Determining the inter- and intraobserver reproducibility of the diagnosis of endometrial hyperplasia subgroups and well-differentiated endometrioid carcinoma in endometrial curettage specimens.

BACKGROUND: Many studies have attempted to identify histologic features that help to distinguish atypical hyperplasia from hyperplasia without atypia and well-differentiated endometrioid carcinoma of the endometrium; however, few have evaluated the reproducibility of these diagnoses. METHODS: Five pathologists independently reviewed 100 endometrial curettage specimens chosen to represent the spectrum of proliferative lesions of the endometrium. This included simple hyperplasia, complex hyperplasia, atypical hyperplasia, and well-differentiated endometrioid carcinoma. Slides were reviewed once for interobserver agreement among the five pathologists and twice for intraobserver agreement by one of them. RESULTS: The results were assessed using the weighted kappa statistic. The mean intraobserver kappa value was 0.86. The mean interobserver kappa values by diagnostic category were as follows: simple hyperplasia without atypia: 0.74; complex hyperplasia without atypia: 0.33; atypical hyperplasia: 0.34, and well-differentiated endometrioid carcinoma: 0.64; with a kappa value of 0.53 for all cases combined. CONCLUSION: A major interobserver discrepancy exists in the diagnosis of complex and atypical hyperplasia which are the most similar mimics of endometrioid carcinoma.

Reproducibility determination of WHO classification of endometrial hyperplasia/well differentiated adenocarcinoma and comparison with computerized morphometric data in curettage specimens in Iran.

BACKGROUND: Management of endometrial precancerous lesions has been of much debate due to inconsistencies in their classification, natural history and histologic diagnosis. Endometrial hyperplasia constitutes a wide range of histomorphologic features associated with high intra and interobserver diagnostic variability.Although traditional microscopic diagnosis is by far the most applicable method and the gold standard for histomorphologic diagnosis, digitized image analysis has been used as a powerful adjunct to maximize the histologic data retrieval and to add some detailed objective criteria for correct diagnosis in difficult cases. METHODS: A series of 100 endometrial curettage specimens with diagnosis of endometrial hyperplasia or well differentiated adenocarcinoma were blindly reviewed by 5 pathologists; their intra and interobserver reproducibility determined and further compared to the objective morphometric data i.e. D-score and volume percent of stroma (VPS). RESULTS: The results were assessed using the weighted kappa statistics. Mean intraobserver kappa value was 0.8690 (99.44% agreement). Mean interobserver kappa values by diagnostic category were: simple hyperplasia without atypia: 0.7441; complex hyperplasia without atypia: 0.3379; atypical hyperplasia: 0.3473, and well-differentiated endometrioid carcinoma: 0.6428; with a kappa value of 0.5372 for all cases combined.Interobserver agreement was in substantial rate for simple hyperplasia (SH) and well differentiated adenocarcinoma (WDA) but was in fair limit for complex hyperplasia (CH) and atypical hyperplasia (AH). Intraobserver agreement was almost perfect. The specimens were divided in two groups according to the computerized morphometric analysis: Endometrial Hyperplasia (EH) ( D Score > or = 1 or VPS > or = 55%) and Endometrial Intraepithelial Neoplasia (EIN) (D-Score < 1 or VPS < 55%). Morphometric findings were closely compatible with routine WHO classification made by one expert pathologist; however; diagnosis of (CH) and (AH) made by other pathologists were not concordant with morphometric data. CONCLUSION: It may be necessary to make some revisions in WHO classification for endometrial hyperplasia and precancerous lesions.