Introduction

Diabetes mellitus (DM) is a global healthcare burden and is significantly associated with higher rates of cardiovascular morbidity and mortality [1]. Among acute stroke patients, ~ 20% are newly diagnosed whereas 60% have a known diagnosis of DM or impaired glucose tolerance [2]. Admission hyperglycemia tends to increase tissue lactate levels and infarct volume, reduce penumbral tissue and portend worse clinical outcomes even among stroke patients treated with acute reperfusion therapies [3, 4]. Consequently, health regulatory authorities have mandated cardiovascular safety assessments of new DM medications [5].

Glucagon-like peptide-1 receptor (GLP-1R) agonists are cost-effective newer antihyperglycemic medications that decrease HbA1c levels by reducing the glucagon secretion, gastric emptying, and appetite [6]. These medications have pleiotropic effects in controlling cholesterol levels, blood pressure and body weight along with lowering the risk for hypoglycemia. Randomized-controlled clinical trials (RCTs) demonstrated a variable benefit of GLP-1R agonists for major adverse cardiovascular (MACE) events in patients with type 2 DM and cardiovascular risk factors, however, their potential benefit on stroke outcomes remain unidentified. We performed a systematic review and meta-analysis to evaluate the effect of GLP-1R agonists on stroke outcomes among type 2 DM patients.

Methods

Our manuscript is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [7]. We used publicly available published studies, and our study was exempt for approval from the Institutional Review Board. Our study protocol for inclusion and exclusion criteria was designed a priori, and the protocol of the present study has been registered to PROSPERO database (CRD42020150260). The authors declare that all supporting data are available within the article and supplementary files.

Data sources and searches

Eligible studies were identified by systematically searching in PubMed (MEDLINE) and Scopus databases. The combination of search strings used to query all the databases included: “glucagon-like peptide-1 receptor agonist”, “exenatide”, “liraglutide”, “lixisenatide”, “semaglutide”, “dulaglutide”, “albiglutide”, “placebo”, and “stroke”. The search algorithm used for the MEDLINE database is available in Table e-1. We restricted our search to randomized clinical trials, while no language restrictions were applied for our database search spanning from database inception to October 15, 2019. An additional manual search of bibliographies of articles meeting study criteria for a comprehensive literature search was conducted.

Study selection and data extraction

We identified RCTs that provided data on the incidence of stroke outcomes among patients with type 2 DM treated with GLP-1R agonists vs placebo. The primary outcomes included the incidence rates of fatal, non-fatal, and all types of strokes. Secondary outcomes included the incidence rates of all-cause and cardiovascular mortality, fatal or nonfatal myocardial infarction (MI), and MACE. We additionally evaluated the recurrence rates of non-fatal strokes and MACE among the patients with prior history of MI or non-fatal strokes. These outcomes were decided a priori to focus and evaluate the effect of GLP-1R agonists on stroke and/or cardiovascular outcomes.

Per the study protocol, we excluded studies that reported (1) no stroke outcomes, (2) observational studies and (3) case reports, case series or conference abstracts. Reference lists of all articles that met the inclusion criteria and of relevant review articles were examined to identify studies that may have been missed by the initial database search. All retrieved studies were scanned independently by two reviewers (KM, AK), and in any case of disagreement, the third author (GT) was consulted to resolve any disagreements. We primarily documented the incidence rates of fatal and/or non-fatal strokes, all-cause and cardiovascular mortality, major adverse cardiovascular events (MACE) i.e. cardiovascular death, non-fatal myocardial infarction and non-fatal stroke. Additional analyses were conducted among patients with a prior history of MI or nonfatal strokes to assess for stroke and MACE recurrence. We also performed sensitivity analyses based on the frequency of administration and homology to native GLP-1 structure.

Risk of bias assessment

The risk of bias for each included study was assessed with the Cochrane Collaboration risk of bias tool independently by the two authors who performed the literature search (GT and AHK) and all emerging conflicts were resolved with consensus.

Data synthesis and statistical analysis

In the current meta-analysis, all the intended outcomes of interest were handled as dichotomous variables, and all the associations were reported as odds ratios (OR) evaluating the effect of GLP-1 analogs with different clinical outcomes. A random-effects model (DerSimonian-Laird) was used to calculate the pooled effect estimates. Heterogeneity between studies was assessed with the Cochran Q and I2 statistics, with I2 values of at least 50% considered to represent substantial heterogeneity and values of at least 75% indicative of considerable heterogeneity [8]. Publication bias was evaluated with a visual inspection of the funnel plot, whereas the Egger linear regression test was not performed because of the small number of included studies (n < 10).

We used the Cochrane risk of bias assessment to explore sources of bias in included RCTS [9]. All statistical analyses were carried out with Cochrane Collaboration’s Review Manager Software Package (RevMan 5.3) and the Comprehensive Meta-analysis version 2 software (Biostat, Englewood, NJ, USA, https://www.meta-analysis.com).

Data availability statement

Our study adheres to the AHA Journals' implementation of the Transparency and Openness Promotion (TOP) Guidelines. The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Results

Study characteristics and publication bias

The complete literature search is presented in Figure e-1. Seven studies were excluded (Table e-1) due to the non-availability of data and the absence of randomized design. Eight eligible RCTs [1, 10,11,12,13,14,15,16] comprising 56,251 patients met the inclusion criteria and their baseline characteristics are summarized in Table e-2. The risk of bias was considered low in all the included RCTs (Figures e-2 and e-3). No asymmetry was observed in the funnel plot for nonfatal stroke (Figure e-4) and all strokes (Figure e-5).

Primary (cerebrovascular) outcomes

Compared to the control group, GLP-1R agonists reduced the odds of nonfatal strokes (7 RCTs; OR 0.84; 95% CI 0.76–0.94, p = 0.002; p for Cochran Q statistic = 0.61, I2 = 0%; Fig. 1, Table 1), and all strokes (5 RCTs; OR 0.84; 95% CI 0.75–0.93, p = 0.001; p for Cochran Q statistic = 0.92, I2 = 0%; Fig. 2). No association was noted with fatal strokes (5 RCTs; OR 0.82; 95% CI 0.62–1.09, p = 0.17; p for Cochran Q statistic = 0.71, I2 = 0%; Figure e-6).

Fig. 1
figure 1

Forest plot comparing the risk of non-fatal strokes in GLP-1R agonists and placebo

Full size image
Table 1 Overview of cerebrovascular outcomes among patients treated with GLP-1R agonists
Full size table
Fig. 2
figure 2

Forest plot comparing the risk of all strokes in GLP-1R agonists and placebo

Full size image

Secondary outcomes

Treatment with GLP-1R agonists (vs placebo) was associated with reduction of MACE (7 RCTs; OR 0.87; 95% CI 0.81–0.94, p = 0.0003; p for Cochran Q statistic = 0.11, I2 = 42%; Figure e-7), all-cause mortality (7 RCTs; OR 0.88; 95% CI 0.82–0.95; p = 0.0007; p for Cochran Q statistic = 0.31, I2 = 15%; Figure e-8) and cardiovascular mortality (8 RCTs; OR 0.88; 95% CI 0.81–0.95; p = 0.002; p for Cochran Q statistic = 0.44, I2 = 0%; Figure e-9). However, no clear association of GLP-1R agonists was noted for either nonfatal MI (7 RCTs; OR 0.90; 95% CI 0.80–1.01, p = 0.08; p for Cochran Q statistic = 0.07, I2 = 49%; Figure e-10) or fatal MI (4 RCTs; OR 1.14; 95% CI 0.72–1.80; p = 0.58; p for Cochran Q statistic = 0.11, I2 = 50%; Figure e-11).

Analyses for secondary prevention

Among patients with prior history of MI or nonfatal strokes, GLP-1R agonists were associated with reduced incidence of recurrent MACE (5 RCTs; OR 0.86; 95% CI 0.80–0.92; p < 0.0001; p for Cochran Q statistic = 0.72, I2 = 0%; Figure e-12), however, no association was noted for recurrent nonfatal strokes (2 RCTs; OR 0.87; 95% CI 0.67–1.15; p = 0.33; p for Cochran Q statistic = 0.32, I2 = 0%; Figure e-13).

Sensitivity analyses

We performed sensitivity analyses between GLP-1R agonists and placebo groups based on the frequency of administration and homology to native GLP-1 structure. No subgroup differences were noted for weekly or daily administration of GLP-1R agonists and nonfatal (p = 0.21; Figure e-14), fatal (p = 0.48; Figure e-15) or all strokes (p = 0.68; Figure e-16). Similarly, after segregating RCTs that used medications homologous to GLP-1 structure (vs no homology), no significant differences were observed for nonfatal (p = 0.28; Figure e-17), fatal (p = 0.77; Figure e-18) and all strokes (p = 0.76; Figure e-19).

Discussion

Our study demonstrates that the use of GLP-1R agonists among patients with type 2 DM is associated with a reduced incidence of nonfatal strokes, all strokes and MACE. Additionally, these medications are associated with reduced cardiovascular and all-cause mortality. Treatment with GLP-1R agonists for secondary prevention was related only to MACE reduction and did not affect recurrent stroke.

Our results were consistent across all stroke and fatal outcomes among patients with type 2 DM treated with GLP-1R agonists. There was a greater benefit on stroke outcomes in comparison to cardiovascular outcomes across the majority of RCTs. Apart from the reduction of cardiovascular outcomes, GLP-1R agonists are well tolerated, have good adherence rates and are not associated with major adverse or hypoglycemic events. These benefits in primary stroke prevention can be attributed to the reduction of HbA1c, body weight, LDL cholesterol levels and systolic blood pressure. GLP-1R agonists also improve endothelial and platelet functions, provide neuroprotective benefit and attenuate atherosclerosis, vasoconstriction and vascular inflammation [17].

Our study findings are in line with a recent meta-analysis involving 5 RCTs that evaluated the effect of GLP-1R agonists on stroke [18]. Nevertheless, we included more studies in the present report (8 vs. 5), registered our meta-analysis protocol to PROSPERO database and performed a systematic evaluation of the risk of bias of included studies. Furthermore, we performed sensitivity and subgroup analyses providing robustness to our study results.

This study has certain limitations. First, this is an aggregate data meta-analysis that restricts further investigation of subgroups based on heterogeneity effect of individual GLP-1R agonists for stroke prevention, different follow-up time periods, impact of concomitant medications or duration of DM. Second, our results have limited generalizability as these associations were studied among type 2 DM patients with high cardiovascular risk factors. Third, a greater number of patients in the placebo group of the included RCTs received sodium-glucose cotransporter 2 inhibitor. As these drugs are cardioprotective [14], the observed treatment difference could potentially be skewed. Last, included studies do not provide information on stroke type, and thus we were unavailable to assess the association separately for ischemic and hemorrhagic strokes.

In conclusion, the present meta-analysis demonstrates the benefit of GLP-1R agonists for primary stroke prevention and the reduction of fatal cardiovascular outcomes. Individual patient data meta-analysis and RCTs are warranted to evaluate the benefit of these drugs for secondary stroke prevention.