Validation of a simple classification system for endoscopic diagnosis of small colorectal polyps using narrow-band imaging

Abstract

Background & aims: Almost all colorectal polyps ≤ 5 mm are benign, yet current practice requires costly pathologic analysis. We aimed to develop and evaluate the validity of a simple narrow-band imaging (NBI)-based classification system for differentiating hyperplastic from adenomatous polyps.

Methods: The study was conducted in 4 phases: (1) evaluation of accuracy and reliability of histologic prediction by NBI-experienced colonoscopists; (2) development of a classification based on color, vessels, and surface pattern criteria, using a modified Delphi method; (3) validation of the component criteria by people not experienced in endoscopy or NBI analysis (25 medical students, 19 gastroenterology fellows) using 118 high-definition colorectal polyp images of known histology; and (4) validation of the classification system by NBI-trained gastroenterology fellows, using still images. We performed a pilot evaluation during real-time colonoscopy.

Results: We developed a classification system for the endoscopic diagnosis of colorectal polyp histology and established its predictive validity. When all 3 criteria were used, the specificity ranged from 94.9% to 100% and the combined sensitivity ranged from 8.5% to 61.0%. The specificities of the individual criteria were lower although the sensitivities were higher. During real-time colonoscopy, endoscopists made diagnoses with high confidence for 75% of consecutive small colorectal polyps, with 89% accuracy, 98% sensitivity, and 95% negative predictive values.

Conclusions: We developed and established the validity of an NBI classification system that can be used to diagnose colorectal polyps. In preliminary real-time evaluation, the system allowed endoscopic diagnoses of colorectal polyp histology.