined as drug‐induced loss of consciousness where the patient is unresponsive to painful stimulation and requires assistance with airway, respiratory, and cardiovascular function. CS is defined as drug‐induced depression of consciousness, but the patient can respond to verbal or tactile stimuli and is able to maintain airway and cardiovascular function (https://www.asahq.org/standards‐and‐practice‐parameters/statement‐on‐continuum‐of‐depth‐of‐sedation‐definition‐of‐general‐anesthesia‐and‐levels‐of‐sedation‐analgesia). Also, the current report could not provide the anesthetic/sedation details used in the trial; however, GA typically requires endotracheal intubation followed by maintenance with inhalational or intravenous anesthetics. This could be achieved by ≥1 of the combinations of drugs such as volatile anesthetics (isoflurane, sevoflurane, desflurane), intravenous anesthetics (propofol, etomidate, benzodiazepines, barbiturates, ketamine, dexmedetomidine), muscle relaxants (succinylcholine or various nondepolarizing neuromuscular blockers), and opioids (remifentanil, fentanyl, sufentanil, and others). CS is usually achieved by a single or a combination of intravenous anesthetics, and opioids. Because the trial was not intended to randomize the anesthetic approach (GA versus CS), the assignment of anesthetic groups and the choice of anesthetics used were at the discretion of the anesthesiology/interventional team at the corresponding sites.

Outcomes

The primary outcome of the trial and this report was the utility weighted mRS (uwmRS) score at 90 days after stroke onset with scores ranging from 0 (worst) to 10 (best), measured between the GA and CS groups. The secondary outcomes were 90‐day mRS score, mRS score ≤1 at 90 days, mRS score ≤2 at 90 days, thrombolysis in cerebral infarction (TICI) 2B or better, 24‐hour NIHSS score, 24‐hour change in NIHSS score, and 90‐day mortality, assessed between GA and CS groups.

Statistical Analysis

The analysis aimed to determine whether the type of anesthesia used (GA versus CS) had a significant impact on the mean uwmRS score at 90 days while controlling for prespecified confounders such as age, sex, stroke scale, stroke onset to thrombolysis, stroke onset to groin puncture, and thrombolysis to groin puncture. To address potential confounding variables, propensity scores were derived using logistic regression models that predicted the choice of anesthesia based on these confounders. These scores were then used to create balanced pseudo‐data through the inverse probability of treatment weighting. For the primary outcome of uwmRS score, multiple linear regression was employed to estimate the effect of GA versus CS on the mean uwmRS score at 90 days. This approach provided both unadjusted and adjusted mean differences between the GA and CS groups. The unadjusted analysis took baseline confounders into account, and the adjusted analysis used inverse probability of treatment weighting to balance the groups. The 4 assumptions of a propensity score model were verified: (1) conditional independence (unconfoundedness), (2) positivity (overlap), (3) stable unit treatment value assumption, and (4) correct specification of the propensity score model. Please refer to Table S1 for detailed results of the assumption testing.

For secondary outcomes, logistic regression was applied to examine mRS scores of ≤1 and ≤2 at 90 days. A proportional odds logistic regression model was used to analyze mRS score at 90 days. Additionally, logistic regression was used to examine changes in the NIHSS score over 24 hours from baseline, assess 90‐day mortality, and evaluate the achievement of a TICI score of 2B/2C/3. For each of these outcomes, both unadjusted and adjusted analyses were performed. Unadjusted and adjusted odds ratios (ORs) were calculated for mRS scores of ≤2, ≤1, and the overall 90‐day mRS score. Further, for the post‐EVT modified TICI score, unadjusted and adjusted ORs were reported for achieving TICI scores of 2B/2C/3. All ORs represented the likelihood of achieving a lower mRS score (better) outcome.

Sensitivity analyses were performed for each primary and secondary outcome, stratified by anterior stroke location and among patients in the placebo group. These analyses provided insights into the consistency of the treatment effect across key subgroups. To address missing data for the primary outcome, Bayesian multiple imputation was employed, using baseline NIHSS score and the 30‐day mRS score. If neither the 30‐day nor the 90‐day mRS score was available, the subject’s 30‐day mRS score was imputed using a hot deck imputation method. This method involved randomly selecting an observed outcome from a pool of matched subjects based on treatment group, age, baseline NIHSS score, time to symptom onset, EVT received, and enrollment timing.

RESULTS

Demographics and Baseline Characteristics

This trial was performed between October 15, 2019, to July 1, 2023, with a total of 514 patients randomized. Of 514 patients, 219 underwent mechanical thrombectomy receiving either GA or CS and this cohort is included in this analysis. In the GA group, 23 patients arrived intubated to the neurointerventional suite and hence received GA for the EVT procedure; those patients were excluded from the analysis because intubation was indicated for their poor medical status and not for thrombectomy. The baseline and demographic details are presented in Table 1. Of 196 patients included in the final analysis, 78 patients received GA, and 118 patients received CS. A total of 93 patients were male. The median age was 70.5 in the GA group compared with 66.5 in the CS group. Patients in the GA group received more recombinant tissue plasminogen activator treatment than tenecteplase, had a higher percentage of Hispanic and Black individuals, fewer received study drug argatroban, had increased use of aspirin, and had higher rates of diabetes. There were no statistically significant differences found between the groups with respect to the stroke onset to thrombolysis initiation, stroke onset to randomization, stroke onset to study drug bolus, thrombolysis initiation to study drug bolus, baseline NIHSS score, baseline partial thromboplastin time, and Alberta Stroke Program Early CT [Computed Tomography] Score.

Table 1.

Patient Demographics/Baseline Characteristics

Characteristic	General anesthesia*	Conscious sedation	P value†
No	78	118
Age, y, median (IQR)	70.5 (60–81)	66.5 (58–80)	0.29
Sex, n (%)			0.22
Male	35 (44.9%)	58 (49.1%)
Female	43 (55.1%)	60 (50.8%)
Race, n (%)			0.70
Black	23 (29.5%)	15 (12.7%)
White	51 (65.4%)	99 (83.9%)
Other, unknown, or multiple	4 (5.1%)	4 (3.4%)
Hispanic or Latino ethnicity, n (%)	11 (14.1%)	6 (5.1%)	<0.001
Intravenous thrombolysis type, n (%)			0.001
Recombinant tissue plasminogen activator	55 (70.5%)	67 (56.8%)
Tenecteplase	23 (29.5%)	51 (43.2%)
Stroke onset to thrombolysis initiation, min, mean±SD	96.4±34.0	98.9±31.8	0.57
Stroke onset to randomization, min, mean±SD	135.5±35.8	134.2±35.3	0.82
Stroke onset to study drug bolus, min, mean±SD	162.5±33.8	162.8±37.1	0.89
Thrombolysis initiation to study drug bolus, min, mean±SD	67.6±16.8	63.8±19.9	0.21
Study drug type, n (%)			0.03
Argatroban	6 (7.7%)	19 (16.1%)
Eptifibatide	39 (50%)	50 (42.4%)
Placebo	33 (42.3%)	49 (41.5%)
Baseline National Institutes of Health Stroke Scale score, median (IQR)	17.5 (13–22)	16 (12–20)	0.15
Baseline partial thromboplastin time, median (IQR)	26.9 (24.7–30)	26 (24.6–29)	0.07
Prior use of, n (%)
Aspirin	34 (43.6%)	32 (27.1%)	<0.001
Clopidogrel	6 (7.7%)	6 (5.1%)	0.09
Warfarin	0 (0%)	1 (0.8%)	0.57
Other antiplatelet	0 (0%)	3 (2.5%)	0.19
Other anticoagulant	2 (2.6%)	2 (1.7%)	0.65
Statins	30 (38.5%)	41 (34.7%)	0.26
Diabetic medications	12 (15.3%)	11 (9.3%)	0.03
Antihypertensive	43 (55.1%)	63 (53.4%)	0.65
Past medical history, n (%)
Myocardial infarction	5 (6.4%)	12 (10.2%)	0.32
Atrial fibrillation	16 (20.5%)	35 (29.7%)	0.05‡
Hypertension	53 (67.9%)	79 (66.9%)	0.68
Diabetes	25 (32.1%)	25 (21.9%)	0.003
Prior stroke	13 (16.7%)	17 (14.4%)	0.36
Hypercholesterolemia	25 (32.1%)	52 (44.1%)	0.01
Alberta Stroke Program Early CT Score of baseline CT, median (IQR)	9 (8–10)	9 (8–10)	0.98
Historical, prestroke modified Rankin Scale score, median (IQR)	0 (0–1)	0 (0–0)	0.29

CT indicates computed tomography; and IQR, interquartile range.

^*Omits those who received general anesthesia due to preprocedural airway or movement concerns.

^† For binomial outcomes, the P value of a weighted χ² test if all cells in the 2×2 frequency table are at least 5. Otherwise, the P value is from an unweighted exact Fisher or Bernard test. For nonbinary categorical outcomes, the P value of a Mantel–Haenszel χ² test. For continuous outcomes, the P value of a weighted 2‐sample t test. Weights are based on propensity scores.

^‡ This P value is rounded up. The actual P value is <0.05.

Procedural Factors

The procedural details are presented in Table 2. Out of 118 patients in the CS group, 93 patients had anterior circulation stroke, 1 patient had posterior circulation stroke, 18 patients had both, and the location was not clear in 6 patients. In the GA group, 63 patients had anterior circulation stroke, 2 patients had posterior circulation stroke, 8 patients had both, and the location was unclear in 5 patients. The baseline prethrombectomy modified TICI score was different between the 2 groups with more patients having grade 0 in the CS group and grade 1 in the GA group. Also, the number of patients with thrombus at proximal internal carotid artery was higher in the GA group compared with the CS group and there were more patients with thrombus at M2 branch in the CS group compared with the GA group. There were no statistically significant differences found between the 2 groups with respect to time from thrombolysis to groin puncture, time from study drug to groin puncture, time from groin puncture to time of final reperfusion, post‐EVT modified TICI score, intraprocedural approach, complications, and the intravenous thrombolytics used.

Table 2.

Procedural Factors

Characteristic	GA (N=78)	CS (N=118)	P value‡
Time from stroke onset to groin puncture, min, mean±SD	151.6±48.1	162±154.2	0.43
Time from stroke onset to final reperfusion, min, mean±SD	190.5±60.8	198.8±159.1	0.51
Time from tissue plasminogen activator to groin puncture, min, mean±SD*	55.2±40.0	63±153.2	0.50
Time from study drug to groin puncture, min, mean±SD*	−12±41.3	0.8±159.5	0.40
Time from groin puncture to time of final reperfusion, min, mean±SD*	41.1±36.4	35.4±25.9	0.18
Side of lesion,† n (%)			0.80
Left	39/76 (51.3%)	58/116 (50%)
Right	37/76 (48.7%)	58/116 (50%)
Baseline prethrombectomy modified TICI score, n (%)			0.02
0	56 (71.8%)	102 (86.4%)
1	13 (16.7%)	5 (4.2%)
2A	3 (3.8%)	4 (3.4%)
2B	6 (7.7%)	7 (5.9%)
Post‐EVT modified TICI score, n (%)			0.78
1	2/72 (2.8%)	1/113 (0.9%)
2A	9/72 (12.5%)	11/113 (9.7%)
2B	38/72 (52.8%)	67/113 (59.3%)
2C	19/72 (12.5%)	29/113 (25.7%)
3	4/72 (5.6%)	5/113 (4.4%)
Significant stenosis or occlusion of the ipsilateral proximal cervical ICA or vertebral artery for basilar thrombosis (>70%), n (%)	14 (17.9%)	16 (13.6%)	0.08
Location of thrombus, n (%)
Proximal ICA	17 (21.8%)	18 (15.2%)	0.04
Distal ICA/T lesion	18 (23.1%)	21 (17.8%)	0.08
Basilar	1 (1.3%)	5 (4.2%)	0.41
M1	51 (65.3%)	75 (63.6%)	0.59
M2	19 (24.3%)	48 (40.7%)	<0.001
M3	9 (11.5%)	14 (11.9%)	0.99
Anterior cerebral artery	2 (2.6%)	9 (7.6%)	0.2
Posterior cerebral artery	2 (2.6%)	2 (1.7%)	0.65
Stroke location, n (%)			0.18
Anterior	63/73 (86.3%)	93/112 (83%)
Posterior	2/73 (2.7%)	1/112 (0.9%)
Both	8/73 (11%)	18/112 (16.1%)
Procedural details
Primary approach, n (%)			0.41
Stent retriever with aspiration	43 (55.1%)	70 (59.3%)
Stent retriever alone	7 (9.0%)	8 (6.8%)
Aspiration alone	28 (35.9%)	40 (33.9%)
Carotid stent placed, n (%)	3 (3.8%)	3 (2.5%)	0.68
Complications
Vessel perforation, n (%)	0 (0%)	2 (1.7%)	0.34
Arterial dissection, n (%)	2 (2.6%)	0 (0%)	0.09
Embolization to previously unaffected territory, n (%)	1 (1.3%)	2 (1.7%)	0.99
Access site complication requiring surgical repair or blood transfusion, n (%)	1 (1.3%)	2 (1.7%)	0.99
Intraprocedural meds
Intra‐arterial thrombolytics, n (%)	2 (2.6%)	0 (0%)	0.09
Other antithrombotics other than heparinized catheters, n (%)	5 (6.4%)	1 (0.8%)	0.04

CS indicates conscious sedation; EVT, endovascular thrombectomy; GA, general anesthesia; ICA, internal carotid artery; and TICI, thrombolysis in cerebral infarction.

^*Negative numbers indicate the former occurred after the latter.

^† Among those whose primary symptomatic occlusive lesion is not basilar.

^‡ For binomial outcomes, the P value of a weighted χ² test if all cells in the 2×2 frequency table are set at least 5. Otherwise, the P value is from an unweighted exact Fisher or Bernard test. For nonbinary categorical outcomes, the P value of a Mantel–Haenszel χ² test. For continuous outcomes, the P value of a weighted 2‐sample t test. Weights are based on propensity scores.

Primary and Secondary End Points

The primary and secondary outcomes details are presented in Tables 3 and 4. The mean±SD 90‐day uwmRS score for the CS group was 6.6±3.2 compared with 5.4±3.5 in the GA group with a P value of 0.02. There were more patients in the CS group with a 90‐day mRS score 0 to 2, compared with the GA group (OR, 2.88 [95% CI, 1.93–4.28], P<0.001). The median score for 90‐day mRS (OR, 1.96 [95% CI, 1.42–2.71], P<0.001) and the 24‐hour NIHSS score (OR, 2.23 [95% CI, 1.62–3.07], P<0.001) were lower in the CS group compared with the GA group. The 24‐hour NIHSS score change from baseline (negative direction) was higher in the CS group (OR, 2.33 [95% CI, 1.7–3.21], P<0.001). There was no statistically significant difference noted between the groups with respect to the mortality at 90 days, 90‐day mRS score 0 to 1, and TICI 2B or greater.

Table 3.

Ninety‐Day mRS Score Analysis Summary

Clinical efficacy	GA	CS	Unadjusted odds ratio (95% CI)	P value§	Adjusted odds ratio (95% CI)*	P value§,*
	No. (%)	No. (%)
90‐d mRS score 0–1,† n (%)	22 (28.2%)	43 (36.4%)	1.24 (0.64 to 2.39)	0.52	1.35 (0.9 to 2.03)	0.15
90‐d mRS score 0–2,† n (%)	31 (39.7%)	74 (62.7%)	2.4 (1.28to 4.51)	0.006	2.88 (1.93 to 4.28)	<0.001
90‐d mRS score, median (interquartile range)‡	3 (1 to 4)	2 (1 to 3)	1.7 (1.01 to 2.85)	0.04	1.96 (1.42 to 2.71)	<0.001

			Unadjusted mean difference (95% CI)	P value||	Adjusted mean difference (95% CI)*	P value||,*
Utility weighted 90‐d mRS score, mean±SD	5.4±3.5	6.6±3.2	−0.9 (−1.7 to 0)	0.06	−1.1 (−1.9 to −0.2)	0.02

CS indicates continuous sedation; GA, general anesthesia; and mRS, modified Rankin Scale.

^*Adjusted for propensity score weights.

^† Including those that returned to their prestroke mRS score.

^‡ The odds ratio for 90‐day mRS score is the odds of getting a lower (better) score.

^§ The P value of a χ² test.

^|| The P value of a t test.

Table 4.

Other Outcomes Analysis Summary

Outcome	GA (N=78)	CS (N=118)	Unadjusted odds ratio (95% CI)	P value†	Adjusted odds ratio (95% CI)	P value†
Post‐EVT modified thrombolysis in cerebral infarction score 2B or better	61/72 (84.7%)	101/113 (89.4%)	1.11 (0.41 to 3.02)	0.84	0.98 (0.51 to 1.89)	0.96
24‐h NIHSS score, median (IQR)*	9 (2 to 18)	4.5 (1 to 11)	2.14 (1.28 to 3.56)	0.004	2.23 (1.62 to 3.07)	<0.001
24‐h change in NIHSS score, median (IQR)*	−5 (−12 to 1)	−10 (−14 to −5)	2.22 (1.33 to 3.7)	0.002	2.33 (1.7 to 3.21)	<0.001
90‐d mortality, n (%)	12 (15.4%)	13 (11%)	1 (0.38 to 2.68)	0.99	0.97 (0.5 to 1.86)	0.92

CS indicates conscious sedation; EVT, endovascular thrombectomy; GA, general anesthesia; IQR, interquartile range; and NIHSS, National Institutes of Health Stroke Scale.

^*The odds ratio for 24‐hour NIHSS score is based on the odds of getting a lower (better) score.

^† The P value from a χ² test.

Sensitivity Analysis for Patients With Anterior Stroke Location

A sensitivity analysis for patients who had only anterior stroke is presented in Table 5. The mean±SD 90‐day uwmRS score for the CS group was 6.9±3.0, and 5.2±3.5 in the GA group with a P value of 0.02. The number of patients with 90‐day mRS score 0 to 1 (OR 2.27 [95% CI, 1.41–3.67], P<0.001) and 90‐day mRS score 0 to 2 (OR 4.45 [95% CI, 2.81–7.06], P<0.001) was higher in the CS group compared with the GA group. The median score for 90‐day mRS (OR 3.1 [95% CI, 2.13–4.52], P<0.001) and the 24‐hour NIHSS score (OR 3.13 [95% CI, 2.17–4.53], P<0.001) were lower in the CS group compared with the GA group. The change in the baseline NIHSS score in the negative direction at 24 hours was higher in the CS group (OR 3.51 [95% CI, 2.43–5.09], P<0.001). There was no statistically significant difference noted between the groups with respect to the mortality at 90 days and TICI 2B or greater.

Table 5.

Sensitivity Analysis Based on Only Anterior Stroke Location for All These Outcomes

Outcome	GA (N=63)	CS (N=93)	Unadjusted odds ratio (95% CI)	P value||	Adjusted odds ratio (95% CI)	P value||
90‐d mRS score 0–1*	15 (23.8%)	38 (40.9%)	2.06 (0.97 to 4.36)	0.06	2.27 (1.41 to 3.67)	<0.001
90‐d mRS score 0–2*	23 (36.5%)	63 (67.7%)	3.52 (1.73 to 7.20)	<0.001	4.45 (2.81 to 7.06)	<0.001
90‐d mRS score, median (IQR)†	3 (2 to 5)	2 (1 to 3)	2.5 (1.39 to 4.5)	0.002	3.1 (2.13 to 4.52)	<0.001
Post‐EVT modified TICI score of 2B or better, n (%)§	49/58 (84.5%)	80/90 (88.9%)	1.24 (0.43 to 3.61)	0.69	1.08 (0.53 to 2.17)	0.84
24‐h NIHSS score, median (IQR)	10 (3 to 18)	4 (1 to 9)	2.75 (1.54 to 4.92)	<0.001	3.13 (2.17 to 4.53)	<0.001
24‐h change in NIHSS score, median (IQR)‡	−5 (−12 to 2)	−10 (−14 to −5)	3.08 (1.72 to 5.51)	<0.001	3.51 (2.43 to 5.09)	<0.001
90‐d mortality, n (%)	10 (15.9%)	9 (9.7%)	0.69 (0.23 to 2.09)	0.51	0.63 (0.31 to 1.32)	0.22

			Unadjusted mean difference (95% CI)	P value¶	Adjusted mean difference (95% CI)	P value¶
Utility weighted 90‐d mRS score, mean±SD	5.2±3.5	6.9±3.0	−0.9 (−1.7 to 0.02)	0.06	−1.1 (−1.9 to −0.2)	0.02

CS indicates conscious sedation; EVT, endovascular thrombectomy; GA, general anesthesia; IQR, interquartile range; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; and TICI, thrombolysis in cerebral infarction.

^*Including those that returned to their prestroke mRS score.

^† The odds ratio is based on the odds of getting a lower (better) score.

^‡ This P value is rounded up and is actually <0.05.

^§ 5 subjects from the GA and CS groups each had either a missing or unknown post‐EVT TICI score.

^|| The P value of a χ² test. The adjusted P value factors in weights based on propensity scores.

^¶ The P value of a t test. The adjusted P value factors in weights based on propensity scores.

Sensitivity Analysis for Placebo Patients

A sensitivity analysis for patients in the placebo group is presented in Table 6. There were more patients with a 90‐day mRS score of 0 to 2 (OR, 3.91 [95% CI, CI 1.96–7.81], P<0.001) and a better 24‐hour change in NIHSS score (OR, 2.3 [95% CI, CI 1.38–3.81], P=0.001) in the CS group compared with the GA group. A lower 90‐day mRS median score (OR, 2.08 [95% CI, CI 1.24–3.48], P=0.005) and a lower 24‐hour NIHSS score (OR, 2.3 [95% CI, CI 1.38–3.81], P=0.001) were noted in the CS group compared with the GA group. There was no statistically significant difference noted between the groups with respect to the 90‐day uwmRS score, 90‐day mRS score 0 to 1, post‐EVT modified TICI score of 2B or better, and the 90‐day mortality.

Table 6.

Sensitivity Analysis Based on Placebo Group Patients for All Outcomes

Outcome	GA (N=33)	CS (N=49)	Unadjusted odds ratio (95% CI)	P value||	Adjusted odds ratio (95% CI)*	P value||, *
90‐d mRS score 0–1,† n (%)	12 (36.4%)	16 (32.7%)	0.77 (0.27 to 2.19)	0.63	0.84 (0.44 to 1.63)	0.61
90‐d mRS score 0–2,† n (%)	14 (42.4%)	33 (67.3%)	3.54 (1.2 to 10.42)	0.02	3.91 (1.96 to 7.81)	<0.001
90‐d mRS score, median (IQR)‡	3 (1 to 4)	2 (1 to 3)	1.68 (0.75 to 3.74)	0.21	2.08 (1.24 to 3.48)	0.005
Post‐EVT modified TICI score of 2B or better, n (%)§	30 (90.9%)	41/48 (85.4%)	0.32 (0.04 to 2.89)	0.31	0.32 (0.08 to 1.3)	0.11
24‐h NIHSS score, median (IQR)‡	10 (1 to 18)	5 (1 to 11)	2.21 (1 to 4.9)	0.05#	2.3 (1.38 to 3.81)	0.001
24‐h change in NIHSS score, median (IQR)‡	−6 (−11 to 0)	−10 (−15 to −5)	2.43 (1.1 to 5.34)	0.03	2.78 (1.68 to 4.6)	<0.001
90‐d mortality, n (%)	5 (15.1%)	5 (10.2%)	1.48 (0.25 to 8.87)	0.67	1.59 (0.5 to 5.08)	0.44

			Unadjusted mean difference (SE)	P value¶	Adjusted mean difference (SE)	P value¶
Utility weighted 90‐d mRS score, mean±SD	5.7±3.4	6.5±3.3	−0.6 (−1.9 to 0.7)	0.35	−0.7 (−1.9 to 0.6)	0.30

CS indicates conscious sedation; EVT, endovascular thrombectomy; GA, general anesthesia; IQR, interquartile range; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; and TICI, thrombolysis in cerebral infarction.

^*Adjusted for propensity scores.

^† Including those that returned to their prestroke mRS score.

^‡ The odds ratio is based on the odds of getting a lower (better) score.

^§ 3 subjects in the CS group have either missing or unknown post‐EVT modified TICI score.

^|| The P value of a χ² test. The adjusted P value factors in weights based on propensity scores.

^¶ The P value of a t test. The adjusted P value factors in weights based on propensity scores.

^# This P value is rounded down. The exact P value is >0.05.

DISCUSSION

We found that GA was associated with poorer functional outcomes in patients with AIS undergoing mechanical thrombectomy compared with CS. These findings are consistent with the results of many observational studies in the literature, ¹ but are contrary to findings from several randomized controlled trials, ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ where there was either no difference between GA and CS or a beneficial impact of GA in patients with AIS undergoing EVT.

EVT has revolutionized the treatment of patients with AIS and anesthesia has become an integral part of these procedures. EVT procedures can be done under GA, CS, or just under local anesthesia, though it is more commonly performed under GA or CS. The choice of anesthetic technique (GA versus CS) for the EVT procedure depends on multiple factors and GA is preferred for several reasons such as (1) patients inability to tolerate conscious sedation due to their medical complexity, (2) severity of the stroke requiring endotracheal intubation to protect the airway, and (3) interventionalist’s/anesthesiologist’s preference to avoid patient movement and complications such as vessel perforation, and arterial dissection. On the other hand, CS is employed for various reasons, such as (1) to avoid delays in securing the airway for the procedure to commence, (2) avoidance of hemodynamic fluctuations associated with GA, and importantly (3) the ability to perform neurological examination throughout the procedure. The selection of the anesthetic technique could also be influenced by other factors such as local culture or practices. In the MOST trial, some centers primarily used either GA or CS, suggesting local practice informed sedation choice over patient factors (Figure S1). This further supports the need for definitive trials to inform best practice.

Blood Pressure, Carbon Dioxide, and AIS Outcomes

One of several postulated reasons that GA has been associated with poor neurologic outcomes is hypotension. ¹ , ¹⁴ Though controversial, a few cohort studies have shown intraprocedural hypotension as a predictor of poor neurologic outcomes in patients with stroke. ¹ , ¹⁴ , ¹⁵ Although the recent stroke guidelines recommend managing systolic blood pressure between 140 and 180 mm Hg for the EVT procedure, the optimal blood pressure values at which neurologic symptoms are improved in patients with AIS are not known. ¹⁶ Similarly, hypocapnia is also shown to be a predictor of poor neurologic outcomes, and higher end‐tidal carbon dioxide concentration was associated with good neurologic outcomes after AIS. ¹⁷ , ¹⁸ The MOST trial was primarily designed to examine the impact of adjunctive antithrombotic therapies to intravenous thrombolysis and was not intended to examine the impact of anesthetics on patients with AIS. So, this study lacks the intraprocedural physiological details, and it is possible that hypotension, and hypocapnia associated with GA could have resulted in the poor outcomes seen in our study. Future studies aimed at identifying the optimal blood pressure and end‐tidal carbon dioxide targets for patients with AIS are warranted.

Anesthetics and AIS Outcomes

Several experimental studies have supported the role of anesthetic‐induced neuroprotection in AIS models. ¹⁹ A very large observational study noted that patients undergoing general anesthesia with volatile anesthetics for noncardiac surgery had a dose‐dependent protective effect on the incidence and severity of early postoperative IS. ²⁰ In a smaller cohort of patients with AIS undergoing mechanical thrombectomy, Sivasankar et al. showed that patients treated with volatile anesthetics were associated with better neurologic outcomes compared with the patients who received CS, total intravenous anesthesia with propofol, or combined anesthetics (volatile anesthetics+total intravenous anesthesia). ²¹ These findings are in accordance with randomized controlled trials where GA has been associated with favorable outcomes in patients with AIS. However, it is important to note that the majority of randomized controlled trials have used propofol as a primary anesthetic agent for maintenance, and the differential impact of other general anesthetic agents on patients with AIS is not known yet. The optimal anesthetic agent that could provide the maximal benefit in patients with AIS is a matter of future investigation. Unfortunately, the type of anesthetics and the doses used were not recorded in the MOST trial and hence commenting on the impact of the individual anesthetics on the outcomes is not possible in the current study.

Location of Thrombus and AIS Outcomes

It is well known that large vessel occlusion strokes are more severe and result in worse outcomes compared with non‐large vessel occlusion strokes. ²² , ²³ In this study, the GA group had significantly higher internal carotid artery occlusions compared with the CS group, and this could possibly explain the observed poor outcomes in the GA group.

Sensitivity Analysis of Patients With AIS and Anterior Circulation Stroke

Posterior circulation stroke may carry a poorer prognosis compared with the anterior circulation stroke. ²⁴ We performed a sensitivity analysis for the patients with AIS presenting only with the anterior circulation stroke and found that patients who received GA for anterior circulation stroke still had poorer neurologic outcomes as measured by mRS and NIHSS, compared with the patients who received CS for their EVT procedure.

Sensitivity Analysis of Patients With AIS in the Placebo Group

MOST found significantly increased mortality in patients who received argatroban and a nonsignificant increase in those who received eptifibatide compared with the placebo group. ¹² To avoid the confounding effects of the study drugs on the current results, we performed a sensitivity analysis including only the patients in the placebo group. In our subanalysis, we found that there were no statistically significant differences in uwmRS score, 90‐day mRS score 0 to 1, post‐EVT modified TICI score of 2B, and mortality at 90 days between the GA and CS groups, though the 90‐day score mRS 0 to 2, mRS median score, 24‐hour NIHSS score, and 24‐hour change in NIHSS score remained significant indicating that the CS group tended to do better than the GA group.

STRENGTHS AND LIMITATIONS

The strength of the study is that the data used for the analysis is from a multicentered randomized controlled trial, though we recognize that randomization in the study is applicable only to the treatments but not the anesthetic techniques. The study has some limitations. (1) This study is not specifically designed to investigate the impact of GA versus CS, and hence, the necessary physiologic and anesthetic details are lacking, and the results should be interpreted with caution. (2) Patients in the GA group had a higher rate of diabetes, internal carotid artery occlusive disease, and higher use of antiplatelet agents (for a probable atherosclerotic cardiovascular disease), which are known to be associated with poorer outcomes. We did not adjust for these parameters in our analysis, and this could have perhaps led to the observed poor outcomes in the GA group in our study. (3) It is possible that the sample size in the current study is not adequate to examine the impact of anesthetics on the assessed neurologic outcomes.

CONCLUSIONS

In patients with AIS, we found that treatment with GA for mechanical thrombectomy was associated with poorer functional outcomes compared with CS. Though this finding is consistent with several observational studies published in the literature, which anesthetic technique should be used in patients with AIS undergoing EVT to improve neurologic outcomes remains an unanswered question. Future large, randomized studies specifically focusing on identifying the optimal blood pressure, end‐tidal carbon dioxide, and the anesthetic technique/agent to improve neurologic outcomes in patients with AIS are warranted.

Sources of Funding

Dr Athiraman reports funding from the National Institutes of Health/National Institute of Neurological Disorders and Stroke (K08NS125038) and the Brain Aneurysm Foundation.

Disclosures

None.

Supporting information

Table S1

Figure S1

Data S1–S2