Clinical and computed tomography characteristics of high-risk periampullary diverticulum.

Abstract

Background: Periampullary diverticulum (PAD) is a common acquired extraluminal outpouching of duodenal mucosa and is often clinically overlooked. When complicated by biliary-pancreatic stones or inflammation, PAD may lead to severe biliary-pancreatic complications. Although endoscopic retrograde cholangiopancreatography (ERCP) is currently regarded as the gold standard for PAD diagnosis, its invasive nature and other limitations have led to computed tomography (CT) being increasingly preferred as the first-line imaging modality. However, the correlation between imaging features and clinical manifestations remains unclear for high-risk PAD (HRPAD), defined as PAD associated with cholelithiasis and biliary-pancreatic inflammation. Therefore, this study aimed to examine the clinical and CT characteristics of HRPAD.

Methods: The data of 260 patients with HRPAD and 288 patients with non-high-risk PAD (N-HRPAD) who underwent contrast-enhanced abdominal CT scans between January 2018 and May 2024 were retrospectively analyzed. The clinical and CT features were compared between the groups. Independent factors for diagnosing HRPAD were identified through binary logistic regression analysis. Additionally, an external validation cohort comprising 150 patients from another center was used to test the predictive efficiency of the model.

Results: Patients with HRPAD were significantly older than those with N-HRPAD, with an average age of 72±10 and 68±11 years, respectively (P<0.001). Additionally, they exhibited a higher proportion of larger diverticula (>1.95 cm: 65.38% vs. 22.57%), larger common bile duct (CBD) size (>0.69 cm: 72.31% vs. 20.83%), and greater pancreatic duct (PD) dilation (>0.20 cm: 86.92% vs. 60.42%) (all P values <0.001). Furthermore, there was a greater prevalence of diverticular neck swelling in the HRPAD group than in the control group (23.08% vs. 4.86%; P<0.001). Binary logistic regression analysis indicated that a diverticulum size >1.95 cm, CBD size >0.69 cm, PD size >0.20 cm, and the presence of diverticular neck swelling were predictors significantly associated with HRPAD, yielding an area under the curve (AUC) of 0.848; meanwhile, the AUC of the external validation cohort was 0.829.

Conclusions: Patients with HRPAD and those with N-HRPAD exhibited different clinical and imaging characteristics. A thorough understanding of these differences may facilitate early identification of HRPAD and timely treatment interventions, which would reduce complications related to biliary-pancreatic diseases.