Management of endometrial precancers is compromised by longstanding debate over the natural history of endometrial hyperplasias and inconsistencies in their diagnosis. The recent demonstration that some hyperplasias, like cancers, are phenotypically monoclonal is useful in recognizing biological precancers. A clonal analysis has been undertaken of a series of 93 endometrial tissues and their morphology has been evaluated by subjective diagnostic classification and computerized morphometric analysis. A pathologist's diagnosis of atypical endometrial hyperplasia was highly associated with monoclonal growth. Both microsatellite-stable and microsatellite-unstable precancers were classified as atypical hyperplasias, indicating overlapping morphologies for these two groups. Diagnosis of non-atypical endometrial hyperplasias was not reproducible and identified a group of lesions equally likely to be monoclonal as polyclonal. Computerized morphometry resolved these lesions into monoclonal and polyclonal subgroups with a high degree of accuracy and reproducibility. The predictive value of morphometry was dominated by that fraction of the sample which consisted of stroma (volume percentage stroma). This can be measured manually and used to predict monoclonality when below the threshold value of 55%. This study shows that morphometric analysis reproducibly and precisely identifies monoclonal endometrial precancers from histological sections. It may serve, furthermore, to classify accurately lesions judged by pathologists as indeterminate (non-atypical hyperplasias). The material from this study (available at www.endometrium.org from March 1, 2000) and precisely defined architectural diagnostic criteria provide new tools for diagnostic standardization of endometrial precancers.
Copyright 2000 John Wiley & Sons, Ltd.