A Weight-Loss Drug, a Calmer Colon? What a New UC Study Found

Ulcerative colitis is a miserable disease to live with and a frustrating one to treat. The colon's lining inflames, ulcerates and bleeds; people run to the bathroom ten times a day and pass blood doing it. We have good drugs — mesalamine, biologics, JAK inhibitors — and still a large share of patients never reach lasting remission.

So a paper in Inflammatory Bowel Diseases asking an unexpected question is worth a look. Doctors at West Virginia University noticed something in their own records: their colitis patients who happened to start semaglutide or liraglutide — the GLP-1 drugs prescribed for diabetes and obesity — seemed to be doing better in the bowel, too. So they went back through the charts and counted.

At 12 weeks, 66.7% of GLP-1 users met the study's definition of symptomatic remission versus 25.3% of matched controls. This was an observational association, not proof of treatment efficacy.

What they found

1. Two thirds met the study's definition of symptomatic remission. They took 150 colitis patients who started a GLP-1 drug and matched each one — by age, sex, BMI, how active their colitis was, and what colitis medication they were on — to a patient who didn't. At twelve weeks, 66.7% of the GLP-1 group had symptomatic remission versus 25.3% of the matched controls. After statistical adjustment, the odds of remission were nearly six times higher. Everyone in the study was symptomatic at the start, and nobody had their colitis medication changed in the eight weeks beforehand.

2. It didn't wait for the weight to come off. This is the finding that gives the paper its interest. By week four, when the average patient had lost a mere 1.5% of their body weight, remission rates had already split — 34.7% versus 15.3%. And when the researchers sorted patients by how much weight they eventually lost, it made no difference: those who lost 5% or more were in remission 68% of the time, those who lost less, 64%. Statistically indistinguishable. In this dataset, the association between GLP-1 treatment and remission was not statistically explained by the amount of weight lost. That is compatible with a weight-independent effect, but it does not prove one.

Symptomatic improvement appeared before substantial weight loss, and the amount of weight lost was not independently associated with remission. That supports a possible weight-independent effect, but does not prove one.

3. There's a believable reason why. GLP-1 receptors aren't only in the pancreas and the brain. They sit on immune cells in the gut wall, including the lymphocytes tucked between the cells lining the intestine. Switch them on in animals and inflammatory signals — IL-1β, IL-6, TNF-α — go down, mucus production goes up, and the tight junctions that hold the gut lining sealed get stronger. In mouse colitis models, GLP-1 drugs calm the inflammation, and they stop working if you remove the receptor from immune cells. That gives the hypothesis biological plausibility — but these mechanistic findings come largely from preclinical studies, not from this group of patients.

4. A stool test pointed the same way. In the 22 patients who happened to have it measured, fecal calprotectin — a protein that rises when the gut lining is inflamed — fell from 480 to 145 µg/g. In the 20 controls measured, it barely moved (465 to 410). Unlike a symptom score, fecal calprotectin is an objective stool biomarker of intestinal inflammation. Still, this analysis involved only a small subset of patients.

5. The endoscopy subset pointed in the same direction. Follow-up endoscopy was available in 48 GLP-1 users and 45 controls. Endoscopic remission, defined in the paper as a Mayo Endoscopic Subscore of 0 or 1, occurred in 58% versus 38%, and at least a one-point endoscopic improvement occurred in 69% versus 47%. Because endoscopy was not scheduled by protocol and was performed at the treating physician's discretion, these results are exploratory and vulnerable to selection bias.

6. Among patients who remained on treatment for at least 12 weeks, no new IBD-specific safety signal emerged. Nausea and vomiting were common, as expected with GLP-1 drugs. The analyzed cohort did not show more UC-related adverse outcomes than controls, and there were no colectomies or reported pancreatitis or biliary complications during follow-up. However, 87 of the original 312 GLP-1 starters were excluded because they did not complete 12 weeks or were nonadherent, and 48 of those early discontinuations were attributed to gastrointestinal side effects. The study therefore describes tolerability among selected patients who stayed on treatment, not among everyone who started it. Semaglutide was associated with a higher symptomatic-remission rate than liraglutide (72.5% vs 60%), but that comparison was exploratory and nonrandomized.

Now the part the headline can't fit

The score wasn't measured. It was reconstructed. Start here, because this is the paper's soft centre. Nobody filled in a colitis score during these visits. Months or years later, reviewers read through the doctors' notes and worked out what the score would have been — and the paper says outright that when a component wasn't written down, it was "inferred from detailed descriptions of symptoms and physician impressions." The reviewers could see who was on the drug. A doctor writing a cheerful note about a patient doing well on a new weight-loss medication is not a neutral measuring instrument.

Symptoms can improve without proving that mucosal inflammation has healed. The partial Mayo score combines stool frequency, rectal bleeding and the physician's global assessment. GLP-1 drugs can themselves change gastrointestinal symptoms — including nausea, constipation or diarrhea — so symptom-based scores may be difficult to interpret in a retrospective chart review. Requiring a rectal-bleeding score of zero makes the remission definition more meaningful, and the fecal-calprotectin and endoscopy subsets support a possible anti-inflammatory effect. But those objective assessments were available in only a minority of patients and were not protocolized.

Only the patients who stayed on the drug were counted. Of 312 colitis patients who started a GLP-1 drug, 87 were dropped for not lasting twelve weeks — 48 of them because of gastrointestinal side effects. So the analysis is of people who tolerated the drug well. That's a systematically rosier group than "everyone who tries it," which is the group you actually belong to when your doctor writes the prescription.

The size of the association demands caution. The adjusted odds ratio for symptomatic remission was 5.90, which is unusually large for an observational treatment association. But an odds ratio from a retrospective matched cohort should not be compared directly with remission percentages or treatment effects from randomized trials of biologics or small molecules. The important point is simpler: an effect this large makes residual confounding, selection bias and measurement bias especially important possibilities to exclude in a randomized trial.

The two groups may not have been observed in exactly the same way. The authors note that patients starting a GLP-1 drug may have had more structured follow-up for metabolic care, whereas control index dates were based on qualifying outpatient visits. Outcomes came from routinely documented clinical data, and control dates were aligned within ±30 days when feasible. Even with matching, differences in visit frequency, documentation and health-seeking behavior could create residual bias.

And this was a particular kind of patient. Roughly 80% had a BMI over 30 and a third had type 2 diabetes — they were prescribed these drugs for metabolic reasons in the first place. Whether a lean person with colitis would see anything at all is untested. This was also a single hospital system, retrospective, with no correction for multiple comparisons; the semaglutide-versus-liraglutide edge only barely cleared statistical significance and shouldn't be read as a verdict on either drug. To the authors' credit, they declare no funding and no conflicts of interest, and they list most of these problems themselves.

This post is a plain-language summary of a published research paper and is not medical advice. Nothing here is a reason to start, stop or change a medication — talk to your own doctor about that.

Source
Alqinai B, Gayam S, Hadam-Veverka J. GLP-1 receptor agonist therapy is associated with increased symptomatic remission in ulcerative colitis: a matched cohort study. Inflammatory Bowel Diseases. 2026. doi.org/10.1093/ibd/izag167
For the record — retrospective 1:1 matched cohort, single US academic health system, 2022–2024; 150 GLP-1 RA users (80 semaglutide, 70 liraglutide) vs 150 matched controls; symptomatic remission (partial Mayo ≤ 2 with rectal bleeding subscore 0) 34.7% vs 15.3% at 4 weeks (P = .001), 54.7% vs 18.0% at 8 weeks, 66.7% vs 25.3% at 12 weeks (P < .001); adjusted OR 5.90 (95% CI 3.52–9.86); mean partial Mayo 5.8 → 2.1 vs 5.7 → 4.6; weight loss not associated with remission (adjusted OR 1.03 per 1%, 95% CI 0.95–1.12, P = .41); semaglutide 72.5% vs liraglutide 60% (OR 1.77, 95% CI 1.02–3.25, P = .04); fecal calprotectin 480 → 145 µg/g (n = 22) vs 465 → 410 µg/g (n = 20); endoscopic remission 58% vs 38% in the scoped subset (n = 48 vs 45). No funding and no conflicts of interest declared.