Interobserver variation in the reporting of local peritoneal involvement and extramural venous invasion in colonic cancer

Aims: Local peritoneal involvement (LPI) and extramural venous invasion (EMVI) are of prognostic value in Dukes' B colonic cancers and may be used to select patients for adjuvant chemotherapy. There is marked variation in the frequency with which they are reported however, ranging from 7% to 39% and 10% to 90%, respectively. A grading system for diagnosing LPI has been proposed by Shepherd et al. and partially incorporated into the Royal College of Pathologists guidelines for reporting colorectal cancer. This study aimed to determine the degree of interobserver variation in the reporting of LPI and EMVI amongst a group of experienced pathologists with a special interest in gastrointestinal pathology. Methods and results: Four pathologists specialising in gastrointestinal pathology independently assessed LPI according to the grading system described by Shepherd et al. and the presence or absence of EMVI on 138 and 131 slides of pT3 and pT4 colonic cancers, respectively. Kappa statistics were performed to assess interobserver agreement. Kappa values for LPI ranged from kappa = 0.74 (substantial agreement) to kappa = 0.89 (almost perfect agreement). Kappa values for EMVI ranged from kappa = 0.29 (poor agreement) to kappa = 0.59 (moderate agreement). Conclusions: Using Shepherd's grading system there was good agreement between pathologists in reporting LPI in colonic carcinomas. The reporting of EMVI in colonic carcinomas on haematoxylin and eosin-stained slides had only poor to moderate agreement however, even amongst gastrointestinal pathologists working together in a single unit. Introduction of standardized criteria and/or the use of an elastin stain in the diagnosis of EMVI may assist in improving interobserver agreement.