GLP-1RAs caused gastrointestinal adverse reactions of drug withdrawal: a system review and network meta-analysis

Background Glucagon-like peptide-1 receptor agonists (GLP-1RAs) significantly reduce postprandial blood glucose, inhibit appetite, and delay gastrointestinal emptying. However, it is controversial that some patients are intolerant to GLP-1RAs. Methods PubMed, Embase, Web of Science, and Cochrane Library were searched for randomized controlled trials (RCTs) using GLP-1RAs with documented withdrawal due to gastrointestinal adverse reactions (GI AEs) from their inception to September 28, 2022. After extracting the information incorporated into the studies, a random-effects network meta-analysis was performed within a frequentist framework. Results 64 RCTs were finally enrolled, which included six major categories of the GLP-1RA. The sample size of the GLP-1RAs treatment group was 16,783 cases. The risk of intolerable gastrointestinal adverse reactions of Liraglutide and Semaglutide was higher than that of Dulaglutide. Meanwhile, the higher the dose of the same GLP-1RA preparation, the more likely to cause these adverse reactions. These intolerable GI AEs were not significantly related to drug homology or formulations and may be related to the degree of suppression of the appetite center. Conclusion Dulaglutide caused the lowest intolerable GI AEs, while Liraglutide and Semaglutide were the highest. For Semaglutide, the higher the dose, the more likely it is to drive GI AEs. Meanwhile, the risk of these GI AEs is independent of the different formulations of the drug. All these findings can effectively guide individualized treatment. Systematic review registration https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42022359346, identifier CRD42022359346.


Introduction
In 1993, exendin-4, which showed the same primary effect as endogenous GLP-1, was isolated from Heloderma suspectum venom If there is a dispute, the decision will be made by the third author with seniority(Liu su).Excel software was used to extract data, including the title, author, publication year, trial period, experimental and control groups' intervention measures, the number of dropped cases, sample size, average age, average duration, Region, and other information.The Risk of bias table tool in Revman5.4.1(Review Manager (RevMan) [Computer program].Version 5.4.1,The Cochrane Collaboration, 2020) was used to evaluate the methodological quality of each included study according to the Cochrane Collaboration Risk of bias tool (version 5.1.0).The methodological quality of each included study was assessed in terms of 7 aspects: generation of randomized sequences, allocation concealment, blinding of investigators and subjects, blinding of outcome indicators, completeness of outcome data, reporting bias, and other preferences.Literature quality evaluation of the included literature was performed using GRADE profiler 3.6.1.The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) (10) was applied to assess the quality of each study in six primary areas: inter-study bias, reporting bias, indirectness, imprecision, heterogeneity, and inconsistency.All trials were considered to be of "low risk," "some concern," or "high risk." After data extraction and quality evaluation, network metaanalysis and mapping were performed using the network package and mvmeta package in Stata 16.0 and netmeta package in R 4.1.2based on the frequentist framework (11, 12).Our study was divided into four main steps.STEP1, a Network Meta-Analysis was conducted to compare GLP-1RAs as a whole versus non-GLP-1RAs hypoglycemic agents.STEP2, GLP-1RAs specific drugs and non-GLP-1RAs hypoglycemic drugs were analyzed.In STEP2, we unified five preparations of Exenatide, Loxenatide (PEX168), ITCA650, Lixisenatide, and Efpeglenatide modified based on Exenatide-4 as Exenatide for analysis.STEP3, we selected Exenatide and Semaglutide, which exist in different dosage forms, for the network meta-analysis by dosage form classification.STEP4, a network meta-analysis of Semaglutide weekly preparations, was conducted based on dose to determine whether the incidence of intolerable gastrointestinal adverse reactions is related to the dose.
Stata 16.0 software was used to pre-process the data, plot the network relationship, perform Network Meta-Analysis, calculate relative effects, and draw a Ranking of risks.The publication bias was identified by drawing corrected comparison funnel plots.The inconsistency of the Network Meta-Analysis results was tested using the node-splitting method and the loop inconsistency test.P>0.05 for the difference between direct and indirect comparison results was considered insignificant.Thus, the consistency model was used.If the prediction interval crossed the null line, the random-effect model was applied due to the existing heterogeneity.If there was no significant heterogeneity, the fixedeffect model was used.The incidence of GI AEs was a dichotomous outcome, so the relative effect value of the result was expressed as OR and 95% CI.Risk ranking using surface under the cumulative ranking curve(SUCRA) (13) assessment, a greater SUCRA represents a higher risk of occurrence of the drug class in this outcome.P<0.05 was considered a statistically significant difference.

Search results
A total of 13805 articles were searched in PubMed, Embase, Web of Science, and Cochrane Library, and 64 papers were finally included, all of which were RCTs.The screening procedure is shown in Figure 1.

Eligible studies and patient characteristics
The essential characteristics and relevant information of the 64 included studies are detailed in Table 1.Of these, 55 were two-armed, 7 were three-armed, and 2 were four-armed.GLP-1RAs consisting of Exenatide, Liraglutide, Dulaglutide, Semaglutide, Tirzepatide, and Taspoglutide were included in the study.Among them, Exenatide includes six different dosage forms: Exenatide b.i.d, Exenatide q.w, PEX168, ITCA650, Lixisenatide, Efpeglenatide q.w, Efpeglenatide q.m.Semaglutide includes weekly formulation and oral formulation.Other hypoglycemic agents included metformin (Met), Insulin, sulfonylureas (SU), sodium-dependent glucose cotransporter 2 inhibitors (SGLT2i), dipeptidyl peptidase-IV inhibitor (DPP4i), thiazolidinedione (TZD).A total of 37740 patients were included in this study, and 16783 patients received GLP-1RAs.The shortest duration of the included studies was ten days, and the longest was 104 weeks.The mean age of the participants ranged from 14.5 to 74.2 years.Articles with missing data were excluded without contacting the authors for additional data.

Quality assessment
All 64 included documents were RCTs, of which 21 studies did not specify the randomization method, and 1 study had both randomized and non-randomized parts.Based on inclusion and exclusion criteria, this study was classified as low risk; we ranked the 20 open-label papers as high risk because they have a problem with blinding participants and personnel.Blinding of outcome assessment, incomplete outcome data, and selective reporting were all satisfactory in all studies.After a comprehensive analysis, we concluded that the risk of bias in this included literature was low.According to the Cochrane Risk of Bias Assessment Tool, the result of discrimination was shown in Figure 2. GRADE ratings of the included literature showed that the vast majority of comparisons were low risk.This study was considered moderate quality evidence overall.The detailed results were established in the Supplementary Materials.

Outcomes of STEP 1-4
In the reticulated body of evidence for STEP 1-4 (Figure 3) and the GLP-1RAs, eight control interventions, including Met, Insulin, SU, SGLT2i, DPP4i, TZD, Placebo, and Blank, were involved.The most significant direct comparison study with the GLP-1RAs was Placebo.The drug with the largest sample size among GLP-1RAs Flowchart of the literature search for this network meta-analysis.was Semaglutide.In this study, the four Network maps of comparisons included in the analyses were tested using the loop inconsistency test and the node-splitting method.No significant inconsistency or heterogeneity was found.We performed the Network Meta-Analysis in a random-effects model with a consistency model.The results of each Network league table in STEP1-4 are shown in Table 2.A funnel plot assessing the risk of publication bias showed symmetric distribution, indicating a low risk of publication bias(Figure 4).SUCRAs for all results are available in Figure 5. Prediction interval plots also showed low heterogeneity among studies (Figure 6).

STEP 1
Sixty-one studies were included in STEP 1.The network evidence map is shown in Figure 3. STEP 1 results showed that GLP-1RA preparations differed from other hypoglycemic agents or placebo, except for Met and SU.GI AEs analysis found that GLP-1RAs showed a higher risk compared to Insulin (OR=7.54 (48), Blank (45.3),Placebo (41.9),TZD (29.8),Insulin (22.6), and SGLT2i(13.4)had the lowest rate of intolerable GI AEs.The high incidence of adverse reactions with SU may be related to the fact that only one study intervention included SU.Therefore, we do not consider this result to be clinically meaningful.Overall, the results of STEP1 showed that GLP-1RAs were more prone to higher intolerable GI AEs than Insulin, DPP4i, TZD, Placebo, and especially SGLT2i.

STEP 2
Sixty-four studies were included in STEP 2. The network evidence map was shown in Figure 3.The results of STEP 2 indicate that Exenatide has a lower risk of causing GI AEs than

STEP 3
Forty studies were included in STEP 3. The network evidence diagram is shown in Figure 3. STEP 3 results showed no significant difference between various preparations of Exenatide and Semaglutide.The risk ranking showed that Semaglutide q.w (87.1) ranked first, followed by ITCA650 (83.8),Lixisenatide (83.2),Liraglutide (78.1),Taspoglutide (74.1),Efpeglenatide q.m (68.4),Exenatide b.i.d (57.3), oral Semaglutide (55.3),Exenatide q.w (54.9),Efpeglenatide q.w (39.4), and PEX168 (35.8),Placebo (16.5), while PEX168(Loxenatide) had the lowest rate of GI adverse reactions.Although there was a sequential ranking between the different dosage forms of Exenatide and Semaglutide, we could not confirm the difference between the injectable and oral dosage forms of Semaglutide or the modified dosage form or method of use of Exenatide.The STEP3 results showed no significant difference in the incidence of intolerable GI AEs between disparate dosage forms of the same GLP-1RA and disparate modifications.In addition, we found that the incidence of intolerable GI AEs remained at the top for Semaglutide versus Liraglutide, thus confirming the results in STEP 2.

STEP 4
Four articles were included in STEP 4. Figure 3 showed the network evidence plot result.STEP 4 results showed that Semaglutide 0.5 mg q.w caused a lower risk of intolerable GI AEs than Semaglutide (OR=0.66,95%CI 0.47,0.91).The risk ranking results also showed that Semaglutide 1mg q.w (99.6) was associated with a higher risk than 0.5mg q.w (74.8).Meanwhile, the STEP 4 results showed that intolerable GI AEs were more pronounced with higher doses.

Discussion
Research on GLP-1RAs over the past 30 years has been sufficient to confirm their place in glucose-lowering and weight loss (78,79).Given the broad range of indications for GLP-1RAs, we included both diabetic and obese patients.Due to the remarkable efficacy of GLP-1RAs, it has been widely used in the clinic, and many adverse effects have occurred.GLP-1 receptors are expressed in several organs throughout the body, such as the intestine, heart, brain, kidneys, and even the peripheral nervous system (80), and activation of GLP-1 receptors in other organs or systems may result in unpredictable responses (81).Notably, the most common adverse reaction was digestive system adverse reactions (82).Most GI AEs are mild or moderate and can be resolved independently without intervention within a few weeks or months (83, 84).However, some patients may not tolerate them.Some studies have shown that this variation may be due to genetic variation (85).Considering that the intolerable GI adverse reaction will occur relatively shortly after the injection, the present study did not restrict the dosing duration.
Researchers have been concerned about the gastrointestinal adverse effects caused by GLP-1RAs.The influence of these adverse reactions effects on efficacy is still controversial.Lean et al. concluded that patients with GI AEs have greater weight loss (86), while other studies pointed out that nausea caused by GLP-1RAs is not significantly correlated with weight loss (87).Due to the remarkable efficacy of GLP-1RAs, it is increasingly used in clinical practice.Since we cannot prove a positive effect of gastrointestinal adverse effects on efficacy, should we consider sparing patients such suffering?This is the first Network Meta-Analysis based on the number of sample withdrawals due to intolerable GI AEs by GLP-1RAs.The present study found that GLP-1RAs had a higher risk of intolerance than Insulin, SGLT2i, DPP4i, TZD, and Placebo.Among the GLP-1RA drugs, Liraglutide or Semaglutide had a higher risk of intolerant GI AE, and Dulaglutide had the lowest chance.Our findings are consistent with Alatorre et al. (88), who found a lower rate of adverse discontinuation with Dulaglutide.In a real-world study (89) using the disproportionality analysis model, it was also shown that Semaglutide and Liraglutide had the highest rate of adverse reactions among several GLP-1RAs.Still, in this literature, the authors also concluded that the incidence of GI adverse reactions with Dulaglutide was also higher.This differs from our findings, and we consider that this may be related to the fact that the studies we included were RCTs and not real-world studies and that the investigation by Li Chen et al. was based on the FDA adverse event reporting database, which differs from the data we included.Therefore, we believe that under both computational models, it can be assumed that Semaglutide and Liraglutide are more likely to cause intolerable GI adverse reactions in users.Dulaglutide causes GI adverse reactions but has a lower probability of being unbearable.In our study, we also found that there did not appear to be a statistically significant difference in intolerable GI AEs caused by Semaglutide weekly versus oral formulations.At the same time, the incidence of intolerable GI AEs caused by Semaglutide was positively correlated with the dose.In short, the higher the dose, the more the side effects.
Semaglutide, developed based on Liraglutide (90), showed similar results in this study.Semaglutide shares 94% identity with human GLP-1 (91), while Exenatide showed 53% homology to humanderived GLP-1 (92).However, Dulaglutide is also a human-derived GLP-1RA agent, indicating that Dulaglutide shares more than 90% identity with human GLP-1, so the intolerable GI AEs do not appear to be associated with an autoimmune response.Meanwhile, the GI AEs responsible for the intolerability of Semaglutide were unrelated to weekly or oral formulations.Drucker et al. (93) found that compared with the daily formula, the weekly recipe of exenatide has a lower risk of gastrointestinal adverse events, which is consistent with our research results.They also found no association between two Network evidence plots for intolerable GI AEs.The gastrointestinal tract may attenuate this feeling of satiety through specific reactions.We hypothesized that the stronger the appetite suppression of GLP-1RA preparations, the more likely it is to lead to intolerable gastrointestinal adverse effects.However, this is only our conjecture and needs to be verified by further experiments.Several methods were suggested to reduce the risk of GI AEs caused by GLP-1RAs.Choosing a small dose for the initial injection may reduce the risk of GI AEs (97), so pharmaceutical companies have developed a combination of GLP-1RA and insulin as a way to reduce the amount of GLP-1RA used (98, 99).This can both reduce the risk of hypoglycemic response caused by insulin and allow for better efficacy of GLP-1RAs.There are also different methods of injection to use Exenatide to reduce the incidence of intolerable GI AEs.However, this difference was not observed in our study.
Our study has some shortcomings.First of all, almost all the studies were lack of the dose of GLP-1RA when it was discontinued.Therefore, the current study could not conduct further Network Meta-Analysis according to the dose or the course of medication.Secondly, more than 25% of the RCTs included in this study had a sample size of fewer than 100 people, which made the OR and 95%CI of some outcomes with low accuracy; Thirdly, this study did not combine the curative effect with the comprehensive analysis of GLP-1RAs, but only analyzed from the direction of adverse reactions of the digestive tract; Fourthly, due to the lack of data, we only analyzed the correlation between the adverse reactions caused by Semaglutide and the dose.Whether other GLP-1RAs are consistent with semaglutide requires further validation in subsequent studies.
In conclusion, no single GLP-1RA has been proven to be superior across the board to the others.Each of these GLP-1RAs has its own merits (100), and physicians have the flexibility to choose the appropriate medicine based on the patient's actual situation.In terms of our findings, we recommend Dulaglutide, which has a low risk of intolerable GI AEs.

Conclusion
In conclusion, we found that Dulaglutide had the lowest risk of causing intolerable GI AEs.In contrast, Semaglutide and Liraglutide were associated with a higher risk of these AEs.We ruled out the possibility that the difference was due to drug homology or dosage 0.66 (0. Positive results were highlighted in blue.
Comparison-adjusted funnel plots for STEP1-4.Frontiers in Endocrinology frontiersin.orgform.However, no single GLP-1RA has been shown to be superior to the others.Each GLP-1RA has its advantages, and physicians have the flexibility to choose the appropriate drug according to the actual situation of the patient.In the future, we hope that more researchers will pay attention to these patients with adverse reactions in clinical trials, record their detailed data, and verify the reliability of our findings.

FIGURE 2
FIGURE 2Risk of bias summary for each item and the included studies.(+) green circle, favorable; yellow circle, moderate (-).

TABLE 2
(96)ork league table.Exenatide and the production of corresponding anti-exenatide antibodies.So what accounts for the difference in intolerable GI AEs between Semaglutide, Liraglutide, and Dulaglutide?We hypothesized that the incidence of intolerable GI adverse effects caused by GLP-1RA might be related to the degree of central appetite suppression because endogenous GLP-1 has a very short half-life in the body(94).It mainly acts in a paracrine form, while exogenous GLP-1 can cross the blood-brain barrier(95), bind to GLP-1 receptors in the brain, and stimulate continuously, unlike endogenous GLP-1.The sensation of satiety is caused by the stimulation of GLP-1 receptors in the brain by GLP-1 RAs entering the blood-brain barrier.Currently, the brain-gut axis is thought to be associated with several functional gastrointestinal disorders(96).The physiological responses produced by the brain in response to GLP-1RA stimulation affect the gastrointestinal tract via the brain-gut axis. 47,0.91)