Glucagon-like peptide 1 receptor agonists and health outcomes in inflammatory bowel disease

This study is a systematic review and meta-analysis, which was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines ().The study protocol was registered in the OSF Registry with the link. Supplementary Material 15 ¹⁸¹ https://doi.org/10.17605/OSF.IO/WABE5↗

The research question was formulated using the patient, intervention, comparator, and outcome (PICO) framework:

Patients (P): Individuals diagnosed with IBD, including CD and UC, who have coexisting obesity and/or T2DM.

Intervention (I): GLP1-RAs.

Comparator (C): Non-GLP-1 therapy or placebo.

Outcomes (O): Relevant clinical outcomes of IBD, including hospitalization rates, corticosteroid initiation, treatment escalation to advanced therapies, and surgical interventions. Secondary outcomes included weight loss and changes in biomarkers (eg, HbA1c, C-reactive protein [CRP], fecal calprotectin [FCP]).

A comprehensive literature search was performed using the databases PubMed, Embase, Web of Science, and Cochrane Library from inception to March 15, 2024. The following keywords and MeSH terms were used: “GLP-1 receptor agonist,” “inflammatory bowel disease,” “Crohn’s disease,” “ulcerative colitis,” “obesity,” “type 2 diabetes,” “hospitalizations,” “remission,” “biomarkers,” and “weight loss.” Boolean operators (AND, OR) and filters (eg, human studies, English language) were applied to refine the search. The search strategy is detailed in. Supplementary Material 1

Inclusion criteria:

Exclusion criteria:

Two independent reviewers (A.B. and M.T.) extracted data using a standardized data collection form. Extracted information included study characteristics (eg, author, year, design, and sample size), patient demographics, intervention details (type of GLP1-RA), comparator characteristics, and reported outcomes. Discrepancies were resolved through discussion or consultation with a third reviewer (N.d.B.). Risk of bias in included observational studies was assessed using the ROBINS-I tool (risk of bias in non-randomized studies of interventions) by 2 reviewers (A.B. and M.T.).ROBINS-I evaluates 7 domains: bias due to confounding, participant selection, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes, and selection of the reported result. Two reviewers independently assessed each study, disagreements were resolved with a third reviewer (N.d.B.). ¹⁸¹

A meta-analysis was conducted using Review Manager (RevMan) software. Both effect sizes (hazard ratios [HRs]) and dichotomous data (odds ratios [ORs]) were analyzed for hospitalizations, surgeries, corticosteroid initiation, and advanced therapy initiation across IBD, CD, and UC cohorts. The definitions of the outcomes of the included studies can be found in. We use both effect sizes and dichotomous data from different studies to assess whether inconsistencies occurred between the different statistical methods. An additional analysis was performed for intestinal obstruction using dichotomous data. The random-effects model was applied because of expected heterogeneity among studies. Vote counting () was performed by assessing the direction of effect for each of the different meta-analyses. We deemed a significant finding if half or more of studies showed a similar direction of effect. Supplementary Material 2 Table 1

Heterogeneity was assessed using thestatistic, with values of 25%, 50%, and 75% representing low, moderate, and high heterogeneity, respectively. Funnel plots were used to visually inspect potential publication bias, and Egger’s regression test was performed where applicable. I 2

Sensitivity analyses were performed based on IBD subtype (CD or UC), BMI and GLP1-RA drug (for weight loss). Other subgroup analysis such as presence of T2DM, and visceral adipose tissue were not possible due to lack of data. All statistical analyses were performed using RevMan. A-value of 0.05 was considered statistically significant. P

Ethical approval was not required for this study as it involved the synthesis of previously published data.

This systematic review included 11 observational studies ­() evaluating the effects of GLP1-RAs on clinical outcomes in patients with IBD complicated by T2DM and/or obesity. A total of 16 242 patients with IBD using GLP1-RAs were analyzed, with representation from both CD and UC cohorts (). The meta-analyses from those patients can be found inand. The studies were conducted across diverse settings, including academic hospitals in the United States, Spain, and Israel, as well as nationwide registries in Denmark, Israel, and the United States (). The summary of clinical and metabolic outcomes can be found inand. The results of the risk of bias assessments can be found in. Figure 1 Table 2 Figures 2-7 Supplementary Materials 3-5 Table 2 Tables 1 3 Figure 2

The impact of GLP1-RAs on hospitalization rates for IBD was similar to controls. The meta-analysis using effect sizes showed that GLP1-RA therapy lead to similar rates of hospitalizations in patients with IBD compared to non-GLP1-RA therapy (HR: 0.91, 95% confidence interval [CI]: 0.71-1.17,= 69%,= 0.47) (). Sensitivity analysis for UC (HR: 0.88, 95% CI: 0.66-1.18,= 51%) and CD (HR: 0.93, 95% CI: 0.78-1.10,= 0%) showed similar trends (). In the meta-analysis using event frequencies, similar hospitalization events in the GLP1-RA group (OR: 0.57, 95% CI: 0.19-1.68,= 98%) () compared to controls were found. Sensitivity analysis () for BMI revealed that patients with BMI ≥ 30 and GLP1-RA therapy (logHR: 0.79, 95% CI: 0.66-0.96,= 37%,= 0.01) had reduced risks of hospitalization. For patients with BMI < 30 and GLP1-RA therapy, the point estimate also suggested reduced risk compared to no GLP1-RA therapy and BMI <30 (logHR: 0.81), but the 95% CI (0.62-1.06) included 1.0, indicating the result was not statistically significant (= 41%;= 0.12). I P I I I I P I P 2 2 2 2 2 2 Figure 3 Supplementary Materials 3 and 4 Figure 4 Figure 5

The potential impact of GLP1-RAs on the risk of surgery was positive. In meta-analysis, the pooled HR for surgery was in favor of GLP1-RA therapy compared to non-GLP1-RA therapy (logHR: 0.61, 95% CI: 0.44-0.84,= 0%,= 0.003) in patients with IBD (. No differences were found in the sensitivity analysis for UC (logHR: 0.79, 95% CI: 0.43-1.45,= 0%,= 0.44), and for CD (logHR: 0.68, 95% CI: 0.37-1.26,= 53%,= 0.22) (). Using event frequencies, surgery events were significantly lower for the GLP1-RA group in patients with IBD (OR: 0.46, 95% CI: 0.32-0.67,= 42%) (). Sensitivity analysis () for BMI showed that patients with BMI ≥ 30 and GLP1-RA therapy had reduced risk for surgery (logHR: 0.42, 95% CI: 0.27-0.67,= 0%,= 0.0003), while patients with BMI < 30 and GLP1-RA therapy had no reduced risk for surgery compared to no GLP-RA therapy and BMI < 30 (logHR: 1.05, 95% CI: 0.55-2.03,= 25%,= 0.88). I P I P I P I I P I P 2 2 2 2 2 2 Figure 3 Supplementary Materials 3 and 4 Figure 4 Figure 5 )

The effect of GLP1-RAs on corticosteroid initiation varied across studies. Meta-analysis showed no effect of GLP1-RA on corticosteroid initiation in patients with IBD (logHR 1.02, 95% CI: 0.86-1.22,= 46%,= 0.79) (). Similar results were obtained in the sensitivity analysis for UC (logHR 1.02, 95% CI: 0.73-1.42,= 40%,= 0.91) and CD (logHR 0.90, 95% CI: 0.56-1.47,= 74%,= 0.68) (). Meta-analyses using event frequencies showed that corticosteroid initiation was not different between GLP1-RA and control groups among patients with IBD (OR: 0.69, 95% CI: 0.25-1.93,= 99%,= 0.48) (). Sensitivity analysis () for BMI showed that in patients with BMI < 30 and GLP1-RA therapy there was lower risk of corticosteroid initiation (logHR: 0.71, 95% CI: 0.57-0.88,= 0%,= 0.002). The point estimate also indicated lower risk for corticosteroid initiation in patients with BMI ≥ 30 and GLP1-RA therapy, although this was not statistically significant due to broad confidence intervals (logHR: 0.76, 95% CI: 0.57-1.01,= 68%,= 0.06). I P I P I P I P I P I P 2 2 2 2 2 2 Figure 3 Supplementary Materials 3 and 4 Figure 4 Figure 5

Meta-analysis showed no effect of GLP1-RA on advanced therapy initiation (ie, biologicals and/or small molecule therapies) in IBD (logHR: 0.96, 95% CI: 0.74-1.23,= 15%,= 0.72) ­(). Advanced therapy initiation did not differ for UC (logHR: 0.78, 95% CI: 0.53-1.14,= 0%,= 0.20), and CD (logHR 1.10 [95% CI: 0.77-1.55],= 28%,= 0.61) (). The event frequency meta-analysis showed that patients with IBD treated with GLP1-RAs had similar advanced therapy initiation (OR: 0.78, 95% CI: 0.41-1.49,= 85%,= 0.44) events compared to non-GLP1-RA treatment (. Sensitivity analysis () for BMI did not show any difference between GLP1-RAs and controls for patients with BMI ≥ 30 (logHR: 0.86, 95% CI: 0.53-1.41,= 0.56) and BMI < 30 (logHR: 1.02, 95% CI: 0.67-1.57,= 0.92). I P I P I P I P P P 2 2 2 2 Figure 3 Supplementary Materials 3 and 4 Figure 4 Figure 5 )

Meta-analyses were performed for different GLP1-RA therapies (semaglutide, liraglutide, dulaglutide, and tirzepatide) for weight loss in IBD patients (). Statistical significant weight loss was achieved in IBD patients using semaglutide (−9.1 kg, 95% CI: −11.8; −6.4,= 12%,< 0.00001,= 442), liraglutide (−9.0 kg, 95% CI: −12.7; −5.3,= 14%, P < 0.00001,= 171), and tirzepatide (−11.6 kg, 95% CI: −18.3; −4.8,= 0%,= 0.0008,= 113). No statistical significant weight loss was achieved using dulaglutide (−3.9 kg, 95% CI: −9.7; −1.9,= 0%,= 0.19,= 124). No meta-analysis was performed for exenatide, although 3 studies reported on exenatide, the study by Clarke et al.only had 1 patient with follow-up. Furthermore, Levine et al.showed that weight loss could be achieved after 12 months of follow-up in IBD patients with T2DM (BMI 33.5, 95% CI: 29.9-37.0 baseline vs. BMI 31.5, 95% CI: 27.1-35.8 follow-up,< 0.01) and without T2DM (BMI 33.2, 95% CI: 29.3-38.4 baseline vs. BMI 32.6, 95% CI: 28.6-37.5 follow-up,= 0.04), and that significant weight loss could only be achieved in patients with obesity (BMI ≥ 30). Figure 6 ^{181 181} I P n I n I P n I P n P P 2 2 2 2

Intestinal obstruction has been reported in 3 studies in relation to GLP1-RA use.The meta-analysis () showed that in these 3 included studies, the number of events was 286 (3.0%) and 2091 (3.4%) in exposed (= 9571) and non-exposed groups (= 62249), respectively. The OR derived from the meta-analysis for intestinal obstruction was not different between GLP1-RA therapy and controls (OR: 0.51, 95% CI: 0.21-1.23), (= 95%,= 0.13). ^{181 181 181} Figure 7 , , 2 n n I P

shows the summary of the meta-analyses for the different IBD-related outcomes. Vote counting was used for each of the individual analyses to determine the direction of the outcome. For hospitalizations, 2 out of 6 analyses were in favor of GLP1-RA therapy, while the other 4 analyses did not show any difference between GLP1-RA and controls. Vote counting for surgery showed that 3 out of 6 were in favor of GLP1-RA therapy. The majority of studies did not show any difference between GLP1-RA and controls for corticosteroid initiation (5 of 6) and advanced therapy initiation (6 of 6). Table 1

The effect of GLP1-RAs on disease activity scores and biomarkers was mixed. Anderson et al.reported no significant improvement in disease activity scores over 12 months, with only marginal reductions in the Harvey–Bradshaw Index (HBI) and Mayo Score (HBI: 3.52-3.18,= 0.61; Mayo: 1.61-1.54,= 0.67). Similarly, Ramos Belinchón et al.found no significant changes in HBI or partial Mayo scores after 6 months of GLP1-RA treatment. ^{181 181} P P

Inflammatory markers, particularly CRP, showed variable responses to GLP1-RA therapy (). Anderson et al.observed a significant reduction in CRP levels from 12.9 mg/L at baseline to 6.4 mg/L after 12 months (= 0.005), suggesting a potential anti-inflammatory effect of GLP1-RAs. Conversely, St-Pierre et al. reported no significant change in CRP levels, with median values remaining stable at 3 mg/L (interquartile range [IQR] 3-6.5) throughout the study period (= 0.21). Levine et al.observed a modest reduction in CRP levels in patients with IBD, from 6.9 mg/L (IQR: 3.1-11.4) to 5.5 mg/L (IQR: 2.8-11.6), though statistical significance was not reported. Supplementary Material 5 ^{181 181} P P

FCP, a biomarker of intestinal inflammation, showed mixed results across studies (). Ramos Belinchón et al.reported no significant change in FCP levels after 6 months of therapy, with median levels decreasing slightly from 41.0 µg/g (IQR: 14-80) to 32.7 µg/g (IQR: 13.8-51.0) (= 0.74). Sehgal et al.observed a decrease in mean FCP levels from 896 µg/g (SD: 2600) at baseline to 705 µg/g (SD 2297) after 12-24 weeks, though the variability in measurements was high. Desai et al.found no significant differences in mean FCP levels between GLP1-RA users and controls in either the UC (324.7 ± 533 vs. 369.7 ± 624,= 0.67) or CD groups (260 ± 391 vs. 324 ± 487,= 0.29). Supplementary Material 5 ^{181 181 181} P P P

Glycemic control, as measured by HbA1c, improved significantly in a few studies (). No meta-analysis was performed because it was unclear which GLP1-RA caused the decrease in HbA1c value. Levine et al.reported a decrease in HbA1c from 6.5% (IQR: 5.9-7.4) at baseline to 6.2% (IQR: 5.4-7.0) after 12 months of therapy in patients with IBD (= 0.01). Similarly, patients without IBD in the same cohort experienced a comparable reduction. Ramos Belinchón et al.,however, reported no significant change in HbA1c levels, with a reduction from 6.2% (IQR: 5.5-7.8) to 5.9% (IQR: 5.3-7.58) (= 0.14), suggesting variability in glycemic outcomes across studies. Supplementary Material 5 ^{181 181} P P

Obesity has been recognized as a risk factor for hospitalizations in patients with IBD. In 1 nationwide cohort study, 5128 obese patients with IBD were matched to 5128 nonobese IBD patients using propensity score matching. Obese patients with IBD were more likely to have preventable readmissions compared to nonobese patients with IBD (19% vs. 15%), and obese patients with IBD spend more days in the hospital annually (median: 8 vs. 5 days,< 0.01) with higher hospitalization-related costs (median: $17 277 vs. $11 847,< 0.01).In patients with IBD that were hospitalized, obesity was also associated with longer duration of stay (mean: 5.5 vs. 4.9 days,< 0.001) and mean hospital costs ($50 126 vs. $45 001,< 0.001).Patients with IBD and obesity had lower steroids-free clinical remission at week 24 in a prospective registry study of patients starting with new IBD treatment in regular care in the Netherlands.In UC patients, it was reported that each increase of BMI (1 kg/m) was associated with 4% increase in the risk of treatment failure (aHR: 1.04, 95% CI: 1.00-1.08) in patients treated with biologicals and 8% increase in the risk of surgery/hospitalization (aHR: 1.08, 95% CI: 1.02-1.14).In this systematic review, GLP1-RAs were found to reduce weight in our meta-analysis by approximately 10 kg (∼TBW 10%) after >3-6 months of follow-up in IBD patients treated with semaglutide, liraglutide, or tirzepatide. Furthermore, GLP1-RAs reduced the risks of hospitalizations and surgery, in patients with IBD and obesity. We would like to emphasize that BMI alone has been described as an independent risk factor for surgery and hospitalization.Our findings suggest that GLP1-RA may reduce these risks in patients with IBD and BMI ≥ 30. This is specifically relevant considering higher rates of medical complications during hospitalization and higher financial costs for admission in patients with obesity. P P P P ^{181 181 181 181 181 181} 2 ,

In this study, GLP1-RAs did not reduce advanced therapies or corticosteroid initiation, although sensitivity analysis showed benefits of GLP1-RAs in reducing corticosteroid initiation in patients with and without obesity. Yarur et al.showed that poor response to biological therapy (infliximab, vedolizumab, or ustekinumab) was associated with a higher intra-abdominal visceral adipose tissue compartment as a percent of total body mass. Similarly, He et al.showed that non-responders to corticosteroid therapy had higher visceral adipose tissue compared to responders in CD (21.3 vs. 11.6 cm/m,= 0.004). Although for patients with UC, non-responders had lower visceral adipose tissue compared to responders (14.7 vs. 24.4 cm/m,= 0.001). In this systematic review, abdominal visceral fat was not determined in the included studies, hence there is a need for more studies that assess the effects of GLP1-RAs on abdominal visceral fat in relation to corticosteroid and biological therapies. ^{181 181} 2 2 2 2 P P

The meta-analysis showed that patients using GLP1-RA had similar rates of intestinal obstruction compared to non-GLP1-RA therapies (1.0% [= 61] vs. 3.1% [= 1,817]). GLP1-RAs were thought to be associated with higher risks of intestinal obstruction or ileus due to its effects on transit times and intestinal motility due to GLP-1-induced increased intestinal length and villus height.Sodhi et al.reported that GLP1-RAs were associated with increased risks of bowel obstruction (HR: 4.22, 95% CI: 1.02-17.4) compared to bupropion-naltrexone in patients with obesity. However, a Danish study by Ueda et al.in patients with T2DM using GLP1-RAs showed no increased risk of intestinal obstruction (HR: 0.83, 95% CI: 0.69-1.01). The data presented by Ueda et al.suggests that the risks of intestinal obstruction in patients with IBD using GLP1-RA are not increased, which is similar to the finding of this systematic review. n n ^{181 181 181 181 181} ,

This systematic review with meta-analysis has several limitations. The included studies were heterogeneous in design, population characteristics, and outcomes which make it challenging to draw consistent conclusions. Most studies were observational or retrospective, introducing unmeasured potential biases and limiting causal inferences. Furthermore, studies had missing data, did not report clinical outcomes, clinical scores, or BMI/visceral adiposity and data on side-effects were lacking. RCTs are needed to confirm the potential efficacy and safety of GLP1-RAs in IBD. While weight loss and metabolic benefits were consistently observed, the impact of BMI or visceral adiposity on IBD-specific outcomes were not reported consistently. The meta-analyses yielded heterogenous and inconsistent results across other clinical outcomes and sensitivity analyses (eg, BMI and IBD-type). Meta-analyses conducted in this study exhibited moderate to high heterogeneity, suggesting variability in patient populations and reported outcomes. It is important to note that the studies by Desai et al.,Niu et al.,and Alchirazi et al.were performed using the triNetX database. Considering that the studies by Niu et al.and Alchirazi et al.were conference abstracts and may have included patient populations overlapping with those reported by Desai et al.,these studies were not included. Another limitation is that most studies did not report or account for BMI or a measurement of the visceral adipose tissue compartment, which may significantly affect the results. Future research should prioritize RCTs evaluating GLP1-RAs in well-defined IBD populations, stratified by subtype, obesity, and metabolic status. Studies should also assess long-term outcomes, including durability of response, side effects, and explore mechanistic pathways to better understand how GLP1-RAs may influence IBD pathophysiology. Additionally, the potential synergistic effects of combining GLP1-RAs with existing therapies warrant investigation. ^{181 181 181 181 181 181}

Unaddressed issues in this review are the side-effects associated with GLP1-RA, including constipation, gastroparesis, and pancreatitis,and potential drug interactions with other medications like warfarin and thiopurines.A shift toward an “obesity first” approach could reshape IBD care by prioritizing obesity as a key underlying factor. By 2030, we anticipate gaining clearer insights into whether GLP-1 therapies will become a long-term treatment option for obesity and potentially for IBD as well. Currently, 2 clinical trials under investigation for GLP1-RAs combined with mirikizumab in IBD.While the current high cost of these medications limits accessibility and research, the introduction of generics may improve availability and affordability. ^{181 181 181 181 181} , ,

Generic pharmaceutical companies are expected to significantly scale up production once patents expire, which also depends on patent and regulatory exclusivities.However, the future landscape remains uncertain, as it is difficult to predict how other pharmaceutical companies will innovate within the GLP-1 space, whether through combination therapies or other advancements. ¹⁸¹