Prevalence of duodenal ulcer and coexisting endoscopic findings in cirrhotic patients with upper gastrointestinal bleeding: a prospective endoscopic study

Abstract

Objective. To assess the prevalence of duodenal ulcers and coexisting endoscopic findings in cirrhotic patients presenting with upper gastrointestinal bleeding. Materials and methods. This prospective study included 132 patients with liver cirrhosis and upper gastrointestinal bleeding who underwent endoscopy after hemodynamic stabilization at the Institute of Medical Gastroenterology, Madras Medical College, Chennai, India, between January 2024 and December 2025. Duodenal ulcers, Forrest classification, stigmata of recent bleeding, esophageal and fundic varices, and portal hypertensive gastropathy (PHG) were evaluated. Ulcer prevalence was calculated with a 95% confidence interval (CI). Group comparisons and multivariable logistic regression were performed. Results. Duodenal ulcers were identified in 24 of 132 patients (18.18%; 95% CI: 12.01-25.83%). Mean age was 43.13 ± 5.91 years in patients with ulcers and 41.44 ± 7.26 years in those without ulcers. Ulcers occurred in 22 of 78 men and 2 of 54 women (unadjusted odds ratio [OR]: 10.21; Fisher's exact p < 0.001). Male sex remained associated with duodenal ulcers in multivariable analysis (adjusted OR: 10.07; 95% CI: 2.25-45.04; p = 0.00251). Most ulcers were classified as Forrest III; only three patients (12.5%) had stigmata of recent bleeding. Among patients with duodenal ulcers, 21 (87.5%) had esophageal varices, 11 (45.83%) had mild PHG, and four (16.67%) had severe PHG. Three patients (12.5%) had isolated fundic varices. Conclusion. Duodenal ulcers were found in approximately one in five cirrhotic patients presenting with upper gastrointestinal bleeding, but most lacked stigmata of recent hemorrhage. Male sex was associated with ulcer presence. Careful endoscopic assessment is necessary to identify coexisting variceal and nonvariceal lesions without assuming that every detected ulcer is the bleeding source.