558 Accuracy of Endoscopists in Predicting Colorectal Polyps

Abstract

INTRODUCTION: Colonoscopy is the gold standard for the identification of colorectal cancer and polyps. The majority of colorectal polyps found at screening colonoscopy are diminutive (≤5 mm) and have a low prevalence for advanced histological features and rarely harbor cancer. Precise endoscopic determination of the histology of polyps at the time of colonoscopy could prevent unnecessary polypectomies. The ASGE PIVI statement on this topic proposes to leave a suspected recto-sigmoid hyperplastic polyp ≤5 mm, there should be ≤90% negative predictive value (when used with high confidence) for adenomatous histology. The study purpose is to assess the accuracy of experienced endoscopists in predicting histological types of polyps resected at standard screening colonoscopy using white light. METHODS: Patients undergoing screening colonoscopy were enrolled into this prospective study. The study was conducted at two sites – a tertiary community teaching hospital and an outpatient endoscopy unit. Screening colonoscopy was defined as the first colonoscopy performed in an adequately prepped asymptomatic patient between the age of 50-80 with no risk factors. Six endoscopists were asked to complete a form at the end of the procedure, which included their pathology prediction and level of confidence of each resected polyp. Classifications for prediction pathology were adenoma (tubular, villous, tubulovillous), serrated (hyperplastic, sessile serrated adenoma, traditional serrated adenoma), cancer and others. Analysis was conducted to determine the accuracy of predictions. RESULTS: Among the 158 patients eligible for the study, endoscopist predictions were made on 336 resected polyps. The majority of polyps (82%) were <5 mm. Predictive accuracy of all polyp types with high confidence was 84%. The accuracy in predicting adenomatous histology with high confidence was 89%. Negative predictive value of polyps <5 mm located in the recto-sigmoid predicted with high confidence was 90%. When comparing endoscopists, there was not significant variability for predicting polyp type or adenoma classification. Out of the 42 polyps predicted to be hyperplastic with high confidence, 30 (71%) were correctly predicted, 7 (17%), 3 (7%) and 2 (5%) were diagnosed as tubular adenoma, sessile serrated adenoma, and other respectively. CONCLUSION: Our study have met the negative predicative value threshold for adenomatous poly histology, supporting a “diagnose- and- leave” strategy for diminutive predicted polyps in the rectosigmoid.