Use of statistical evaluation of antigen profiles in differential diagnosis between colonic and ovarian adenocarcinomas.

A study was carried out to determine whether it was possible to classify colonic and ovarian adenocarcinomas by their antigen profile. Colonic and ovarian adenocarcinomas were immunostained with a panel of antibodies which have a limited specificity for colon (parlam-4, 19.9, anti-secretory component) and ovary (OV-TL3 and OC125) and the most discriminatory antibodies were selected by stepwise linear discriminant analysis. For frozen material OV-TL3 and OC125 were the best classifying antibodies. Although OC125 had better discriminative power, for paraffin wax embedded material parlam-4 was selected as the best classifying antibody. OC125 had no additional effect on the classification of a tumour. These antibodies were subsequently tested on an independent test set of primary and metastatic adenocarcinomas of colonic and ovarian origin. When ovarian posterior probabilities of less than 0.1 and greater than 0.9 were selected as cut off points for a positive identification of colonic or ovarian origin (jackknifed classification method), no adenocarcinoma was incorrectly identified as ovarian carcinoma in frozen material. The same trend was noticed for paraffin wax embedded material. Statistical analysis of antigen profiles can be helpful in defining the colonic or ovarian origin of an adenocarcinoma when routine microscopy does not yield a definitive result.