CLASSIFICATION OF COLORECTAL ADENOMAS WITH QUANTITATIVE PATHOLOGY - EVALUATION OF MORPHOMETRY, STEREOLOGY, MITOTIC COUNTS AND SYNTACTIC STRUCTURE-ANALYSIS

In search for an objective classification rule for the morphologic cha nges in colorectal adenomas, the results of nuclear morphometry for as sessing geometric characteristics of tumour nuclei, syntactic structur e analysis for assessing the arrangement of nuclei in the epithelium, stereological assessment of glandular changes, and area weighted mitot ic counts were evaluated in a multivariate analysis. The H&E stained t issue sections of 59 colorectal adenomas were studied, of which 20 sho wed mild, 20 moderate and 19 severe dysplasia, according to blind dupl icate assessments by two pathologists. In a stepwise jackknifed discri minant analysis, the combination of variables was selected that could best discriminate the cases into the previously assessed grade of dysp lasia. With the combination of two variables (minimum line length and inner gland surface density), 71% of the cases could be correctly clas sified according to the previously assessed grade. No cases with mild dysplasia were classified as severe dysplasia and vice versa. However, since the reproduction of subjectively assessed grade was not the ult imate goal, we aimed to discriminate the cases into two groups on the basis of measurements results alone, by means of a cluster analysis. T hese two groups would merely reflect the cases with major versus minor morphologic changes. The results of the cluster analysis showed that all mild dysplasia cases were allocated into the low grade cluster, an d that the moderate dysplasia cases were divided over both the low gra de and high grade cluster. Two severe dysplasia cases with borderline morphometric results were allocated into the low grade cluster, and th e remaining 16 into the high grade cluster. It is concluded that objec tive evaluation of dysplasia in colorectal adenomas is possible by a c ombination of morphometric techniques. The results of the morphometric analyses seem to favor a two-grade classification system rather than a three-grade system.