Value of prediction model in distinguishing gallbladder adenoma from cholesterol polyp

Abstract Background and Aim Gallbladder adenomatous polyp is a pre‐cancerous neoplasm, and it is difficult to classify from cholesterol polyps before cholecystectomy. The study aimed to clarify the risk characteristics of gallbladder adenomas and establish a prediction model to differentiate gallbladder adenomas from cholesterol polyp lesions. Methods From May 2019 to December 2021, the patients underwent cholecystectomy in the Shanghai Eastern Hepatobiliary Surgery Hospital were retrospectively reviewed. According to the permanent pathology test, the patients were divided into adenomas and cholesterol polyps groups. All the included cases received ultrasound equipment examinations before cholecystectomy and their clinical information were completely recorded. Then the patients' baseline characteristics and ultrasound imaging variables were analyzed by logistic regression. Finally, a predictive model for gallbladder adenomas will be established and assessed based on the independent risk factors. Results A total of 423 cases including 296 cholesterol polyps and 127 gallbladder adenomas were analyzed in detail. Multivariate logistic regression analysis revealed that solitary polyp lesion (OR = 2.954, 95% CI 1.759–4.960, P < 0.001), the maximal diameter of lesions (OR = 1.244, 95% CI 1.169–1.324, P < 0.001), and irregular shape of polyp lesions (OR = 5.549, 95% CI 1.979–15.560, P = 0.001) were the independent predictive factors of gallbladder adenomas. According to the results, regression equation of logit( P ) = −3.828 + 1.083*number of gallbladder polyps lesions (GPLs) + 0.218*diameter of GPLs + 1.714*shape of GPLs was established. Area under the curve (AUC) was 0.828 (95% CI 0.782–0.874, P < 0.001). When logit P > 0.204, the sensitivity of estimating adenoma was 79.5%, the specificity of recognizing adenoma was 70.6%, and the whole correct ratio was 73.3%. While the AUC of diameter (10 mm) being a predictive factor in this study was only 0.790 (95% CI 0.741–0.839, P < 0.001). And the sensitivity and specificity of 10 mm as the optimal diagnostic cutoff value to diagnose adenomas were 74.8% and 65.9%, respectively. Conclusions The risk factors of solitary polyp lesion, larger diameter, and irregular morphology feature of polyp lesions were significantly related to gallbladder adenomas. And the predictive model established in the study can effectively identify adenomas from cholesterol polyps and help patients to select the optimal treatment protocol.