Adults treated with liraglutide developed Barrett's esophagus at about one third the rate of adults who underwent sleeve gastrectomy over a follow-up of at least three years. Esophageal cancer was rare in both groups and did not separate.
The retrospective comparison, published online April 23 in the Annals of Gastroenterology ↗, used a large electronic health-record database to identify adults with obesity who either had a sleeve gastrectomy or took daily liraglutide. Propensity-score matching across demographics, comorbidities, medication use, and body mass index produced 10,048 patients per arm. The primary outcome was Barrett's esophagus without dysplasia diagnosed at three years or later after treatment. Secondary outcomes were Barrett's with dysplasia and esophageal cancer.
Barrett's esophagus is a precancerous change in the lining of the lower esophagus, where chronic acid exposure remodels the epithelium toward an intestinal type. Most patients with Barrett's never progress to cancer, but the lesion is the recognized stepping stone to esophageal adenocarcinoma and is the reason gastroenterologists run surveillance endoscopies on patients with long-standing reflux.
In the matched cohort, Barrett's without dysplasia was found in 0.3 percent of sleeve patients and 0.1 percent of liraglutide patients, a risk ratio of 2.70 (95% CI 1.31-5.56) and an absolute risk difference of 0.2 percentage points (95% CI 0.1-0.3). Barrett's with dysplasia, the higher-grade form, was 0.1 percent versus zero. Esophageal cancer rates did not differ significantly between arms. The signal sits entirely in the precancerous lesion, not the malignancy, and at three years that is the readout most likely to move first if the comparison ever produces a cancer gap.
The biological reason this comparison was worth running is well established. Sleeve gastrectomy removes most of the greater curvature of the stomach and creates a high-pressure tube, which produces or worsens gastroesophageal reflux disease in a substantial fraction of patients. Reflux is the upstream driver of Barrett's. Liraglutide ↗ is a daily GLP-1 receptor agonist that produces weight loss without changing gastric anatomy. The expectation was that the two interventions would diverge on reflux-mediated outcomes. They did.
Two caveats. The first is the rarity. Barrett's at 0.3 percent versus 0.1 percent means about 30 sleeve patients and 10 liraglutide patients across cohorts of more than ten thousand each. Small absolute event counts produce wide confidence intervals on the ratio, which is why the upper bound here is 5.56. The signal is real but the precision is modest. The second is the follow-up window. Three years is the right cutoff for Barrett's, but esophageal adenocarcinoma develops over a much longer arc, and any cancer-divergence signal from this comparison would need closer to a decade of follow-up.
The result fits into a broader pattern of bariatric-vs-GLP-1 head-to-head readouts that the literature is starting to fill in. The Neurosurgery cohort on idiopathic intracranial hypertension covered earlier in this section showed bariatric surgery beating GLP-1 receptor agonists early on vision-protection endpoints, with the gap shrinking past eighteen months. The Barrett's comparison runs the other direction. A non-glycemic complication where the anatomical intervention carries the higher risk and the drug carries none of it. Both are now in the ledger that any future obesity-care guideline will have to weigh against weight-loss magnitude and cost.
Semaglutide ↗ and tirzepatide ↗, the two weekly injectables that have largely displaced liraglutide in current obesity practice, were not in this analysis. Whether the reflux protection generalizes to weekly agonists or to the dual GLP-1 / GIP class is an open question. The Annals of Gastroenterology authors note that liraglutide is short-acting and treat that as part of the explanation. A parallel comparison with semaglutide is the obvious next paper.