Accuracy of hysteroscopic diagnosis of endometrial hyperplasia: a retrospective study of 323 patients.

Study Objective: To evaluate the diagnostic accuracy of hysteroscopic view in endometrial hyperplasia.

Design: Retrospective study (Canadian Task Force classification II-2).

Setting: Public hospital in northern Italy.

Patients: Three hundred twenty-three patients suffering from endometrial hyperplasia out of 2251 women (1119 premenopausal and 1132 postmenopausal) who underwent office-based hysteroscopy from January 1996 through May 2004.

Intervention: Review of 2251 outpatient hysteroscopies carried out with 5- to 6-mm sheathed hysteroscopes and accomplished with blind or hysteroscopically targeted endometrial biopsies.

Measurements And Main Results: The pathologic report was considered the reference test. Hysteroscopic detection of focal or extensive endometrial thickening, irregular vascular network, architectural distortion and crowding of gland openings, and gland cyst formation were considered endoscopic features consistent with hyperplasia. Overall sensitivity, specificity, negative predictive values (NPV), and positive predictive values (PPV) of hysteroscopy in order to foresee a diagnosis of hyperplasia were calculated. These figures were calculated both in premenopausal and postmenopausal patients. Histopathology yielded a diagnosis of simple, complex, and atypical hyperplasia in 247, 51, and 25 patients, respectively. Hysteroscopy foresaw hyperplasia in 38.4% of patients with simple hyperplasia and in 58.9% of patients with complex or atypical hyperplasia. Normal hysteroscopic findings underestimated simple hyperplasia in 34 patients (13.7%) and complex or atypical hyperplasias in 1 patient (1.3%) (p <.01). To predict the diagnosis of hyperplasia, hysteroscopy showed an overall sensitivity, specificity, NPV, and PPV of 63.7%, 91.7%, 91.3%, and 64.7%, respectively. Among premenopausal patients, hyperplasia was diagnosed in 134 women (11.9%); in this group, hysteroscopy showed sensitivity, specificity, NPV, and PPV of 65.6%, 88.5%%, 93.5%, and 50.5%, respectively. In postmenopausal patients, we found endometrial hyperplasia in 189 women (16.6%); sensitivity, specificity, NPV, and PPV of hysteroscopic view to anticipate hyperplasia were 61.6%, 95.2%, 89.3%, and 79.4%, respectively. A significantly better PPV to foresee hyperplasia was found in postmenopausal women compared with premenopausal patients (p <.01).

Conclusions: Current hysteroscopic criteria suggesting endometrial hyperplasia are inaccurate; in order to exclude hyperplasia, a pathologic assessment is warranted in all hysteroscopies showing an irregularly lined or thick endometrium.