Histological assessment of intra- and inter-institutional reliabilities in detection of desmoplastic reaction in biopsy specimens of early colorectal carcinomas

Pathol Int. 2013 Nov;63(11):539-45. doi: 10.1111/pin.12110.

Abstract

We previously reported a relationship between depth of submucosal invasion of early colorectal carcinomas and desmoplastic reaction (DR). However, poor inter-observer agreement on the histopathological diagnosis of DR in biopsy specimens with hematoxylin and eosin (H&E) staining has been the major critique of this tool. In this study, reproducibility of the histopathological diagnosis of DR was evaluated. Furthermore, we investigated the possible improvement of the reproducibility after education about histological characteristics and tried to identify histological characteristics that are most important in the recognition of DR. A total of 34 H&E stained slides were included in this study and analyzed by three pathologists. Slides were reviewed before and after education about histological characteristics of DR. Kappa statistics were used to compare the inter-observer variability. We investigated the relationship between DR and histopathological factor. The inter-observer agreement during the first session varied between 0.30 and 0.63, which improved during the second session toward an agreement between 0.58 and 0.71. Myofibroblast proliferation associated with cancer invasion was found to be the most useful in the diagnosis of DR. In conclusion, the correct detection of myofibroblasts may facilitate the standardization of diagnosis of DR.

Keywords: Myofibroblast; biopsy; desmoplastic reaction; early colorectal carcinoma; inter-observer agreement; prospective study; retrospective study; submucosal invasion.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adenocarcinoma / pathology*
  • Adult
  • Aged
  • Biopsy
  • Cell Proliferation
  • Colorectal Neoplasms / pathology*
  • Female
  • Humans
  • Male
  • Middle Aged
  • Myofibroblasts / pathology*
  • Observer Variation
  • Reproducibility of Results