Cardiovascular benefits of semaglutide: a systematic review and meta-analysis of randomized controlled trials

Abstract

Background

Semaglutide, a GLP-1 receptor agonist (GLP-1RA), has shown potential in controlling type 2 diabetes due to their ability to lower blood glucose and improve cardiovascular outcomes. Although semaglutide is the newest GLP-1RA available, its cardiovascular effects have not yet been systematically reviewed.

Methods

This meta-analysis adhered to a pre-registered protocol (PROSPERO ID: CRD42024538847). PubMed, Scopus, and Web of Science were searched comprehensively on February 26, 2024, to locate randomized controlled trials (RCTs) assessing semaglutide's impact on cardiovascular outcomes. A random effects model was used to analyze the data and the Cochrane risk of bias 2 tool was used to assess the quality of included studies.

Results

There were four RCTs involving 27,617 individuals (13,809 receiving semaglutide and 13,808 receiving placebo). There was a significant reduction in the risk of major adverse cardiovascular events (MACE) after administration of semaglutide (RR: 0.81, 95% CI: 0.74–0.88, p < 0.001, I²: 0.00%). Additionally, benefits were observed for particular findings such as cardiovascular death and nonfatal myocardial infarction. However, semaglutide showed no significant effect on nonfatal stroke, heart failure hospitalizations, or admissions for unstable angina. The treatment also led to fewer serious adverse events (RR: 0.91, 95% CI: 0.88–0.94, p < 0.001, I²: 0.01%), though it was linked to a higher rate of treatment discontinuation (RR: 1.67, 95% CI: 1.27–2.21, p < 0.001, I²: 91.09%), which was underemphasized in its clinical implications and should be interpreted more cautiously considering the trade-off with benefits.

Conclusion

Semaglutide effectively reduces the risk of MACE and modestly improves cardiovascular mortality rates and nonfatal myocardial infarction, without significantly affecting nonfatal stroke incidence and hospitalizations (for angina and heart failure). While it lowers the occurrence of serious adverse events, it also increases the likelihood of discontinuing treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-025-05278-3.

Introduction

The global rise in diabetes mellitus cases, now estimated to cause 11.3% of all deaths worldwide, poses an important public health issue [1, 2]. Individuals with diabetes are at a 2- to fourfold greater risk of developing cardiovascular diseases (CVDs) [3], which tend to occur approximately fifteen years earlier than in those without diabetes [4]. As CVDs represent the primary cause of morbidity and mortality among patients with diabetes [5], efforts have been made to alleviate the risk of CVDs through measures such as exercise, nutrition, weight management, smoking cessation, and medical interventions such as glucose-lowering medications [6].

However, it should be noted that not all glucose-lowering agents have the same impact on the health of the cardiovascular system. Although epidemiological research hints at a link between blood sugar levels and cardiovascular events among diabetes patients, major trials and systematic reviews have not confirmed these findings [7]. In fact, studies on glucose-lowering drugs have often failed to show significant benefits in reducing CVD-related morbidity and mortality, and some drugs, such as with rosiglitazone and saxagliptin have even shown detrimental effects on cardiovascular health [8, 9]. With that said, many agents of the sodium-glucose cotransporter 2 (SGLT2) inhibitors and GLP-1 receptor agonists (GLP-1RA) classes have indicated clear cardiovascular improvements [10].

Semaglutide, commercially known as Ozempic®, is a GLP-1RA with a long half-life of about one week [11] and is available both as an oral agent and weekly subcutaneous injections [12]. This potent glucose-lowering drug has been shown to modify key risk factors of CVDs, as it significantly reduces the levels of HbA1c, body mass, and systolic blood pressure [13]. This agent is also associated with anti-inflammatory effects and impact on endothelial function in the prevention of CVDs [14, 15]. Of course, this medication is accompanied by side effects that put into question its tolerability and discontinuation [16].

Indirect comparisons of semaglutide with other GLP-1RAs via network meta-analysis suggest that semaglutide has the highest probability of having cardiovascular benefits [17]. Furthermore, a network meta-analysis comparing GLP-1RAs and SGLT2 inhibitors has found that semaglutide and albiglutide reduce MACE more than others, suggesting that they are the most cardioprotective agents [18]. With that said, to date, this is the first meta-analysis and systematic review study focusing solely on semaglutide. In this study, we aim to perform the first systematic review and meta-analysis examining the effect of semaglutide on cardiovascular outcomes relative to a placebo, ensuring a homogeneous comparator group and minimizing confounding from variations in efficacy and safety profiles of active comparators, thereby isolating the cardiovascular effects attributable to semaglutide itself. In addition, we aimed to evaluate the potential tolerability concerns or discontinuation of this agent.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) criteria were followed in the conduct of this systematic review and meta-analysis [19]. Under registration ID CRD42024538847, the study protocol was pre-registered with the International Prospective Register of Systematic Reviews (PROSPERO).

We made three main deviations from the registered protocol. First, the risk of bias was assessed using the Cochrane Risk-of-Bias 2 (RoB 2) tool instead of the pre-specified JADAD scale. This change was made to provide a more comprehensive and domain-specific assessment of bias, aligning with current Cochrane recommendations for systematic reviews of RCTs. Second, we conducted an exploratory subgroup meta-analysis based on the route of administration (oral vs. subcutaneous) to investigate a clinically important variable that emerged as a potential source of heterogeneity. Finally, meta-analyses were performed on specific adverse events of high clinical interest like pancreatitis, acute kidney injury (AKI), and malignant neoplasms which were not pre-specified in the protocol.

Search strategy

An extensive search of databases including Web of Science, PubMed, and Scopus was carried out on February 26, 2024. The search strategy utilized both keywords and MeSH terms related to "Semaglutide" and "Cardiovascular outcomes," without any restrictions on language or time of publication. The full search strategy for each database is detailed in Supplementary Table S1.

Screening process

Search findings from the databases were combined, followed by the removal of duplicates using EndNote 21 (Thomson Reuters, Toronto, ON, Canada). Two reviewers (S.S. and A.A.) individually conducted a two-stage screening process, first assessing titles and abstracts, followed by full-text evaluations. A third reviewer was consulted to resolve any disagreements (S.R.F.).

Study selection

Studies that satisfied the following predefined PICOS criteria were selected for inclusion:

Population: Individuals aged 45 years or above with established cardiovascular disease were considered eligible, such as previous myocardial infarction, stroke, or peripheral artery disease. This age threshold aligns with major clinical guidelines that recommend routine cardiovascular risk assessment for middle-aged and older adults [20, 21]. Exclusion criteria included recent use of dipeptidyl peptidase-4 (DPP-4) inhibitors within 30 days or GLP-1RAs within 90 days before screening, planned vascular revascularization procedures, or ongoing long-term dialysis.

Intervention: The intervention tested was semaglutide, administered in any dose or by any route. Participants receiving placebo or standard-of-care treatment served as the comparison group.

Outcomes: The study aimed to evaluate both efficacy and safety outcomes. To determine efficacy, major adverse cardiovascular events (MACE) were tracked, including heart attack, cardiovascular death, stroke, heart failure leading to hospitalization, unstable angina requiring hospitalization, and instances where treatment was terminated. Safety was assessed by monitoring adverse events, especially those that led to permanent discontinuation of semaglutide. It also looked at serious adverse events and specific concerns such as malignant tumors, AKI, and acute pancreatitis over the course of the trial.

Study design: This study only included randomized controlled trials (RCTs) to ensure high-quality evidence. Non-randomized trials, observational studies (such as cohort, case-controlled, or cross-sectional studies), review articles, conference abstracts, editorial letters, and animal studies were excluded.

Outcomes and data extraction

The primary outcome focused on semaglutide's ability to enhance cardiovascular outcomes. To compare semaglutide with placebo, we evaluated major adverse cardiovascular events (MACE)—including cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke—as well as hospitalization because of unstable angina, hospitalization for heart failure, and treatment discontinuation, without imposing any restrictions on the follow-up period. The secondary outcome focused on safety, assessed through adverse events reported during the study periods, again without specific timepoint limitations.

Two independent reviewers (A.A. and S.S.), supervised by a third reviewer (S.R.F.), performed a standardized data extraction form and gathered data from the chosen RCTs, such as the names of the authors, the year the study was published, the study's location, the trial's duration, the total number of participants, the number of participants in the semaglutide and placebo groups, the type of placebo group (including use of other medications), rates of successful revascularization, patient comorbidities, and all relevant outcome data.

Quality assessment

Using the Cochrane Risk of Bias 2 tool, two reviewers (S.S. and A.A.) independently assessed the RCTs' risk of bias under the supervision of a third (S.R.F.) [22]. Five domains are examined by this tool: randomization bias, deviations from the planned interventions, missing outcome data, assessment of outcomes, and selection of reported results. Domains were rated as High risk," "Some concerns," or "Low risk" depending on the responses to signaling questions. Specifically, judgments were based on the criteria specified by the RoB 2 tool. Bias from the randomization process was judged as 'Low risk' if allocation was adequately concealed and baseline differences were unlikely. Bias due to deviations from intended interventions was assessed based on the appropriateness of blinding of patients and personnel. Bias due to missing outcome data was considered 'Low risk' if the amount of missing data was minimal and unlikely to be related to the true outcome. Finally, bias in the selection of the reported result was judged based on whether the study's outcomes were analyzed in accordance with a pre-specified plan.

A detailed breakdown of the assessment for each study across all domains is provided in Fig. 8. Results were visually summarized utilizing the Risk of Bias Visualization (ROBVIS) [23].

Fig. 8.

Quality assessment using the Cochrane risk of bias tool

Statistical analysis

Pairwise meta-analyses were conducted with Stata version 17 (StataCorp LP, College Station, TX, USA). Only event counts and total participants per group (semaglutide vs. placebo) were used. Relative risk (RR) and 95% confidence intervals were provided, with statistical significance determined at p < 0.05. Due to anticipated differences in study protocols, a random effects model was applied. The I² statistic was used to assess heterogeneity, with I² > 50% indicating substantial heterogeneity [24]. Sensitivity analyses (leave-one-out method) and publication bias assessments (Egger’s regression) were not performed because the number of studies did not meet minimum thresholds (five for sensitivity, ten for publication bias). Furthermore, an analysis of subgroups was carried out based on the semaglutide administration route (oral vs. subcutaneous). Meta regressions on age, body weight (kg), glycated hemoglobin levels (%), treatment duration (months), and exclusion of diabetic patients were conducted for all cardiovascular outcomes excluding hospitalizations due to the limited number of studies. For other outcomes, meta regressions were conducted in case of significant heterogeneity. A meta regression using dose was conducted on studies with a subcutaneous administration route. Due to ambiguity regarding the exact dosage in the study by Marso et al. [25], we opted to use the average of the two doses used in their study. For time to event analyses, reported hazard ratios (HR) were pooled for each cardiovascular outcome excluding hospitalizations due to inconsistent reporting. For safety outcomes, by R software (version 4.3.2) and the meta package, proportional meta-analyses were carried out, once more utilizing a random effects model and evaluating heterogeneity using the I² statistic. Statistical significance was defined by p < 0.05.

Results

Based on the comprehensive systematic search, 1,362 research papers were identified. A total of 1050 articles were available for screening after 312 duplicates were removed. As a result of title and abstract screening, 1013 articles were excluded, leaving 36 articles for full-text retrieval. Among these, four RCTs fulfilled the criteria to be included in the review. Of the 36 papers excluded during full-text evaluations, 17 did not report separate outcomes for the semaglutide cohorts, six used a control group receiving heterogeneous interventions, and nine were post hoc analyses of previous studies (Fig. 1).

Fig. 1.

PRISMA flowchart

Study characteristics

The four included RCTs were conducted between 2016 and 2024, and included a pooled population of 27,617 individuals, with a mean age of 63.25, 77.41% male, and follow-up periods ranging between 15.9 to 39,8 months. A total of 13,809 participants received semaglutide, of which 8,803 received subcutaneous semaglutide 2.4 mg once a week, 1,591 received oral semaglutide 14 mg once a day, and 3,415 received subcutaneous semaglutide 0.5 or 1 mg once a week, while 13,808 individuals received placebos for comparison. Among the patients of these 4 RCTs the number of patients who had successful revascularization was determined (Table 1).

Table 1.

Study characteristics

Author (year), country	Study design	NO of all patients	Mean age, Gender (male%)	Dose and frequency of semaglutide	Type of placebo (use of other medications)	Comorbidities	Revascularization	Follow-up period
Husain et al. [26] (2019), Canada	RCT	3183	semaglutide: 66, 68.1% placebo: 66, 68.6%	14 mg, oral, once per day	Investigators were encouraged to maintain and intensify patients’ existing glucose-lowering and cardiovascular medication, in accordance with local and international guidelines, in addition to semaglutide or placebo	Type 2 diabetes, CVD, CKD	were excluded	15.9 months
Lincoff et al. [27] (2023), USA	RCT	17,604	semaglutide: 61.6, 72.2% placebo: 61.6, 72.5%	2.4 mg, subcutaneous, once weekly	If diabetes developed during the trial, the patient continued to take the assigned trial product. The use of glucose lowering medications was at the discretion of the investigator, although initiation of open label treatment with a GLP-1 receptor agonist was prohibited)	CVD, Obesity	Semaglutide:473 Placebo: 608	39.8 ± 9.4 months
Marso et al. [25] (2016), USA	RCT	3297	semaglutide, 0.5 mg: 64.6, 59.9% semaglutide, 1 mg: 64.7, 63% placebo, 0.5 mg: 64.8, 58.5% placebo, 1 mg:64.4, 61.5	0.5 and 1 mg, subcutaneous, once weekly	All investigators were encouraged to treat all the patients according to local guidelines to achieve the most effective glycemic control, and additional non-investigational antihyperglycemic medication (non-incretin-based therapy) could be added or adjusted	Type 2 diabetes, CVD	Semaglutide:83 Placebo:126	104 weeks
Mahaffey et al. [28] (2024), USA	RCT	3533	Mean age: 67.65 semaglutide:70.61% placebo: 68.85%	1 mg, subcutaneous, once weekly	The use of SGLT2 inhibitors and mineralocorticoid-receptor antagonists was permitted, and randomization was stratified according to SGLT2 inhibitor use at baseline	CKD, Type 2 diabetes	NP	156 weeks

Abbreviations: RCT Randomized controlled clinical trials, CVD Cardiovascular disease, CKD Chronic kidney disease, IHD Ischemic heart disease, MI Myocardial infarction, HF heart failure, BMI Body mass index

The initial study by Marso et al. [25] found that administration of 0.5 mg and 1 mg semaglutide weekly for 104 weeks lowered the risk of cardiovascular death, nonfatal heart attack, and nonfatal stroke when compared to placebo among 1,648 semaglutide-treated patients and 1,649 placebo-treated patients with diabetes. According to this study, 83 patients from the group receiving semaglutide and 126 patients from the placebo group had a history of successful revascularization. In another study, Lincof et al. [27] found that among the participants, 8,803 non-diabetic individuals who received semaglutide and 8,801 who were given a placebo, treatment with 2.4 mg semaglutide weekly over a mean duration of 39.8 months resulted in decreased incidence of cardiovascular death, nonfatal myocardial infarction and nonfatal stroke. A history of successful revascularization was documented in 473 participants in the semaglutide group and in 608 participants in the placebo group. Husain et al. [26] found that 14 mg of semaglutide administered once daily for 15.9 months was not inferior to placebo in reducing cardiovascular mortality, nonfatal myocardial infarction, and nonfatal stroke among 1,591 patients receiving semaglutide and 1,592 patients receiving placebo. In this study, people who underwent revascularization for their CVD were excluded. Furthermore, Mahaffey et al. [28] found that among 1,767 patients treated with semaglutide and 1,766 given placebo, subcutaneous semaglutide 1 mg administered for 156 weeks significantly decreased the overall risk of MACE. An overview of the detailed characteristics of the RCTs is provided in (Table 1). It is worth mentioning that all major events, including deaths and serious adverse events such as pancreatitis or malignancy, were adjudicated by independent external event adjudication committees in the four included studies. These committees, composed of specialists with no conflicts of interest, reviewed pre-defined clinical data and source documents in a blinded manner according to pre-specified diagnostic criteria and published endpoint definitions.

Cardiovascular outcomes

When pooling the four RCTs comparing semaglutide with placebo, a significant relationship was identified between semaglutide treatment and decreased incidence of MACE (relative risk, (95CI), p-value, I²; 0.81, (0.74, 0.88), < 0.001, 0.00%); Fig. 2). When assessing individual adverse cardiovascular events, however, this relationship appears to dissipate, as the incidence of nonfatal stroke did not differ significantly between participants receiving semaglutide and those receiving placebo (stroke: relative risk, (95CI), p-value, I2; 0.90, (0.68, 1.20), 0.47,51.28%). Unlike these events, analyses indicated improvement in cardiovascular mortality and nonfatal myocardial infarction following treatment with semaglutide. (cardiovascular mortality: relative risk, (95CI), p-value, I²; 0.80, (0.70, 0.91), < 0.001,5.76%; MI: relative risk, (95CI), p-value, I²; 0.78, (0.67, 0.90), < 0.001, 8.2%; Fig. 3). Pooled HRs for these outcomes also followed the same pattern with only stroke remaining insignificant (HR, (95CI), p-value, I²: MACE: 0.80, (0.73, 0.87), < 0.001, 0%; stroke: 0.89, (0.68, 1.18), 0.437, 48.5%; cardiovascular mortality: 0.79, (0.67, 0.92), 0.003, 20.5%; MI: 0.77, (0.66, 0.89), < 0.001, 6.1%). Meta regressions for age, body weight, glycated hemoglobin, treatment duration, and exclusion of diabetic patients were insignificant in all outcomes (Table 2). Meta regression using dose as a moderator in studies using subcutaneous administration were insignificant for all outcomes (MACE: coefficient = 0.013, p-value = 0.826; stroke: coefficient = 0.083, p-value = 0.82; cardiovascular mortality: coefficient = 0.033, p-value = 0.801; MI: coefficient = −0.037, p-value = 0.717).

Fig. 2.

Results of meta-analysis for risk of MACE

Fig. 3.

Results of meta-analysis for risk of cardiovascular mortality, non-fatal stroke, and non-fatal myocardial infarction

Table 2.

Meta regressions for meta-analyses on cardiovascular outcomes

Outcome	Moderator	Coefficient (95CI)	P-value
Major adverse cardiovascular events	Age	0.001 (−0.036, 0.038)	0.950
	Weight (kg)	−0.001 (−0.027, 0.026)	0.968
	Glycated hemoglobin (%)	−0.011 (−0.081, 0.059)	0.755
	Average duration of treatment	0.003 (−0.008, 0.015)	0.596
	Exclusion of diabetics	0.012 (−0.154, 0.179)	0.885
Stroke	Age	0.022 (−0.160, 0.204)	0.811
	Weight (kg)	−0.015 (−0.148, 0.118)	0.830
	Glycated hemoglobin (%)	−0.087 (−0.394, 0.221)	0.579
	Average duration of treatment	0.023 (−0.003, 0.049)	0.085
	Exclusion of diabetics	0.082 (−0.740, 0.904)	0.844
Cardiovascular mortality	Age	−0.035 (−0.090, 0.020)	0.211
	Weight (kg)	0.024 (−0.016, 0.063)	0.236
	Glycated hemoglobin (%)	−0.038 (−0.193, 0.117)	0.632
	Average duration of treatment	0.006 (−0.016, 0.029)	0.587
	Exclusion of diabetics	0.133 (−0.165, 0.431)	0.382
Myocardial Infarction	Age	0.043 (−0.023, 0.108)	0.206
	Weight (kg)	−0.029 (−0.077, 0.020)	0.245
	Glycated hemoglobin (%)	0.060 (−0.088, 0.208)	0.427
	Average duration of treatment	−0.013 (−0.031, 0.005)	0.146
	Exclusion of diabetics	−0.167 (−0.501, 0.166)	0.325

Hospitalizations due to unstable angina or heart failure indicated no meaningful difference between the groups (unstable angina: relative risk, (95CI), p-value, I²; 0.90, (0.72, 1.12), 0.35,0.00%; heart failure: relative risk, (95CI), p-value, I²; 0.89, (0.71, 1.13), 0.35,17.70%; Fig. 4).

Fig. 4.

Results of meta-analysis for risk of unstable angina and heart failure

Adverse events

Incidence of adverse events was also investigated among the RCTs, with our pooled analyses showing that semaglutide lowers the risk of severe adverse events (relative risk, (95CI), p-value, I2; 0.91, (0.88, 0.94), < 0.001, 0.01%; Fig. 5). A subsequent proportional meta-analysis showed that 32% (95CI: 22%, 44%; Additional file 1: Figure. S1.) of semaglutide users experienced serious adverse events. Treatment terminations following treatment-related adverse events were also compared between semaglutide and placebo arms, the pooled data revealed an increased risk of discontinuation among those in the semaglutide group (relative risk, (95CI), p-value, I2; 1.67, (1.27, 2.21), < 0.001, 91.09%); Fig. 6), with a pooled prevalence of about 14% (95CI: 12%, 16%; Additional file 1: Figure. S2.) among semaglutide treated participants. To investigate the source of this heterogeneity, a meta-regression was conducted. None of the tested variables (age, body weight, glycated hemoglobin levels, or treatment duration) were found to be statistically significant sources of the observed variance (all p > 0.05).

Fig. 5.

Results of meta-analysis for risk of serious adverse events

Fig. 6.

Results of meta-analysis for risk of treatment terminations

When assessing individual adverse events, acute pancreatitis, AKI, and malignant neoplasm were among the most reported across the 4 RCTs. As far as acute pancreatitis is concerned, semaglutide did not reduce the risk in a significant way (relative risk, (95CI), p-value, I²; 0.82, (0.53, 1.25), 0.35, 0.00%; Fig. 7). Similarly, semaglutide did not seem to have much effect on the occurrence of AKI as well (relative risk, (95CI), p-value, I²; 0.88, (0.77, 0.99), 0.04, 0.00%; Fig. 7). The incidence of malignant neoplasm complications was also not statistically different between semaglutide and placebo arms (relative risk, (95CI), p-value, I²; 1.01, (0.91, 1.13), 0.83,0.00%; Fig. 7).

Fig. 7.

Results of meta-analysis for risk of acute pancreatitis, AKI, and malignant neoplasm

Quality assessment

All four RCTs were assessed for quality using the Cochrane Risk of Bias 2 tool. As shown in Fig. 8, all the RCTs were of high quality, with a low risk of bias.

Discussion

This is the first systematic review and meta-analysis focused exclusively on RCTs evaluating major cardiovascular events (MACE). Our findings indicate that semaglutide exerts a favorable effect on MACE as well as other significant clinical outcomes. According to our analyses, semaglutide reduces the risk of MACE and, to some extent, cardiovascular mortality and nonfatal myocardial infarction. The risk of stroke does not differ significantly between individuals who receive semaglutide and those who do not. In terms of safety, semaglutide reduces the potential for major unfavorable outcomes without increasing the incidence of AKI, acute pancreatitis, or malignant neoplasms. However, semaglutide treatment was associated with higher discontinuation rates compared to placebo.

According to international guidelines, GLP-1RAs are recommended for patients with type 2 diabetes who have established atherosclerotic cardiovascular disease or are considered at high cardiovascular risk. [29, 30]. Semaglutide, a GLP-1RA, has been linked to reduced rates of MACE, including heart attacks and strokes, in patients with diabetes [31, 32]. Semaglutide is known to improve cardiovascular health markers such as blood pressure, cholesterol levels, and inflammation [33]. These improvements in traditional risk factors may partially explain the observed reduction in cardiovascular events. Although the exact mechanisms by which semaglutide confers cardioprotection are still under investigation, current evidence suggests that its effects extend beyond glucose control, supporting its use in patients with type 2 diabetes and established cardiovascular disease [34, 35].

Although semaglutide lowered the overall risk of MACE, it did not significantly reduce the incidence of nonfatal stroke or heart failure hospitalizations. These nonsignificant findings may be explained by limited statistical power for less frequent events, variability in baseline comorbidities such as atrial fibrillation or hypertension, and differences in trial duration, which ranged from 15.9 to 39.8 months. It is possible that longer follow-up is required to detect an effect on stroke outcomes, or that stroke risk is influenced by distinct mechanisms less responsive to semaglutide therapy. Future trials with larger event numbers and stratified analyses may help clarify these discrepancies.

Our findings align with those of prior studies, including previous network meta-analyses, which have demonstrated that semaglutide reduces the incidence of MACE [36, 37], yet our study is the first synthesis of direct evidence proving its benefits in this regard. Another meta-analysis found that GLP-1 RAs, including semaglutide, significantly reduced the relative risk of MACE by 10%, whereas our study found a 22% reduction. However, no significant association was observed between GLP-1 RAs and the risk of fatal or non-fatal myocardial infarction, fatal or non-fatal stroke, hospitalization due to unstable angina or heart failure, or the occurrence of pancreatitis [38]. A meta-analysis conducted by Gao et al. supports our findings regarding higher discontinuation rates due to adverse events, demonstrating an increased incidence compared to placebo. However, in contrast to our results, their study found that semaglutide was associated with a higher risk of serious adverse events [39].

Semaglutide exerts its beneficial cardiovascular effect by augmenting various aspects of metabolic and cardiovascular health. First, it activates GLP-1 receptors, enhancing incretin activity, and reduces both fasting and postprandial glucose levels through various mechanisms, including increased insulin secretion (glucose-dependent), suppression of glucagon release, and decreased hepatic gluconeogenesis [40]. In addition, similar to other GLP-1 RAs, it affects different stages of the pathophysiological process of CVDs, from risk factors like dyslipidemia, hypertension, obesity, diabetes, and coronary artery atherosclerosis to more advanced stages such as myocardial ischemia, atherothrombosis, myocardial infarction, left ventricular remodeling, and death, without raising the risk of arrhythmia, hospitalization, or exacerbating heart failure [35, 41]. Semaglutide has been shown to improve heart structure and reduce left ventricular hypertrophy through decreasing fibrosis and fibrosis mediators. Additionally, semaglutide has been shown to enhance heart function by reducing myocardial diastolic stiffness [42]. Oral semaglutide also reduces systemic blood pressure, potentially reducing the incidence of MACE [43–45]. Finally, semaglutide has been associated with a lower incidence of gastrointestinal adverse events compared to other GLP-1 RAs, such as dulaglutide and liraglutide [46, 47].

Preclinical studies have also highlighted potential mechanisms of benefit. In animal models, subcutaneous semaglutide alters genes linked to atherosclerosis and dose-independently reduces aortic plaque lesions in non-diabetic low-density lipoprotein receptor-deficient mice [33]. Moreover, it lowers proatherogenic inflammation by lowering immune cell recruitment and plasma levels of the inflammatory cytokines TNF-α and IFN-γ [33]. These cytokines are mechanistically relevant, as activating the TNF-α signaling pathway leads to hypertension, atherogenesis, vascular dysfunction, and unfavorable cardiac remodeling following MI [48], while IFN-γ contributes to the progression of atherosclerotic CVDs through the JAK/STAT pathway [49]. While these findings are biologically plausible, they should be interpreted with caution and considered hypothesis-generating rather than confirmatory. Importantly, no direct causal evidence has been demonstrated in human RCTs. Bridging studies are needed to determine whether these mechanisms translate into the cardiovascular benefits observed in clinical populations [33].

An important consideration in semaglutide therapy is treatment discontinuation, which we found to be higher among semaglutide recipients than placebo. This is clinically relevant, as an observational study reported an increased risk of major cardiovascular events following discontinuation of GLP-1 RA therapy [50]. Another important concern is that treatment discontinuation could affect the included samples if high-risk patients dropped out earlier, potentially introducing bias in the overall results. While the literature suggests that the route of delivery (subcutaneous vs. oral) could be an important factor in treatment adherence [51], our analyses found no difference between the two routes; however, this finding is limited by the fact that only one trial evaluated oral semaglutide, warranting further investigation. In real-world settings, additional causes of discontinuation include high out-of-pocket costs, restrictive insurance coverage, prior authorization requirements, and drug shortages [52]. Notably, while persistence with semaglutide was lower compared to placebo, existing evidence indicates that semaglutide has a lower discontinuation rate relative to other GLP-1 RAs [53]. Moreover, many of the patients discontinuing GLP-1 RAs eventually reinitiate therapy, meaning the proportion of long-term users is greater than suggested by analyses limited to discontinuation rates [54].

Limitations

Our study had some limitations. First, the number of RCTs available for analysis was limited, thus precluding any formal statistical assessment for publication bias, such as funnel plot analysis. Consequently, the potential for unpublished studies with null or negative results to influence our findings cannot be ruled out. Therefore, the pooled effect estimates presented in this review should be considered preliminary and warrant confirmation in future, larger-scale studies. Nonetheless, the lack of heterogeneity in our meta-analyses strengthens the quality of our evidence. Second, although the included RCTs were large-scale, recent, and of high methodological quality, they were all industry-sponsored, which may increase the risk of selective reporting or publication bias. Third, the RCTs exclusively enrolled patients with a history of CVD, with the population being predominantly male and of aged 45 years and older. This reflects the current focus of the primary literature and highlights a gap in the evidence regarding the effects of this intervention in younger adults who may also be at risk for CVD. This limits the applicability of our results to not only the younger female patients, but also to individuals who may be overweight, obese, or have renal disease, diabetes, or psychiatric disorders but no history of CVD. We recommend future studies to explore these populations, as semaglutide may have a diminished therapeutic effect in these individuals. Another limitation of our study is that the included RCTs used different semaglutide doses (0.5, 1.0, and 2.4 mg) and administration routes (oral and subcutaneous), which may introduce heterogeneity and complicate the interpretation of pooled outcomes. Future studies directly comparing different doses and routes are needed to clarify whether cardiovascular benefits are consistent across treatment regimens. Also, the variability in follow-up duration across the included RCTs can affect event capture rates and risk estimates, as shorter trials may underreport long-term outcomes while longer trials may reflect cumulative risks. Additionally, since all the RCTs were conducted in the USA and Canada, the lack of genetic, ethnic, and racial diversity limits the generalizability of the findings to Asian, European, Hispanic, and African populations. Importantly, treatment discontinuation was considerably higher with semaglutide compared to placebo, which may affect long-term adherence, tolerability, and ultimately the effectiveness of semaglutide in practice. In addition, it should be noted that RCTs usually do not reflect a population of real-world patients with multiple comorbidities, medication burdens and socioeconomic factors affecting their adherence. Future studies in real-world settings may prove beneficial in gauging the efficacy of semaglutide more realistically.

Conclusion

This systematic review and meta-analysis demonstrates that semaglutide reduces the risk of major adverse cardiovascular events, cardiovascular mortality, and non-fatal myocardial infarction. However, semaglutide does not significantly reduce the risk of non-fatal stroke and hospitalizations (for angina and heart failure), warranting caution when interpreting its overall benefit. In addition, adverse events such acute pancreatitis, AKI, or malignant neoplasms were not significantly affected by semaglutide use. Importantly, semaglutide was associated with a significantly higher rate of treatment discontinuation, which represents a key clinical consideration when weighing its overall benefit-risk profile. These findings reinforce the role of GLP-1RAs, and semaglutide in particular, as an important therapeutic option for patients with type 2 diabetes and established cardiovascular disease, supporting their inclusion in evidence-based guideline recommendations. Future research should focus on evaluating semaglutide in more diverse populations, comparing different doses and administration routes, and exploring strategies to improve long-term adherence in real-world settings.

Authors’ contributions

Conceptualization: S.S.; Methodology: M.A.; Formal analysis and investigation: S.R.F.; Writing—original draft preparation: S.S., A.A., M.A., I.M., S.R.F.; Writing—review and editing: A.G.J., S.S., I.M.; Supervision: I.M.

Funding

This study received no funding.

Data availability

Supporting data for this study are available in the PubMed, Web of Science, and Scopus databases, drawn from publicly available resources.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.