Association between Glucagon-Like Peptide-1 Receptor Agonist Use and Ischemic Stroke Severity

Abstract

Background:

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) reduce stroke incidence, but their impact on the severity of acute ischemic stroke (AIS) when events occur remains unclear. The aim of this study is to evaluate whether pre-stroke GLP-1RA use is associated with reduced ischemic stroke severity and improved post-stroke outcomes.

Method:

We conducted a retrospective cohort study using the 2016-2024 data from the TriNetX US Collaborative Network. Patients were categorized based on documented use of GLP-1RAs for at least six months prior to the index stroke event. After, propensity score matching, primary outcome was measured by the National Institutes of Health Stroke Scale (NIHSS), with secondary outcomes including rates of severe stroke (NIHSS ≥10), intensive care unit admission, neurological symptoms, in-hospital complications, and 30-day mortality.

Results:

A total of 209,180 first-time AIS patients were identified, of whom 4,773 (2.3%) were GLP-1RA users. After PSM, 4,769 patients remained in each group. GLP-1RA users had lower median NIHSS (3 vs 4, p<0.001). GLP-1RA users also had lower rates of severe stroke (18.4% vs 25.5%, p<0.001), dysphagia/dysarthria (43.5% vs 48.0%, p<0.001), aphasia (25.2% vs 28.7%, p<0.001), hemiplegia/hemiparesis (37.6% vs 43.3%, p<0.001), intracranial hemorrhage (13.2% vs 14.7%, p=0.031), and 30-day mortality (6.3% vs 7.6%, p=0.016).

Conclusions:

Among adult patients who suffered a first-time AIS, pre-stroke use of GLP-1RA was associated with reduced stroke severity, lower rates of symptoms and complications, and lower mortality. These findings suggest that the benefits GLP-1RA may extend beyond preventing the occurrence of cerebrovascular events.

Introduction

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have become foundational therapies for type 2 diabetes and obesity, demonstrating cardiometabolic benefits beyond glycemic control^1–3. Large cardiovascular outcomes trials for GLP-1RAs have shown reductions in major adverse cardiovascular events, including ischemic stroke⁴. In the REWIND trial of type 2 diabetes mellitus patients with a history of cardiovascular disease or cardiovascular risk factors, dulaglutide was associated with a 25% reduction in ischemic stroke risk among GLP-1RA users⁵. In the SUSTAIN-6 trial, which enrolled a similar patient population as REWIND, semaglutide led to a 39% reduction in nonfatal stroke⁶. Finally, in a meta-analysis of 11 cardiovascular outcomes trials, GLP-1RA as a medication class was associated with a 16% reduction in overall stroke risk⁷. While GLP-1RA use among patients with diabetes likely reduces overall stroke risk, little is known about whether GLP-1RA exposure may influence the severity and outcomes of stroke if it occurs. In the REWIND trial, no significant differences in modified Rankin score among patients who suffered strokes were observed; however, as only 363 patients suffered strokes in the study, these results were limited by a lack of statistical power⁸. Recently, GLP-1 receptor agonists have also shown a beneficial effect on several other central nervous system disorders, such as subarachnoid hemorrhage, Alzheimer’s disease, Parkinson's disease, head injuries, and epilepsy^9–13

In this study, we aimed to compare stroke severity and outcomes in AIS patients with versus without pre-stroke GLP-1RA use. We hypothesize that GLP-1RA use would be associated with milder stroke presentations and lower risks of in-hospital complications and mortality.

Methods

Study Design and Data Source

This retrospective cohort study utilized the TriNetX US Collaborative Network, aggregating de-identified electronic health records from 69 healthcare organizations. Because data are deidentified, institutional review board approval was not required. Cohorts, outcomes, and covariables were defined using International Classification of Diseases, Tenth Revision (ICD-10) and RxNorm codes (Table S1)

Participants and Exposure

Adults (≥18 years) with acute ischemic stroke and available NIH stroke scale information were identified. Primary exposure was prescription of GLP-1 RAs (dulaglutide, liraglutide, semaglutide, tirzepatide, lixisenatide) documented before at least six months prior to the AIS event. Two cohorts were defined: patients with prior GLP-1 RA use versus unexposed controls.

Outcomes

Primary outcome was stroke severity measured by NIH Stroke Scale (NIHSS). Secondary outcomes included severe stroke (NIHSS ≥10), ICU admission, symptoms (gait instability, dysphagia/dysarthria, aphasia, visual disturbance, hemiplegia/hemiparesis), reperfusion treatments (IV thrombolysis, endovascular thrombectomy), complications (intracranial hemorrhage, endotracheal intubation, hemicraniectomy, gastrostomy, tracheostomy), palliative care consultation, and 30-day mortality.

Statistical Analysis

Propensity score matching (1:1) balanced demographics (age, sex, race/ethnicity), comorbidities (hypertension, diabetes, metabolic disorders, obesity, sleep disorders, mood/anxiety disorders, chronic kidney disease, coagulopathy, atrial fibrillation, cerebrovascular/cardiovascular conditions), and medications (antilipidemic, antithrombotic, diabetes, antihypertensive agents). Absolute standardized differences less than 0.10 indicated adequate balance. Outcomes were compared using the Mann-Whitney U test and chi-square tests; relative effect sizes for binary outcomes were calculated using odds ratios with 95% confidence intervals. Statistical significance was defined as a two-sided p<0.05. All statistical analyses were conducted on the TriNetX platform. Since NIHSS was pre-specified as the primary outcome of our study, other secondary outcomes may be prone to type 1 error due to multiple comparisons and should be considered exploratory in nature.

Results

Patient population

A total of 209,180 first-time AIS patients were identified, of whom 4,773 were GLP-1RA users for at least six months prior to AIS. Patient characteristics were significantly different across multiple characteristics (Table 1). A large majority of GLP-1RA users had baseline diabetes mellitus (94.0%), and many were also overweight or obese (54.6%). Overall, GLP-1RA users were significantly younger, less likely White and more likely Black, and more likely male (all p<0.05; Table 1). GLP-1RA users also had significantly higher rates of various comorbidities such as hypertension, diabetes, metabolic disorder, overweight or obesity, sleep disorders, mood disorders, anxiety disorders, chronic kidney diseases, carotid artery stenosis, heart failure, atherosclerosis, other peripheral artery disease, and chronic ischemic heart disease (all p<0.001), while they were less slightly less likely to have atrial fibrillation (25.4% vs. 28.5%, p<0.001), possibly due to younger age. Finally, GLP-1RA users were also significantly more likely to be on antilipidemic, antithrombotic, other diabetes medications, and blood pressure medications.

Table 1:

Patient characteristics

	Before Matching				After Matching
Characteristic - mean ± SD, % (n)	GLP-1RA users (N=4,773)	Control (N=204,407)	P-value	ASD	GLP-1RA users (N=4,769)	Control (N=4,769)	P-value	ASD
Demographics
Age (years)	64.8 ± 11.8	68.1 ± 14.5	<0.001	0.244	64.8 ± 11.8	65.0 ± 13.0	0.51	0.013
Race/ethnicity
White	70.0% (3,343)	71.4% (145,978)	0.038	0.03	70.0% (3,339)	68.5% (3,266)	0.11	0.033
Hispanic or Latino	7.6% (361)	7.0% (14,239)	0.11	0.023	7.6% (361)	8.3% (395)	0.20	0.026
Black	20.3% (967)	16.5% (33,713)	<0.001	0.097	20.3% (967)	21.1% (1,006)	0.32	0.02
Male	45.4% (2,169)	51.8% (105,975)	<0.001	0.128	45.5% (2,169)	45.5% (2,172)	0.95	0.001
Comorbidities
Hypertension	96.9% (4,623)	87.3% (178,452)	<0.001	0.359	96.9% (4,619)	96.6% (4,606)	0.46	0.015
Diabetes mellitus	94.0% (4,485)	39.1% (79,896)	<0.001	1.429	94.0% (4,481)	95.2% (4,539)	0.009	0.054
Metabolic disorder	95.1% (4,539)	80.9% (165,280)	<0.001	0.449	95.1% (4,535)	95.1% (4,537)	0.92	0.002
Overweight or obesity	57.6% (2,750)	22.9% (46,848)	<0.001	0.756	57.6% (2,746)	56.8% (2,708)	0.43	0.016
Sleep disorders	41.4% (1,977)	16.8% (34,290)	<0.001	0.564	41.4% (1,973)	40.3% (1,923)	0.30	0.021
Mood disorders	37.5% (1,792)	19.6% (39,974)	<0.001	0.406	37.5% (1,788)	36.7% (1,752)	0.45	0.016
Anxiety disorders	35.6% (1,697)	19.7% (40,363)	<0.001	0.359	35.5% (1,693)	33.6% (1,602)	0.050	0.04
Chronic kidney disease	39.0% (1,862)	21.5% (43,878)	<0.001	0.389	39.0% (1,860)	39.1% (1,865)	0.92	0.002
Coagulopathy	14.2% (678)	14.0% (28,545)	0.64	0.007	14.2% (678)	14.4% (689)	0.75	0.007
Atrial fibrillation	25.4% (1,213)	28.5% (58,281)	<0.001	0.07	25.4% (1,213)	25.9% (1,235)	0.61	0.011
Cerebral atherosclerosis	8.7% (415)	8.0% (16,351)	0.08	0.025	8.7% (415)	9.2% (439)	0.39	0.018
Carotid artery stenosis	27.1% (1,294)	23.7% (48,504)	<0.001	0.078	27.1% (1,291)	27.3% (1,301)	0.82	0.005
Heart failure	33.1% (1,581)	22.7% (46,500)	<0.001	0.233	33.1% (1,580)	33.6% (1,601)	0.65	0.009
Atherosclerosis	14.2% (678)	8.9% (18,188)	<0.001	0.167	14.2% (678)	13.8% (659)	0.58	0.011
Other peripheral artery disease	12.6% (600)	7.0% (14,362)	<0.001	0.187	12.6% (599)	12.0% (570)	0.37	0.019
Chronic ischemic heart disease	45.1% (2,151)	29.8% (60,894)	<0.001	0.32	45.0% (2,148)	45.1% (2,153)	0.92	0.002
Antilipidemic agent	80.7% (3,852)	58.7% (120,087)	<0.001	0.492	80.7% (3,848)	81.6% (3,893)	0.24	0.024
Antithrombotic medications
Aspirin	64.2% (3,062)	54.3% (111,034)	<0.001	0.201	64.2% (3,060)	63.9% (3,046)	0.77	0.006
Clopidogrel	30.3% (1,446)	19.6% (40,045)	<0.001	0.249	30.3% (1,445)	30.1% (1,434)	0.81	0.005
Ticagrelor	2.2% (104)	1.2% (2,402)	<0.001	0.078	2.2% (104)	2.3% (109)	0.73	0.007
Warfarin	3.5% (169)	3.1% (6,333)	0.08	0.025	3.5% (169)	3.7% (177)	0.66	0.009
Apixaban	11.5% (551)	7.4% (15,200)	<0.001	0.141	11.6% (551)	11.0% (525)	0.40	0.017
Rivaroxaban	3.6% (172)	2.3% (4,726)	<0.001	0.076	3.6% (172)	3.7% (175)	0.87	0.003
Other diabetes medications
Metformin	33.9% (1,616)	7.1% (14,506)	<0.001	0.703	33.9% (1,616)	34.7% (1,655)	0.40	0.017
Other oral agents	53.4% (2,550)	10.7% (21,895)	<0.001	1.029	53.4% (2,546)	53.5% (2,552)	0.90	0.003
Blood pressure medications
ACE-inihibor	31.3% (1,495)	18.0% (36,791)	<0.001	0.313	31.3% (1,495)	31.4% (1,496)	0.98	<0.001
ARB	33.5% (1,599)	13.8% (28,124)	<0.001	0.478	33.4% (1,595)	32.8% (1,563)	0.49	0.014
Diuretics	47.5% (2,268)	24.6% (50,209)	<0.001	0.492	47.5% (2,265)	47.3% (2,258)	0.89	0.003
Beta blocker	69.3% (3,309)	59.4% (121,376)	<0.001	0.209	69.3% (3,306)	68.4% (3,263)	0.34	0.019
Calcium channel blocker	44.8% (2,140)	36.6% (74,786)	<0.001	0.168	44.9% (2,139)	45.5% (2,169)	0.54	0.013
Other BPM	42.7% (2,040)	34.2% (70,000)	<0.001	0.175	42.8% (2,039)	42.3% (2,015)	0.62	0.01

After PSM, 4,769 patients remained in each arm. Characteristics were well balanced for all variables (all ASD < 0.10; Table 1).

Study outcomes

For the primary study outcome, compared to PSM controls, GLP-1RA users had lower median NIHSS scores (3 [IQR 1-7] vs 4 [IQR 1-8]), which was statistically significant (p<0.001). Secondary outcomes also generally favored GLP-1RA use (Figure 1). Severe stroke occurred less frequently among GLP-1RA users (18.4% vs 25.5%, p<0.001), and the proportion requiring ICU admission was lower (34.9% vs 39.1%, p<0.001). The presence of neurological symptoms was also less common among GLP-1RA users, including dysphagia/dysarthria (43.5% vs 48.0%), aphasia (25.2% vs 28.7%), and hemiplegia/hemiparesis (37.6% vs 43.3%). No differences in rates of gait instability and visual disturbance were observed.

Figure 1:

Secondary outcomes of GLP-1RA users versus PSM controls.

In terms of reperfusion treatments, GLP-1RA users were less likely to receive IV thrombolysis (10.5% vs 12.2%; OR 0.84, p=0.007) and thrombectomy (4.1% vs 5.5%; OR 0.73, p=0.001), which could be a reflection of the observed milder stroke severity, though other factors such as time to hospital arrival, time from stroke onset, and anticoagulant use may have also contributed to these differences. Finally, GLP-1RA use was associated with lower rates of post-stroke complications and adverse outcomes, including intracranial hemorrhage (13.2% vs 14.7%; OR 0.88, p=0.031), hemicraniectomy (0.6% vs 1.0%; OR 0.59, p=0.023), gastrostomy (2.7% vs 3.5%; OR 0.75, p=0.013), tracheostomy (0.8% vs 1.4%; OR 0.58, p=0.005), palliative care consultation (8.6% vs 11.1%; OR 0.76, p<0.001), and thirty-day mortality (6.3% versus 7.6%, OR 0.82, 95% CI 0.70-0.96, p=0.016).

Discussion

In this multicenter retrospective analysis of real-world data, we found that GLP-1RA use was associated with milder AIS symptoms and lower rates of adverse clinical outcomes compared with matched controls. These results extend prior evidence of GLP-1RA cardiovascular protection by suggesting that, beyond reducing stroke incidence, GLP-1RA exposure may also mitigate stroke severity when cerebrovascular events occur.

Previous clinical trials collectively found reductions in ischemic stroke risk among GLP-1RA users^5–7, an effect that is likely mediated by superior glycemic control offered by GLP-1RAs compared to legacy medications among patients with diabetes^8,14. The mechanisms underlying milder stroke presentations observed in our study, however, may be more complex. GLP-1 RAs have been shown to have anti-inflammatory effects, reduce oxidative stress and apoptosis, inhibit advanced glycation end-product formation, improve endothelial function, enhance cerebral blood flow and collateral circulation, and promote angiogenesis and neurogenesis, which could have collectively been associated with better cerebrovascular reserve prior to AIS, potentially making patients more resilient to acute cerebral ischemia^15,16. While confirmation of our findings in future prospective studies would be ideal, studying differences in stroke severity with GLP-1RA in a randomized trial is challenging as even large trials of GLP-1RA yield a limited number of stroke events (e.g., 363 patients in REWIND⁸), and patients cannot be prospectively assigned to have prior drug exposure at the time of stroke. Thus, large observational studies such as ours may be ideal for investigating stroke severity with GLP-1RA use. Nevertheless, validation with additional studies is needed to confirm our findings.

Limitations

Our study has several limitations. First, despite our best efforts to balance patient groups, hidden confounders maybe present, which may have biased our outcomes. Notably, time from last-known-well to hospital presentation is not available in the database, and it is possible that GLP-1RA users presented earlier in the treatment time window, thus leading to more favorable outcomes. However, given that utilization of acute reperfusion treatment with intravenous thrombolysis and endovascular thrombectomy was low overall, and even lower among GLP-1RA users, our conclusion that GLP-1RA use is associated with milder strokes is likely robust to potential biases that could arise from differences in time from last-known-well to stroke presentation. Additionally, socioeconomic information was not available, and it is possible that GLP-1RA users may have better access to care and thus overall better baseline health status. NIHSS missingness in electronic health records may be informative, and our study’s inclusion of only patients with available NIHSS may have also biased outcomes. Second, while we were able to identify patients who were prescribed GLP-1RA, whether they were compliant on the medication was not clear. Finally, as this study was based on administrative health records data, patient characteristics and outcomes are subject to reporting errors. Future prospective studies are needed to confirm our study findings.

Conclusions

For patients with first-time AIS, pre-stroke use of GLP-1RA was independently associated with lower stroke severity, lower rates of complications, and improved survival. These findings highlight potential neuroprotective properties and treatment benefits of GLP-1RAs that may extend beyond reducing the incidence of stroke. Prospective randomized studies are needed to confirm these preliminary observational results.

Highlights.

• GLP-1RA users had lower median NIHSS scores after propensity score matching

• Severe stroke rates were significantly lower in GLP-1RA users (18.4% vs 25.5%)

• GLP-1RA use linked to less post-stroke aphasia, hemiplegia, and dysphagia.

• 30-day mortality was lower among GLP-1RA users (6.3% vs 7.6%)

• 1GLP-1RA use may have a neuroprotective effect on AIS severity

DATA AVAILABILITY STATEMENT

Study data are available upon reasonable request from the corresponding author.