Defined morphological criteria allow reliable diagnosis of colorectal serrated polyps and predict polyp genetics

Contains fulltext : 137748pub.pdf (publisher's version ) (Closed access)
Criteria for the diagnosis of serrated colorectal lesions (hyperplastic polyp, sessile serrated adenoma without or with dysplasia-which we called mixed polyp-and traditional serrated adenoma) for which consensus has been reached should be validated for applicability in daily practice in terms of inter-observer reproducibility and their association with clinical features and (epi)genetic events. A study set was created from a consecutive series of colorectal polyps (n = 1,926) by selecting all sessile serrated adenomas, traditional serrated adenomas and mixed polyps. We added consecutive series of hyperplastic polyps, classical adenomas and normal mucosa samples for a total of 200 specimens. With this series, we conducted an inter-observer study, encompassing ten pathologists with gastrointestinal pathology experience from five European countries, in three rounds in which all cases were microscopically evaluated. An assessment of single morphological criteria was included, and these were correlated with clinical parameters and the mutation status of KRAS, BRAF and PIK3CA and the methylation status of MLH1. Gender, age and localisation were significantly associated with certain types of lesions. Kappa statistics revealed moderate to good inter-observer agreement for polyp classification (kappa = 0.56 to 0.63), but for single criteria, this varied considerably (kappa = 0.06 to 0.82). BRAF mutations were frequently found in hyperplastic polyps (86 %, 62/72) and sessile serrated adenomas (80 %, 41/51). KRAS mutations occurred more frequently in traditional serrated adenomas (78 %, 7/9) and less so in classical adenomas (20 %, 10/51). Single morphological criteria for sessile serrated adenomas showed significant correlation with BRAF mutation (all p </= 0.001), and those for classical adenomas or traditional serrated adenoma correlated significantly with KRAS mutation (all p < 0.001). Therefore, single well-defined morphological criteria are predictive for genetic alterat