Development and validation of the linked color imaging classification for endoscopic prediction of colorectal polyp histology.

Abstract

Linked color imaging (LCI) is a recently developed technique that emphasizes differences in mucosal color. In this study we aimed to develop a LCI classification based on the Narrow-band Imaging International Colorectal Endoscopic Classification for predicting colorectal polyp histology and evaluate the validity and performance of the endoscopists in differentiating hyperplastic polyps from adenomas using the LCI classification. A workshop involving six international experts from China and Japan with substantial experience with LCI developed the classification. Three experienced and seven less-experienced endoscopists used the LCI images to predict the histology of polyps independently, recording their degrees of confidence in these predictions before and after completing the training test for the LCI classification. Of the 50 polyps included, 30 (60.0%) were adenomas. Overall diagnostic accuracy before training was 75.4% (95% confidence interval [CI] 71.4%-79.1%), which increased to 85.2% (95% CI 81.8%-88.2%) after training. After training, the experienced and less-experienced endoscopists achievedanoverallaccuracyof 87.3% and 84.3%for the prediction of polyp histology.Polyp prediction using the color criterion alone had the highest specificity and positive predictive value, whereas the vessel criterion achieved the highest accuracy and negative predictive value among all three individual LCI criteria. After training, both the experienced and less-experienced endoscopists had high degrees of interobserver agreement. We developed and validated the first LCI classification for endoscopic differentiation of adenomas and hyperplastic polyps. The LCI classification significantly improved the diagnostic accuracy of colorectal polyps.