[Subjective grading of Barrett's neoplasia by pathologists: correlation with objective histomorphometric variables]

Even though pathologists are trained to recognize the same histological features for the diagnosis and grading of different histological images, not all pathologists are influenced to a similar level of intensity by the same morphological characteristics of the tissue when scoring Barrett's dysplasia/neoplasia. The variables which most pathologists have intuitively chosen to use for scoring of the severity of Barrett's changes are mainly those related to the general tissue architecture, such as nuclear crowding, orientation and stratification. Interestingly, nuclear size is not used by most pathologists but nuclear pleomorphism and symmetry does influence a significant number of pathologists. Maybe the most difficult variables for the human eye to recognize are variables of chromatin texture (such as margination or heterogeneity), the predictive importance of which has been demonstrated in a previously published work. Textural variables may therefore remain the subject of a computerized analysis. Nevertheless, the fact that a few pathologists do actually correlate with nuclear texture in scoring, argues in favor of making further attempts to train pathologists to also rely on texture, similar to cytologists, when scoring Barrett's dysplasia.