Morphometric and multivariate statistical detection of cancer cells in endometrial cytology

Endometrial carcinoma was discriminated in cytologic studies using morphometric and multivariate statistical methods. On aspirated samples from 70 cases (10 well-differentiated adenocarcinomas, 4 hyperplasias and 56 normal controls), clumps of epithelial cells that could be regarded as the most diagnostically relevant were selected in each case. Their cytologic character was reduced to a combination of five quantitative parameters (nuclear size, degree of anisokaryosis, nuclear form index, homogeneity of nuclear chromatin texture and regularity of nuclear arrangement). The five-variate cluster analysis demonstrated that the 70 cases could be classified into three groups: group A (17 cases) was characterized by cells with small nuclear size, slight anisokaryosis, homogeneous chromatin texture and regular nuclear arrangement; group C (12) by cells with large nuclear size, marked anisokaryosis, heterogeneous chromatin texture and irregular nuclear arrangement; and group B (41) by cells with intermediate parameter values. Group C was derived from 10 cases of adenocarcinoma and 2 of atypical hyperplasia, while groups A and B were not derived from any cases of malignancy. This preliminary study indicated that morphometric-statistical classification can be of great help in improving the cytodiagnostic validity and reproducibility of endometrial carcinoma, and it awaits further, practical testing within a significantly larger series of malignancy cases.