Our objective was to assess clinical and pathological factors associated with a final diagnosis of endometrial carcinoma in patients with atypical endometrial hyperplasia with a particular emphasis on the grading of atypia.

Materials and methods
A retrospective review over five years on patients (N = 97) who underwent hysterectomy for a diagnosis of atypical endometrial hyperplasia at a statewide public tertiary gynaecologic oncology centre. Clinical and pathological characteristics were obtained.

The rate of concurrent endometrial carcinoma was 34% (n = 33) with most being stage 1A endometrioid. A significant group difference was reported for age at diagnosis (t = −2.20 P = 0.031 d = 0.43) with carcinoma patients on average older (Mage = 60.2 (8.9) years) than patients without carcinoma (Mage = 55.5 (12.3) years). No significant group differences were found for body mass index, endometrial thickness or time between diagnosis and treatment. Significantly higher rates of carcinoma were reported in patients with moderate atypical hyperplasia (27.6%) and severe atypical hyperplasia (66.7%), compared to mild atypical hyperplasia (7.1%). Only severe atypical hyperplasia (odds ratio (OR) = 21.5, 95% CI 2.8–163.1, P = 0.003) and postmenopausal status (OR = 13.2, 95% CI 1.3–139.0, P = 0.032) significantly increased the risk of carcinoma in a multivariate model.

Severe atypical hyperplasia and postmenopausal status were significant predictors of concurrent endometrial carcinoma in patients with atypical endometrial hyperplasia. The grading of atypical hyperplasia may be utilised by gynaecologic oncologists in the triage and referral process of managing these patients; however, the grading system requires external validation in larger prospective studies.