Accuracy of in vivo optical diagnosis of colon polyp histology by narrow-band imaging in predicting colonoscopy surveillance intervals

Background: The American Society for Gastrointestinal Endoscopy (ASGE) recently developed thresholds for the performance characteristics of technologies for real-time assessment of histology of diminutive (≤5 mm) colon polyps. Narrow-band imaging (NBI) has been shown to predict polyp histology with moderate to high accuracy in several studies. Objective: To determine whether in vivo optical diagnosis of polyp histology by using NBI can reach the 2 benchmarks set forth by the ASGE. Design: Retrospective analysis of data from 3 prospective clinical trials. Setting: Two tertiary referral centers. Patients: Subjects undergoing screening or surveillance colonoscopy. Interventions: In vivo optical diagnosis of polyp histology by using NBI. Main Outcome Measurement: Accuracy in predicting colonoscopy surveillance intervals, negative predictive value (NPV) for diagnosing adenomatous histology in the rectosigmoid. Results: A total of 410 patients met the inclusion/exclusion criteria and had at least 1 polyp seen and resected during colonoscopy. Using in vivo optical diagnosis instead of histopathology for all diminutive polyps predicted the correct colonoscopy surveillance interval in 86% to 94% patients. When optical diagnosis was limited to diminutive polyps in the rectosigmoid only, the NPV for diagnosing adenomatous histology with NBI was 95%. Limitations: Retrospective analysis from tertiary referral centers. Conclusions: The threshold NPV for diagnosing adenomatous histology in diminutive rectosigmoid polyps recently set forth by the ASGE can be achieved by using NBI. The threshold accuracy rate for predicting surveillance interval recommendations can be reached by using NBI, but only if patients with 1 to 2 small adenomas without advanced features have a repeat colonoscopy in 10 years.