Hyperplasia and carcinoma in secretory endometrium: a diagnostic challenge.

The diagnosis of endometrial hyperplasia or carcinoma in a background of secretory endometrium can be difficult. We attempt to establish the diagnostic criteria to be used in such cases. We examined 80 cases of endometrial hyperplasia, carcinoma, and other conditions with glandular crowding arising in secretory endometrium, analyzed their morphologic features, assessed the volume percentage stroma in each case and performed Ki67 immunostaining on 27 cases. Thirteen cases each of secretory and gestational endometrium served as controls. The mean age of the patients was 45 yr. The non-neoplastic diseases included simple hyperplasia without atypia (56%), endometrial polyps (12.5%), and chronic endometritis with glandular crowding (3%). The proportion of cases with complex hyperplasia without atypia was 10%. Neoplastic diseases included atypical complex hyperplasia (12.5%) and endometrioid carcinoma (6%). The secretory changes were usually less advanced in the hyperplastic glands than in the background endometrium. The morphologic features that best distinguished hyperplasia or carcinoma from secretory endometrium included glandular crowding that stood out from the background; architectural disorder (the long axes of the glands pointing in different directions or parallel to the endometrial surface); dilated, irregularly shaped glands, including budding or branching glands and staghorn-shaped glands; stroma of a polyp; cribriform or confluent glands in cases of carcinoma; nuclear atypia in cases of atypical hyperplasia and carcinoma; and crowded nonsecretory glands. The volume percentage stroma of neoplastic lesions was less than that of non-neoplastic ones (34% vs. 61%, P=0.000001) and that of secretory endometrium (34% vs. 68%, P=0.000038). Non-neoplastic lesions did not have significantly more crowded glands than secretory endometrium (61% vs. 68%, P=0.11). Gestational endometrium had more crowded glands than non-neoplastic lesions (39% vs. 61%, P=0.000004), an approximately equal volume percentage stroma with complex hyperplasia without atypia (39% vs. 43%, P=0.51), and less crowded glands than neoplastic lesions (39% vs. 34%, P=0.03). The Ki67 index of the neoplastic lesions was higher than that of the controls, including secretory and gestational endometria (positive nuclei per 100 epithelial cells, 44.8 vs. 4.6, P=0.0004), of the non-neoplastic lesions (44.8 vs. 5.4, P=0.002) and of complex hyperplasia without atypia (44.8 vs. 9.3, P=0.007). Hyperplasia and carcinoma in secretory endometrium can be diagnosed on the basis of increased glandular crowding, architectural irregularity, nuclear atypia, and increased Ki67 index.