Back

Non-Endoscopic Screening for Barrett's Esophagus Using a DNA Methylation-Based Assay: 18-Month Real-World Experience in 11,991 Patients

Smith, M. S.; Chen, W. T.; Kotfila, R. P.; Panzarella, P. S.; Le, I. T.; Verma, S.; Lee, V. T.; Aklog, L.

2025-11-21 gastroenterology
10.1101/2025.11.20.25340693 medRxiv
Show abstract

BackgroundBarretts esophagus (BE), characterized by specialized intestinal metaplasia (SIM), is the precursor to esophageal adenocarcinoma (EAC). Despite published BE screening guidelines for at-risk individuals, uptake of endoscopic screening remains low. We present 18 months of real-world data on non-endoscopic BE screening using EsoGuard(R) (EG), the first commercially available U.S. molecular biomarker test for this purpose, performed on esophageal cell samples collected with the swallowable EsoCheck(R) (EC) balloon-capsule device. MethodsWe retrospectively analyzed EC performance and EG results in patients tested commercially from January 2023 to June 2024. A subset enrolled in a registry underwent follow-up endoscopy. Multivariable logistic regression was used to evaluate risk factors associated with (1) positive EG results, and (2) confirmed BE (SIM [&ge;]1 cm). ResultsAmong 11,991 tested patients, 11,355 (94.7%) had successful EC cell collection, averaging under 2 minutes with no serious adverse events. EG was positive in 16.6% of patients, with positivity increasing by age; age > 50 years was the strongest individual risk factor for predicting a positive EG result. Among 177 EG-positive registry patients who underwent endoscopy, 59 (33.3%) had SIM, of which 33 met American College of Gastroenterology criteria for BE and 26 had ultra-short SIM (<1 cm). Dysplasia was found in 3 patients: 1 HGD, 1 LGD, and 1 indefinite for dysplasia (IND). ConclusionsWe report here the largest real-world experience of EG and EC to date, demonstrating excellent safety, tolerability, and scalability. EG detects both guideline-recognized and ultra-short SIM, supporting its utility as a non-invasive BE screening tool.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.