Impact of GLP-1 Receptor Agonists in Gastrointestinal Endoscopy: An Updated Review

Stark et al. conducted a retrospective cohort study of patients on GLP-1 RAs (n = 59) undergoing esophagogastroduodenoscopy (EGD) and compared them with matched controls (n = 118) (Figure 1 and Table 1) [28]. Retained food content was found in four patients in the GLP-1 RA group vs. two in the control group, without any significant difference between the groups (6.8% vs. 1.7%, p = 0.08). One patient in each group required lavage (1.7% vs. 0.8%, p = 0.62), and none of the patients needed repeat EGD. The study was limited due to low sample size, possible missed recording of retained food, variable dose, treatment duration, and adherence to GLP-1 RAs.

In a single-arm retrospective analysis by Firkins et al., clinical outcomes in GLP-1 RA users undergoing EGD were assessed [29]. Of the 1512 EGDs included in the study, 142 were found to have RGCs (9.4%), with solid retained contents being present in 112 cases (7.4%) and liquid contents in 27 cases (1.6%). Around 93 patients (11.6%) had upper GI symptoms such as nausea, vomiting, and abdominal pain. The procedure was aborted in 30 cases (2%), with 21 cases requiring a repeat endoscopy (1.4%). Four patients had intraprocedural hypoxia (0.3%), two experienced aspiration (0.1%) and three (0.2%) had other adverse events such as bronchospasm and cholangitis. Univariate analysis revealed that younger age was associated with an increased RGC risk (p = 0.003), whereas colonoscopy performed on the same day exerted a protective effect against RGC (p < 0.001). Multivariate analysis confirmed the roles of young age (p = 0.026) and same-day colonoscopy (p < 0.001). While some GLP-1 RAs, including tirzepatide and oral semaglutide, demonstrated a higher RGC risk in univariate analysis, this was not reflected in the subsequent multivariate results. Timing of the procedure or anesthesia type did not have any individual effect over the RGCs. Procedure discontinuation and requirement of repeat endoscopy were significantly associated with the presence of RGCs (p < 0.001 for both).

In an observational study from the Mayo Clinic Health System, Anazco et al. found only two cases of pulmonary aspiration among 4134 endoscopies in patients on GLP-1 RAs, amounting to a rate of 4.8 per 10,000 endoscopies [30]. This rate was similar to the previous study from the same hospital system conducted in a period before GLP-1 RAs became popular (2010–2016), with 4.6 cases of aspiration per 10,000 endoscopies. In another case series by Maselli et al., 57 patients on GLP-1 RAs scheduled for endoscopic sleeve gastroplasty were instructed to maintain a liquid-only diet for 24 h and a fasting period of 12 h prior to the procedure [31]. No cases of retained gastric solids, pulmonary aspiration, regurgitation of gastric contents, or hypoxia were observed.

Silveira et al. performed a retrospective analysis of patients having elective upper endoscopy and stratified the cases based on semaglutide use within 30 days before the procedure [25]. Of the 404 patients—with 33 in the semaglutide group and 371 in the non-semaglutide group—increased RGC was higher in the former group (24.2% vs. 5.1%, p < 0.001). In the unadjusted analysis, increased RGC was found to be associated with semaglutide use (p < 0.001) and ongoing GI symptoms (p < 0.001). On the contrary, RGCs were lower in patients who had combined upper endoscopy and colonoscopy (p = 0.011). Patients with or without increased RGCs had no difference pertaining to the fasting duration for clear liquids (p = 0.084) or solids (p = 0.457). Subsequent propensity weighted analysis also showed similar effects of semaglutide use, GI symptoms, and combined procedures on RGCs (p < 0.001 for all). Interestingly, the amount of RGC found did not correlate with semaglutide use (p = 0.99). Further, only one aspiration event was noted; the patient was on semaglutide prior to the endoscopy but did not have any ongoing GI symptoms.

In a larger sample size study by Nadeem et al., conducted at Geisinger Medical Center (2019–2023), a total of 35,183 patients undergoing upper endoscopy were divided into GLP-1 (n = 922) and non-GLP-1 groups (n = 34,261) [32]. The former group had older patients (57.1 ± 12.9 vs. 53.9 ± 17.5 years, p < 0.0001), with a higher average body mass index (BMI) (36.4 ± 8.9 vs. 30.5 ± 7.8 kg/m², p < 0.0001) and proportion of diabetic patients (756 vs. 5407, p < 0.0001). As compared to the patients not on GLP-1 RAs, those using these medications had higher RGC rates (13.6% vs. 2.3%, p < 0.0001), aborted procedures (1.5% vs. 0.3%, p < 0.0001), and requirements for repeat EGD (2.4% vs. 1.1%, p = 0.0001). Multiple logistic regression performed after adjustments for demographics and comorbidities confirmed the higher risk of RGCs (4 times), aborted procedures (4 times), and need for repeat EGD (2 times) among patients on GLP-1 RAs. Upon stratifying the patients based on presence of diabetes, GLP-1 RAs were associated with increased RGCs and aborted procedures in both diabetic and non-diabetic patients. While repeat EGDs were higher among GLP-1 RA users who were diabetic, there was no significant difference compared to the non-GLP-1 RA group in patients without diabetes (p = 0.655). Postprocedural adverse events such as bronchospasm did not differ between the GLP-1 RA users and non-users (0.2% vs. 0.2%, p = 0.707). The one aspiration event reported was in the non-GLP-1 RA cohort.

In the 1:1 matched case-control study by Chapman et al., patients scheduled for EGD and on GLP-1 RAs were compared with controls (not on GLP-1 RAs) [33]. Among the 84 patient pairs included, the GLP-1 RA group had higher BMI (40.70 ± 13.29 vs. 31.23 ± 10.65 kg/m², p < 0.001), no underweight patients (0 vs. 7.1%, p < 0.001), and a greater number of patients who received monitored anesthesia care (MAC) (76.2% vs. 58.3%, p = 0.014). Clean gastric mucosa was less common (57.1% vs. 67.9%) along with higher RGCs (13.1% vs. 4.8%, p = 0.025) in the GLP-1 RA users. Similarly, the gastric mucosal visibility score (POLPREP) was lesser in patients on GLP-1 RAs compared to the control group (2.14 ± 1.03 vs. 2.57 ± 0.74, p = 0.0012). Further, BMI increase by every unit led to a lower visibility score (p < 0.001). Logistic regression revealed the association of GLP-1RA use with lower mucosal visibility score (p < 0.001), without any significant difference among the various GLP-1 RA medications. While the GLP-1 RA group had a higher number of aborted EGDs (4.8% vs. 0, p = 0.043), no differences were observed compared to the non-GLP-1 RA group with respect to adverse events (procedure- or anesthesia-related) or mortality at 30 days.

Barlowe et al. conducted a retrospective cohort study based on the nationwide claims database of diabetic patients who had an upper endoscopy during the period of 2005–2021 [34]. The outcomes were compared between GLP-1 RA (n = 15,119), dipeptidyl peptidase 4 inhibitor (DPP4i, n = 14,407), and chronic opioid (n = 7257) users. The opioid group had a higher comorbidity burden, including asthma, chronic obstructive pulmonary disease (COPD), and 2+ Charlson Comorbidity Index. The three groups had comparable rates of aspiration (0.05% vs. 0.07% vs. 0.11%) and aspiration pneumonia (0.07% vs. 0.07% vs. 0.11%). The GLP1-RA group had a lower risk of pneumonia (0.18% vs. 0.59%), respiratory failure (0.10% vs. 0.41%), and composite pulmonary adverse events (0.31% vs. 1.01%) when compared with the chronic opioid group. Further, post-EGD hospitalization and emergency room visits were both lower in patients on GLP-1 RAs vs. DPP4is/opioids (2.22% vs. 2.33%/4.93% and 4.15% vs. 4.94%/10.69%, respectively). Sensitivity analyses accounting for baseline risk factors and stricter use criteria showed consistent results.

In the multicenter cross-sectional study by Phan et al., patients (n = 815) who underwent upper endoscopy while being on GLP-1 RAs were analyzed [35]. RGCs were present in 8.6% of cases, with the amount being minimal in 14.3%, moderate in 44.3%, and large in 41.4% of patients. Unplanned intubation was required in only one patient due to a large RGC, and the procedure was aborted in fourteen patients. As compared to those without RGCs, patients with RGCs had higher odds of being diabetic (p = 0.009) and on insulin therapy (p < 0.01), with lesser likelihood of holding GLP-1 RAs (p < 0.001). Among diabetic patients, male gender (p = 0.01) and absence of gastroparesis (p = 0.004) were associated with RGCs. Furthermore, those with uncontrolled diabetes had higher chances of having RGCs compared with other diabetic groups (p = 0.006). When grouped based on the ASA recommendations, patients who held their GLP-1 RAs had lesser RGCs compared to those who did not hold (4.4% vs. 12.7%, p < 0.001); however, holding the GLP-1 RAs did not affect the quantity retained (p = 0.37), intubation rate (p = 0.53), or procedure abortion rate (p = 0.40). Multivariate analysis showed an association of RGCs with not holding GLP-1 RAs (p < 0.001), an HbA1C increase by every 1% (p < 0.001), and preoperative glucose levels >150 mg/dL (p = 0.002).

Garza et al. conducted a case-control study where patients on GLP-1 RAs undergoing upper endoscopy had a higher rate of gastric residue as compared with controls (14% vs. 4%, p < 0.01) [36]. This was further shown in the subgroups of diabetic patients (14% vs. 4%, p < 0.01), those on insulin (17% vs. 5%, p < 0.01), and those with diabetic complications (15% vs. 2%, p < 0.01). Prolonged fasting and maintenance of a clear liquid diet for colonoscopy (p < 0.01) and performing the procedure in the afternoon (p < 0.01) lowered the risk of gastric residue. No aspiration events occurred in any of the procedures.

In another cohort study by Wu et al., gastric contents were observed in 18.7% of EGDs in patients on GLP-1 RAs vs. 4.9% in the control group (p = 0.004) [37]. The GLP-1 RA group had more procedures with endotracheal anesthesia (21% vs. 11%), with an additional four cases being converted from MAC due to presence of RGCs (6% vs. 0). One aspiration event was recorded in the GLP-1 RA group vs. none in the control group. Differentiating the GLP-1 RA users into groups of patients on regular diet vs. clear liquid/low-residue diet for 24 h prior to the endoscopy, with fasting from midnight, Ghazanfar et al. reported a higher residual food rate among those on regular diet (10% vs. 1.5%, p = 0.03) [38]. Further, four patients who had residual food and ongoing GI symptoms were on a regular diet. No aspiration events were reported.

In an observational study by Tong et al., patients with diabetes undergoing colonoscopy (bowel preparation with polyethylene glycol) were divided into liraglutide (GLP-1 RA), sitagliptin (DPP-4i), and control groups [39]. Inadequate bowel cleaning was not statistically different across the three groups (17.5% vs. 20.5% vs. 21.7%, p = 0.927); similarly, the mean Boston Bowel Preparation Scale (BBPS), cecal intubation time, and rates of polyp detection were comparable (p > 0.0.05). However, liraglutide use was associated with increased GI symptoms (such as nausea, vomiting; p < 0.05) and a higher incidence of inadequate bowel cleaning in a subgroup of patients with diabetic peripheral neuropathy (61.3% vs. 32.1% sitagliptin, p = 0.022; 61.3% vs. 32.8% control, p = 0.025). Along the same lines, Yao et al. reported patients in the GLP-1 RA group to have a lower mean BBPS (p = 0.046), along with a higher proportion of patients with BBPS < 5 (p = 0.01) and those requiring repeat colonoscopy due to inadequate bowel preparation (p = 0.041) [40].

Nakatani et al. assessed GI motility and residue rates via capsule endoscopy in 14 diabetic patients before and after administration of liraglutide [41]. While the gastric transit time in patients with diabetic neuropathy (DN) was not impacted significantly due to liraglutide (p = 0.19), it was significantly increased after liraglutide administration in the non-DN group (p = 0.03). A similar trend was noticed in the duodenal and small intestine transit time as well. GI residue rate was higher in both DN and non-DN groups after receiving liraglutide (p < 0.001). In the recent matched cohort study of patients with diabetes undergoing capsule endoscopy by Odah et al., the GLP-1 RA group had increased gastric transit time (99.3 ± 134.2 min vs. 25.3 ± 31.6 min, p < 0.001) and a greater rate of incomplete passage through the small intestine (23.5% vs. 4.4%, p < 0.01) as compared with the control group [42].

Hiramoto et al. conducted a meta-analysis to evaluate the effect of GLP-1 RAs on gastric emptying [8]. Among the studies quantifying gastric emptying by scintigraphy, GLP-1 RA use caused longer t1/2 compared to placebo (138.4 vs. 95.0 min, p < 0.01). While the type of GLP-1 RA formulation or duration of treatment did not affect gastric emptying (scintigraphy), it was significantly increased in obese patients with BMI ≥ 30 kg/m² (p = 0.018). No significant difference was noted between GLP-1 RA and placebo groups with respect to gastric emptying measured via the acetaminophen absorption test (p = 0.432). The study concluded that a gastric emptying delay of around 36 min (scintigraphy) in patients on GLP-1 RAs would not have a great impact on the standard fasting time of 6–8 h, and hence these medications can be continued in the perioperative period. Further, a liquid diet does not delay gastric emptying (as seen with the acetaminophen absorption test) and can be given on the day prior to the endoscopy.

In the meta-analysis by Facciorusso et al., 13 studies (84,065 patients) were included demonstrating the effects of GLP-1 RAs on upper endoscopy [43]. As compared to non-users, the GLP-1 RA users had significantly higher RGC rates (odds ratio [OR] 5.56, 95% CI 3.35–9.23), and the results were consistent in the subgroup analysis of full-text papers—those with propensity score matching, diabetic patients, and fasting duration (greater or less than 12 h). While the GLP-1 RA group had a higher rate of aborted procedures (1% vs. 0.3%) and need for repeat procedures (2% vs. 1%), no significant differences were noted with regards to adverse events (0.3% vs. 0.1%) or bronchial aspiration (0.3% vs. 0.2%).