Gastrointestinal safety evaluation of semaglutide for the treatment of type 2 diabetes mellitus: A meta-analysis

Abstract

Background:

Semaglutide, as an innovative weekly formulation, has attracted much attention. Nevertheless, the predominant occurrence of gastrointestinal adverse events (GIAEs) poses a noteworthy challenge linked to the use of this medication, substantially affecting its clinical applicability and the overall well-being of patients. Therefore, this systematic review aims to comprehensively discuss the GIAEs, providing a basis for clinical therapeutic decisions.

Methods:

We systematically searched 4 independent databases for randomized controlled trials investigating the application of semaglutide in managing type 2 diabetes mellitus. The search period spanned from the inception of the databases to December 2023. We conducted a comprehensive meta-analysis, employing Review Manager 5.4.1 software, to systematically analyze and evaluate potential biases. Our primary emphasis was on assessing the gastrointestinal safety profile of semaglutide.

Results:

The outcomes unveiled a noteworthy rise in the collective occurrence of GIAEs across all dosage groups of semaglutide in comparison with the control group (P < .05). Upon further analysis, it was observed that semaglutide showed a heightened occurrence of GIAEs in contrast to the placebo. However, statistically significant distinction was not observed when compared to the reduction of conventional doses or the transition to other types of glucagon-like peptide-1 receptor agonist. Additionally, an extended treatment duration with semaglutide (>30 weeks) demonstrated an association with a certain degree of decrease in the incidence of gastrointestinal events. Funnel plot assessment for publication bias demonstrated high-quality inclusion of studies with no apparent publication bias.

Conclusion:

The frequency of GIAEs in using semaglutide was observed to be elevated in comparison to the control group. However, it was comparable to other glucagon-like peptide-1 receptor agonist or low-dose treatment regimens. Additionally, an extended treatment duration played a role in decreasing the frequency of GIAEs. These findings provide valuable insights for clinical practice. Nonetheless, further research is crucial to explore supplementary data indicators, informing clinical practices and better serving the interests of patients.

1. Introduction

In recent years, type 2 diabetes mellitus (T2DM) has arisen as a significant societal challenge.^[1–3] Pharmacological interventions stand as crucial measures in controlling type 2 diabetes.^[4,5] Among them, GLP-1RAs demonstrate distinctive benefits in sustaining stable blood glucose levels, promoting weight reduction, and mitigating the risk of cardiovascular events. Consequently, they have gained widespread use globally.^[6–9] Newer GLP-1RAs, exemplified by semaglutide, in weekly formulations, significantly enhance treatment adherence by reducing dosing frequency. Studies have shown that long-acting GLP-1RAs are more effective in reducing fasting blood glucose compared to short-acting GLP-1RAs. Moreover, semaglutide exhibits greater efficacy in reducing blood glucose levels and body weight compared to other GLP-1RAs on the market.^[10] Additionally, as the first orally administered formulation of GLP-1RA developed, semaglutide currently offers the most effective weight reduction for patients with type 2 diabetes.^[11,12] Consequently, semaglutide has become one of the most extensively utilized antidiabetic medications worldwide, aiding diabetic patients in overcoming multiple challenges.^[13,14] However, akin to any medication, semaglutide is accompanied by drug-related adverse reactions, with GIAEs being the most prevalent.^[15–17]

It is a well-established fact that patients frequently experience digestive symptoms including nausea, diarrhea, vomiting, abdominal pain, and bloating when using semaglutide. These discomforts significantly impact the patient’s medication experience and, in some cases, even limit the clinical application of semaglutide, leading to the failure of treatment plans.^[18,19] However, regarding the GIAEs associated with semaglutide, the typical clinical approach is to inform patients that these are considered normal occurrences. When patients cannot tolerate them, medication replacement is pursued. Moreover, it seems challenging to provide additional information to optimize treatment plans and ensure full patient awareness. Specifically, uncertainties persist regarding whether dose adjustments, altering regimens, or changing treatment medications would decrease the incidence of GIAEs. Therefore, in-depth research on the GIAEs of semaglutide is crucial for optimizing the treatment plans for diabetic patients.

This study will employ meta-analysis to evaluate the GIAEs linked to the utilization of semaglutide in treating type 2 diabetes, with the goal of furnishing a more scientifically robust foundation for clinical practice. Through this research, we hope to enhance clinical understanding of the GIAEs associated with semaglutide, ultimately offering more personalized and effective treatment plans for diabetic patients.

2. Methods

This report was developed and conducted following the PRISMA 2020 guidelines.^[20] Considering its nature as a meta-analysis, as the data involved in this study were collected without direct patient participation and the article does not disclose any private patient information, ethical review and approval are not needed.

2.1. Search of the literature

The preliminary work of this article involved comprehensive searches conducted independently by 2 researchers across 4 databases: PubMed, Web of Science, Cochrane Library, and Embase. The search period spans from December 2013 to December 2023, and the inclusion criteria were limited to articles published in English. The search strategy incorporated the following keywords: semaglutide, rybelsus, Wegovy, Ozempic, diabetes mellitus, diabetes, diabetic Mellitus, experimental diabetic, diabetic, type 2 diabetes mellitus, II diabetes, type 2 diabetes, T2DM. Subsequently, Boolean logic was applied to merge subject headings and free-text terms, determining the final data retrieval strategy.

2.2. Inclusion and exclusion

2.2.1. Inclusion criteria

Inclusion criteria were defined as follows:

1. Participants. Individuals diagnosed with type 2 diabetes mellitus.

2. Intervention. The test group received semaglutide, while the control group was administered other antidiabetic medications or a placebo.

3. Comparator. Control group received other antidiabetic medications or a placebo.

4. Outcome measures. GIAEs, including but not limited to nausea, diarrhea, vomiting, reduced appetite, indigestion, constipation, abdominal pain, served as the study endpoints.

5. Study type. randomized controlled trials.

2.2.2. Exclusion criteria

Studies published in the form of conference abstracts, reviews, animal experiments, case reports; duplicate publications; studies with incomplete data or lacking clear conclusions; articles where full-text information cannot be obtained; studies involving subjects using oral formulations of semaglutide; studies of poor quality.

2.3. Screening of the literature, bias risk, and data statistics

2.3.1. Screening of the literature

Extraction of data for all eligible studies was independently conducted by 2 authors (MH and XY). Through a comprehensive review of titles, abstracts, and full texts, the subsequent details were extracted from the 12 included studies: the first author’s name, publication year, the country of the first author, sample size of the trial, gender ratio, dosage and administration of semaglutide within the intervention group, treatment regimens within the control group, and the follow-up duration of the studies. Any discrepancies or queries during data summarization were resolved through consultation with a senior physician (JJ).

2.3.2. Bias risk

For the evaluation of bias in the included studies, this article utilized the Cochrane Assessment Tool. The assessment primarily considered factors including randomization process, concealment of allocation, masking of participants and personnel, masking of outcome assessment, fullness of outcome data, reporting bias, and other additional bias sources.

2.3.3. Data statistics

We utilized Revman software (version 5.4.1) for data input in our study, strictly adhering to a series of procedures to ensure the accuracy and consistency of the analysis.

During data summarization, individual trial odds ratios (OR) were computed for binary outcomes. In this study, heterogeneity was estimated based on the magnitude of I². According to the literature references, I² values <25% indicate low heterogeneity, while values between 25% and 75% indicate moderate heterogeneity, and values >75% indicate high heterogeneity.^[21] In the presence of significant heterogeneity, as indicated by an observed I² value exceeding 50%, we employed a stepwise exclusion method and conducted sensitivity analysis to identify the source of clinical heterogeneity. If substantial clinical heterogeneity persisted, subgroup analysis was performed to explore potential sources of heterogeneity.

If the summary result’s P value > .1 and I² < 50% (indicating internal consistency among the included studies), we employed a fixed-effects model for analysis. This model assumes that the true effect size remains constant across all studies. Conversely, if the summary result’s P value < .1 and I² ≥ 50%, signifying significant heterogeneity, and when heterogeneity remained high (I² > 50%) despite our efforts to minimize it, we ultimately utilized the random-effects model for analysis. This model accounts for both within-study and between-study variability, providing a more robust analysis of the data.^[21,22]

Additionally, based on the analysis conducted using a random-effects model, we employed the inverse variance method in STATA software (version 14) to analyze the extracted data and conduct publication bias estimation.^[23]

3. Results

3.1. Literature search results

Ultimately, in the database search, 5390 articles were initially identified, and 1074 duplicates were subsequently eliminated. After a thorough review of titles, abstracts, and full texts, 4304 articles were excluded. Finally, 12 articles were included (Fig. 1).^[24–35]

Figure 1.

The flow chart for the selection of eligible studies in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) 2020 guidelines.

3.2. Baseline characteristics

In the end, 12 clinical studies were encompassed, involving a total participant of 8055 individuals. Among them, the intervention group comprised 4083 individuals, while the control group consisted of 3972 individuals. Six studies reported on the comparison between semaglutide and a placebo, while 7 studies reported on the comparison between semaglutide and other GLP-1RA formulations (Table 1).

Table 1.

Characteristics of included studies.

Study	Year	Country	Sample size	Female percentage (%)	Follow-up (wk)	Control group plan
Frias et al	2021	USA	1879	53	40	Tirzepatide
Davies et al	2021	UK	1210	51	68	Placebo
Heise et al	2022	Germany	117	27	28	Tirzepatide, Placebo
Frias et al	2023	USA	61	36	32	Cagrilintide
Davies et al	2017	UK	140	37	26	Placebo
Sorli et al	2017	USA	387	46	30	Placebo
Pratley et al	2018	USA	1201	45	40	Dulaglutide
Rodbard et al	2018	USA	396	44	30	Placebo
Capehorn et al	2020	UK	576	43	30	Liraglutide
Ji et al	2021	China	868	58	30	Sitagliptin
Nauck et al	2016	Germany	411	77	12	Liraglutide, placebo
Ahmann et al	2018	USA	809	45	56	exenatide

USA = United States of America, UK = United Kingdom.

3.3. Total GI AEs

Incorporating the findings from the analysis of the 12 selected articles, it was observed that the incidence of GIAEs among patients undergoing semaglutide treatment was consistently higher compared to the control group. This disparity demonstrated statistical significance (P < .0001, OR = 2.14 [1.49, 3.07]) (Fig. 2).

Figure 2.

Overall gastrointestinal adverse event incidence in the semaglutide group compared to the control group.

3.4. Semaglutide versus placebo

The synthesis analysis of the included 6 articles revealed that the incidence of GIAEs among patients receiving semaglutide treatment was consistently higher compared to the placebo group. This difference exhibited statistical significance (P < .00001, OR = 3.40 [2.79, 4.14]), with low internal heterogeneity, the analysis was conducted using a fixed-effects model (Fig. 3).

Figure 3.

Overall gastrointestinal adverse event incidence in the semaglutide group compared to the placebo group.

3.5. Semaglutide versus other GLP-1RA

The synthesis analysis of the included 7 articles indicated that, in comparison to other types of GLP-1RA, the occurrence of gastrointestinal events in the semaglutide group did not exhibit a statistically significant distinction, as indicated by a P value of .06 and an odds ratio of 1.26 [0.99, 1.60] (Fig. 4).

Figure 4.

Overall gastrointestinal adverse event incidence in the semaglutide group compared to other GLP-1 receptor agonists. GLP-1 = glucagon-like peptide-1.

3.6. Semaglutide < 1 mg versus semaglutide = 1 mg

Involving a total of 4 articles related to studies on semaglutide at both 1 mg and doses <1 mg, the forest plot analysis yielded a P value of .8, display an odds ratio value of 1.05 [0.71, 1.57]. The comparison between low-dose (<1 mg) and standard dose (1 mg) did not demonstrate a statistically significant variance during the occurrence of gastrointestinal events (Fig. 5).

Figure 5.

Overall gastrointestinal adverse event incidence in semaglutide <1 mg group compared to semaglutide 1 mg group.

3.7. Semaglutide versus control (more than 30 weeks)

Selecting 5 trials with semaglutide treatment lasting more than 30 weeks for meta-analysis, it was noted that the overall occurrence of GIAEs during the extended treatment period did not exhibit a statistically significant distinction if compared to the control group (including placebo). The odds ratio was 1.38 [0.85, 2.21] (Fig. 6). These findings imply that the duration of medication could be a crucial factor influencing the occurrence of adverse reactions.

Figure 6.

Overall incidence of gastrointestinal adverse events in the semaglutide group compared to the control group (treatment duration more than 30 wk).

3.8. Publication bias

After consolidating data from the 12 included articles, observation of the funnel plot revealed a generally symmetrical distribution of points on both sides, indicating the absence of publication bias (Fig. 7). Following subgroup stratification of semaglutide versus placebo, the funnel plot displayed a symmetric distribution of points on both sides, and internal consistency was good, suggesting no publication bias (Fig. 8).

Figure 7.

Funnel plot of included studies.

Figure 8.

Funnel plot for semaglutide versus placebo.

To further investigate the possibility of publication bias, we conducted Begg and Egger tests using STATA software (version 12.0). The results of Begg test showed P values of .493 (uncorrected) and .537 (corrected), both of which were greater than the significance level of .05. Similarly, the P value of Egger test was .974, which was also substantially higher than .05. Therefore, we failed to reject the null hypothesis, indicating no significant publication bias was detected.

4. Discussion

GIAEs are prevalent side effects associated with numerous drugs. These reactions may lead to a decrease in treatment compliance, with patients potentially discontinuing medication due to intolerable adverse effects.^[36–38] Therefore, a comprehensive assessment of GIAEs to drugs is crucial for ensuring the effectiveness of treatment and maintaining patients’ quality of life.

In this study, we conducted to assess the GIAEs associated with semaglutide in the management of T2DM. Our analysis revealed a higher occurrence of GIAEs among patients receiving semaglutide treatment in comparison with other group, and this disparity was statistically significant. This outcome aligns with findings from prior research.^[39–41] Furthermore, multiple meta-analyses targeting various diseases and populations have also yielded similar conclusions.^[16,42,43]

Analyzing the mechanism of action of GLP-1RAs, it is evident that most GLP-1 is produced in the GI tract. GLP-1Rs are widely distributed among enteric neurons throughout the entire intestine, where they play a crucial role in regulating gastrointestinal motility.^[44] However, with pharmacological administration of GLP-1RAs, these neurons undergo rapid but incomplete desensitization, leading to a diminished regulatory effect on GI function.^[45] Additionally, other regions expressing GLP-1R are also modulated by the drugs; for instance, Brunner glands increase secretion of intestinal wall mucus, while parietal cells suppress gastric acid secretion.^[46,47] Consequently, under the intricate and extensive regulatory effects, patients are highly prone to GIAEs.

When compared to other classes of GLP-1RA, the incidence of GIAEs in the semaglutide treatment group did not demonstrate a statistically significant difference. This suggests that the common occurrence of GIAEs with semaglutide is comparable to other GLP-1RA, and it may even be more favorable than certain formulations. Additionally, the once-weekly dosing of semaglutide significantly reduces the frequency of administration for patients, making it an ideal GLP-1RA treatment option. According to our research results, the gastrointestinal event incidence of the dual-agonist Tirzepatide is largely comparable to that of semaglutide, affirming a partial overlap in their pharmacological mechanisms, consistent with prior research outcomes.^[40,48,49] It is noteworthy that a recent preclinical study has revealed the potential of GIP receptor agonism in reducing GIAEs occurrences.^[50] Additionally, research suggests that Tirzepatide, when initiated at lower doses and with smaller dose increments, may mitigate adverse reaction symptoms.^[51] However, due to the limited number of relevant studies at present, further analysis requires a greater quantity of high-quality data. Nonetheless, these findings provide valuable insights to guide future research endeavors.

The outcomes also suggest that the dosage of semaglutide (<1 mg or 1 mg) is not significantly correlated with the GIAEs. This suggests that, in standard dose treatment, reducing the dosage of semaglutide may have a relatively limited effect on alleviating GIAEs, and alternative measures may need to be considered in clinical practice. Subgroup analysis further revealed that in long-term treatment (more than 30 weeks) with semaglutide, the frequency of GIAEs showed no statistically significant difference in comparison to the control group. The analysis suggests that with prolonged drug use, patient tolerance to the medication gradually increases, displaying a characteristic of high early adverse reaction rates and low later rates.

The analysis results also offer insights for clinical practice. First, from a clinical perspective, long-term therapy presents certain advantages. When devising treatment plans for patients, priority should be given to medication regimens lasting more than 30 weeks. Additionally, initiating treatment with lower starting doses and gradually increasing them is beneficial not only for glycemic control but also for establishing drug tolerance, thereby reducing the occurrence of related adverse events. This strategy contributes to enhancing overall diabetes management effectiveness.

Secondly, from the patient’s viewpoint, clinical focus should emphasize medication education. Prior to treatment, patients should be thoroughly informed about the characteristics of semaglutide therapy, including specific symptoms and probabilities of adverse events. When patients understand that these adverse reactions are temporary and may alleviate with continued medication, they are more inclined to adhere to treatment, thus improving treatment compliance.

Lastly, concerning the management of adverse events, individualized treatment plans should be tailored based on specific circumstances when patients experience GIAEs. For mild to moderate GIAEs, immediate switching to other types of GLP-1RA or dose reduction may not necessarily decrease the occurrence of adverse events. We recommend patients adjust their dietary habits, adopt smaller, more frequent meals, and reduce postprandial activities, while continuing medication. For severe adverse reactions, discontinuation and regimen assessment should be conducted only after symptomatic treatment proves ineffective. Consideration for switching to other classes of antidiabetic medications should occur only upon confirmation that the patient cannot tolerate GLP-1RA therapy.

While this aligns with clinical treatment experience, the results still require further discussion based on the aggregation of more high-quality trials. This study’s strengths lie in the incorporation of high-quality randomized controlled trials that encompassed diverse control and dosage groups, enhancing the credibility and representativeness of the meta-analysis to some extent. Additionally, strict inclusion and exclusion criteria, along with scientific statistical methods, were employed to control potential biases and heterogeneity, ensuring the quality and accuracy of the meta-analysis.

The limitations of this study include the restriction to English-language publications, introducing a potential language bias. Furthermore, the analysis only focused on the GIAEs of semaglutide, without considering other adverse reactions such as hypoglycemia, rash, allergies, etc, which may impact the overall safety assessment of semaglutide. Additionally, we did not analyze the severity and duration of GIAEs to semaglutide, which could affect the further assessment of semaglutide’s tolerability and compliance.

5. Conclusion

This study suggests a heightened occurrence of GIAEs with semaglutide treatment for T2DM when compared to the control group. However, switching to other GLP-1RA formulations and reducing the 1 mg treatment dosage did not lead to a decrease in the occurrence of GIAEs. Interestingly, prolonged use of semaglutide was linked to a reduced occurrence of GIAEs. These findings provide valuable insights for clinical practice, but additional research is warranted to investigate supplementary safety and efficacy indicators of semaglutide, providing improved guidance in clinical practice and patient care.

Author contributions

Funding acquisition: Xiaoyan Huang.

Methodology: Xiaoyan Huang.

Validation: Xiaoyan Huang, Jiaojiao Lin, Lunpan Mou.

Writing – original draft: Xiaoyan Huang, Miaohui Wu.

Writing – review & editing: Xiaoyan Huang, Miaohui Wu, Lunpan Mou, Yaping Zhang.

Conceptualization: Miaohui Wu, Jianjia Jiang.

Data curation: Miaohui Wu, Jiaojiao Lin.

Formal analysis: Miaohui Wu.

Software: Miaohui Wu.

Investigation: Jiaojiao Lin, Yaping Zhang.

Visualization: Jiaojiao Lin, Lunpan Mou.

Resources: Lunpan Mou.

Supervision: Jianjia Jiang.