RE-EXAMINING THE RISKS OF GASTROINTESTINAL ADVERSE EVENTS ASSOCIATED WITH GLUCAGON-LIKE PEPTIDE-1 RECEPTOR AGONISTS FOR WEIGHT LOSS WITH MORE STRINGENT CRITERIA ON A COMPREHENSIVE LARGE DATABASE.

Introduction:
Sodhi et al. (2023) recently reported that glucagon-like peptide-1 receptor agonists (GLP-1 RA) usage for weight loss was associated with increased risks of pancreatitis, gastroparesis, and bowel obstruction but not biliary disease when compared to bupropion-naltrexone (BN). One notable limitation was the lack of standardization for initial BMI and an inability to examine BMI change. We sought to re-examine this association with a more stringent criterion that includes BMI covariates.
Methods:
We identified patients (2011- 2023) in the TriNetX healthcare database who had a BMI of ≥ 30 or a diagnosis of obesity 90 days prior or up to 30 days after cohort entry . GLP-1 RA cohort entry was defined as the date of new use of semaglutide or liraglutide (Cohort 1) with the active comparator being the date of new use of BN (Cohort 2), with both cohorts being exclusive. BN users were explicitly documented as taking “Contrave” or were taking BN at the weight loss dosage ratio of 8mg Naltrexone/90mg Bupropion. Cohorts were limited to USA. Patients could never have a diagnosis of type 2 diabetes nor an A1c ≥ 6.5%. Kaplan-Meier and Cox-proportional hazards models were performed on matched and unmatched cohorts. 1:1 propensity score matching included demographics, alcohol use, smoking, hyperlipidemia, abdominal surgery, and BMI. A second analysis specifically did not match BMI. Patients were observed until the first-ever occurrence of acute pancreatitis, bowel obstruction, gastroparesis, or biliary disease or censored on last day of follow-up or death. Proportionality was tested using Schoenfeld residuals.
Results:
Cohort 1 had 105,793 patients while Cohort 2 had 8,794 patients. Unmatched Cohort 1 had an initial average BMI of 37.9 ± 6 SD while Cohort 2 had 36.9 ± 5.8. One year after initiation, the average BMI in Cohort 1 was 36.9 ± 6.1 and 36.4 ± 5.9 in Cohort 2. After propensity matching, there were 8,792 patients in each cohort. The initial average BMI was 37.2 ± 5.9 for Cohort 1 and 36.9 ± 5.8 for Cohort 2. One year after, the average BMI in Cohort 1 was 36.4 ± 5.9 and 36.4 ± 5.9 in Cohort 2. We did not observe any relationships between GLP-1 RAs and acute pancreatitis [adjusted HR 1.19; 95% CI 0.66 – 2.14] nor bowel obstruction [HR 1.30; 95% CI 0.69 – 2.18]. We did observe an increased risk of biliary disease [HR 1.27; 95% CI 1.02 – 1.59] and gastroparesis [HR 2.30; 95% CI 1.19 – 4.46]. In the second analyses without BMI matching, we no longer observed an increased risk of biliary disease [HR 1.21; 95% CI 0.96 – 1.52]. There were no violations of proportionality assumption as confirmed by Schoenfeld residuals.
Discussion:
We observed different associations than Sodhi et. al (2023) except for gastroparesis. Notably, all confidence intervals of adjusted hazards ratios overlap their reported outcomes. Some differences may be explained by initial BMI.