Background and aim: Colorectal laterally spreading tumor granular type (LST-G) is generally divided into two subtypes based on morphology. Here, we retrospectively investigated the clinical significance of a concrete, objective LST-G subclassification.
Methods: This study examined 636 consecutive cases that were resected endoscopically or surgically. LST-G was subclassified as follows: Type 1, a lesion with homogenous uniform granules with uniform (<5 mm) nodules; Type 2, a lesion with granules and small nodules (≥5 mm, <10 mm); or Type 3, a lesion accompanied by large nodules (≥10 mm). For the validation study, 194 images were compiled from 97 cases investigated using conventional colonoscopy and chromoendoscopy with indigo carmine dye spraying. Images were distributed in a randomized order to students without prior endoscopy experience, less-experienced endoscopists (LEE group), and highly experienced endoscopists (HEE group). Diagnostic accuracy and interobserver agreement were then evaluated.
Results: There was no submucosal invasion in Type 1 lesions. The incidence of deep submucosal invasive carcinoma was higher for Type 3 lesions than for Type 2 lesions. Interobserver agreement was good in each group. Diagnostic accuracy was higher in the HEE group than in the student and LEE groups. Chromoendoscopy had a higher accuracy rate than conventional colonoscopy in the LEE and HEE groups (LEE, 0.74 vs 0.69, P < 0.05; HEE, 0.84 vs 0.78, P < 0.05).
Conclusions: This subclassification of LST-G according to the diameters of granules and nodules was both useful for choosing therapeutic strategies in the clinical setting and universally applicable.
Keywords: LST granular type; diagnostic accuracy; kappa value; size of nodule; subclassification.
© 2015 Journal of Gastroenterology and Hepatology Foundation and John Wiley & Sons Australia, Ltd.