Gallbladder‐Related Adverse Events With Semaglutide 2.4 mg: Pooled Analysis of Placebo‐Controlled STEP Trials

1. Background

Glucagon‐like peptide‐1 receptor agonists (GLP‐1 RAs) are widely prescribed for chronic weight management. Semaglutide 2.4 mg (once‐weekly subcutaneous) achieves mean weight loss of approximately 15% in people with overweight or obesity [1]. An increased risk of gallbladder and biliary events has been documented across GLP‐1 RA class meta‐analyses, attributed both to rapid weight loss‐induced changes in bile composition and to direct GLP‐1‐mediated inhibition of gallbladder motility [2, 3]. However, estimates specific to semaglutide 2.4 mg remain limited by small event counts in individual trials. We therefore conducted a pooled analysis of the placebo‐controlled randomized trials to estimate the incidence and relative risk of gallbladder‐related events attributable to semaglutide.

2. Methods

The STEP programme constitutes the complete phase 3 randomized evidence base for semaglutide 2.4 mg in weight management, with consistent dosing, populations and safety reporting. No other placebo‐controlled randomized trials of semaglutide 2.4 mg in this indication with comparable design were available. Consequently, we pooled gallbladder‐related event data from STEP trials (STEP 1–5) evaluating once‐weekly subcutaneous semaglutide 2.4 mg in adults (BMI ≥ 30 kg/m², or ≥ 27 kg/m² with comorbidity) with available safety data [4, 5, 6, 7]. Trials of semaglutide at lower doses in type 2 diabetes and studies of other GLP‐1 receptor agonists were not included due to differences in dose, population and pharmacological profile.

Four trials used a parallel‐group design (STEP 1, 2, 3, 5); STEP 4 used a withdrawal design with a 20‐week open‐label semaglutide run‐in before randomization. Because in the STEP 4 trial both arms had pre‐randomization drug exposure, STEP 1, 2, 3 and 5 were prespecified as the primary analysis. A sensitivity analysis incorporating all five trials was also performed.

The primary outcome was the incidence of gallbladder‐related events, encompassing cholelithiasis, cholecystitis and gallbladder disease not otherwise specified. Gallbladder‐related events were defined according to each trial's prespecified adverse event categories, which consistently encompassed these diagnoses without additional diagnostic harmonization across trials. Event data were independently extracted by two reviewers from published reports and Supporting Information. Risk ratios (RRs) with 95% confidence intervals (CIs) were calculated per trial. For rare events or arms with zero events (STEP 2 semaglutide arm), a continuity correction of 0.5 was applied. Pooled RRs were estimated using the DerSimonian–Laird random‐effects model; heterogeneity was quantified with the I ² statistic. Incidence rates were calculated as events per 100 patient‐years using reported follow‐up durations.

3. Results

Across the four parallel‐group trials (STEP 1, 2, 3 and 5; primary analysis), 3683 participants were included (semaglutide: 2269; placebo: 1414). STEP 2 enrolled individuals with type 2 diabetes; the remaining trials enrolled participants without diabetes. Duration was 68 weeks for STEP 1–3 and 104 weeks for STEP 5 (Table 1). Gallbladder‐related events were uncommon in all trials. In the primary analysis, semaglutide was associated with a statistically significant approximately two‐fold increased risk of gallbladder‐related events compared with placebo (pooled RR 2.08, 95% CI 1.16–3.72; I ² = 4.3%; Figure 1). Cholelithiasis accounted for the majority of events in trials where individual diagnoses were reported; cholecystitis and cholecystectomy were rare or unreported in most trials. Semaglutide‐treated participants experienced approximately 2.0 gallbladder events per 100 patient‐years, versus 0.8 per 100 patient‐years with placebo across these four trials (Table 1).

TABLE 1.

Gallbladder‐related events per 100 patient‐years in STEP trials.

Trial	Treatment	Events (n)	N	Follow‐up (years)	Patient‐years	Incidence/100 PY	RR (95% CI)
STEP 1	Semaglutide 2.4 mg	34	1306	1.31	1711	1.99	2.13 (0.99–4.58)
	Placebo	8	655	1.31	858	0.93
STEP 2 a	Semaglutide 2.4 mg	1	404	1.31	529	0.19	0.33 (0.03–3.18)
	Placebo	3	403	1.31	528	0.57
STEP 3	Semaglutide 2.4 mg	20	407	1.31	533	3.76	3.34 (1.00–11.11)
	Placebo	3	204	1.31	267	1.12
STEP 4 b	Semaglutide (cont.)	15	535	1.31	701	2.14	0.75 (0.34–1.65)
	Placebo (after run‐in)	10	268	1.31	351	2.85
STEP 5	Semaglutide 2.4 mg	4	152	2.00	304	1.32	2.00 (0.37–10.76)
	Placebo	2	152	2.00	304	0.66
Primary analysis (STEP 1, 2, 3, 5)	Semaglutide	59	2269	—	3077	1.92	2.08 (1.16–3.72)
	Placebo	16	1414	—	1957	0.82	I 2 = 4.3%
All 5 trials	Semaglutide	74	2804	—	3772	1.96	1.47 (0.74–2.95)
	Placebo	26	1682	—	2305	1.13	I 2 = 45%

Abbreviations: PY, patient‐years; RR, risk ratio.

^a STEP 2 enrolled participants with type 2 diabetes.

^b STEP 4 used a 20‐week open‐label semaglutide run‐in before randomization; excluded from primary analysis. A continuity correction of 0.5 was applied to STEP 2 semaglutide arm.

FIGURE 1.

Random‐effects pooled RR for gallbladder‐related events, STEP 1, 2, 3, 5; excludes STEP 4 due to pre‐randomization semaglutide exposure. Pooled RR 2.08, 95% CI 1.16–3.72, I ² = 4.3%.

In the sensitivity analysis including all five trials (STEP 1–5; n = 4486; semaglutide: 2804; placebo: 1682), the pooled estimate was attenuated and no longer statistically significant (RR 1.47, 95% CI 0.74–2.95; I ² = 45%), with the elevated heterogeneity driven primarily by the atypical design of STEP 4 (where event rates were slightly higher in the placebo arm: 3.73% vs. 2.80%), and by differences between STEP 2 and STEP 3 in the parallel‐group subset. This attenuation is consistent with the expected dilution of the semaglutide effect when STEP 4—where both arms received pre‐randomization drug—is included.

4. Conclusions

In this pooled analysis of four parallel‐group, randomized, placebo‐controlled trials, semaglutide 2.4 mg was associated with an approximately two‐fold increased risk of gallbladder‐related events compared with placebo. The absolute incidence was low (approximately 2 events per 100 patient‐years), substantially below rates reported after bariatric surgery [8]. Cholelithiasis accounted for most events; cholecystitis and cholecystectomy were rare.

GLP‐1 receptors are expressed on gallbladder smooth muscle and inhibit cholecystokinin‐stimulated gallbladder contraction, resulting in impaired fasting gallbladder emptying and bile stasis—effects demonstrated in human pharmacodynamic studies with GLP‐1 RA therapy [3]. Concurrently, the rapid and substantial weight loss achieved with semaglutide increases hepatic cholesterol secretion and reduces bile acid synthesis, shifting bile towards a more lithogenic composition—a mechanism well‐characterized in the bariatric surgery literature. The STEP trials likely capture both pathways acting in concert, which may explain the higher event rates observed in STEP 3, where a more intensive weight‐loss intervention was employed.

A critical limitation of all trial‐based estimates of GLP‐1 RA‐associated gallbladder disease is that adverse events were captured through spontaneous clinical reporting. Asymptomatic cholelithiasis—which accounts for the majority of gallstones in the general population and is detectable only by systematic ultrasound—is not reflected in these figures. Studies including protocol‐mandated ultrasound before and during treatment would likely reveal substantially higher rates of gallstone formation, as demonstrated in the bariatric surgery setting. Furthermore, individual gallbladder diagnoses (cholelithiasis, cholecystitis and cholecystectomy) were not consistently reported across trials in sufficient granularity to permit formal pooled analysis of each subtype, and the small number of trials precludes formal meta‐regression to quantify sources of heterogeneity. When STEP 4 is included in the sensitivity analysis, elevated heterogeneity (I ² = 45%) is attributable to several factors beyond the structural run‐in design: prevalent gallstones formed during the run‐in are distributed across both arms at randomization, attenuating post‐randomization differences; the placebo arm in STEP 4 experienced weight regain after semaglutide discontinuation, which may independently affect gallbladder dynamics; and the STEP 4 placebo event rate (3.73%) was higher than in any other placebo arm, consistent with carry‐forward of prevalent disease.

The clinical implications are modest given the low absolute risk. Our results correspond to approximately one additional gallbladder event per 90–100 treated patients over 1 year compared to placebo. Nevertheless, clinicians should counsel patients about biliary symptoms, particularly during the early phase of rapid weight loss. These findings are particularly relevant given the rapidly expanding use of semaglutide for obesity treatment in routine clinical practice. Although time‐to‐event data were not available, prior evidence suggests that gallbladder events cluster during periods of rapid weight reduction. If symptomatic cholelithiasis develops during treatment, standard biliary evaluation should be performed. Overall, the benefits of semaglutide for weight reduction are likely to outweigh this risk for most patients. Whether prophylactic ursodeoxycholic acid—shown to reduce symptomatic gallstone formation after bariatric surgery [8]—might provide similar benefit in patients initiating GLP‐1 RA therapy remains unknown, and direct evidence in this population is lacking. This question warrants prospective evaluation, particularly in high‐risk individuals such as those with pre‐existing biliary sludge, prior cholelithiasis, or undergoing rapid weight loss.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Baseline characteristics of included trials.

Ukkonen M., Kivivuori A., Starckjohann P., Lammi P., Junttila A., and Panula V., “Gallbladder‐Related Adverse Events With Semaglutide 2.4 mg: Pooled Analysis of Placebo‐Controlled STEP Trials,” Diabetes, Obesity and Metabolism 28, no. 10 (2026): 9701–9703, 10.1111/dom.71087.

Data Availability Statement

The data underlying this study were extracted from published studies, all of which are cited in the manuscript. No new datasets were generated or deposited in a public repository.