Transvaginal ultrasonography (US) identifies abnormal pathologic conditions in the endometrium as diffuse or focal areas of thickening. Transvaginal hysterosonography is useful when further evaluation of endometrial thickening or better visualization is needed. However, little is known about how well the latter technique can distinguish benign abnormalities from malignancies. Dubinsky and associates established criteria for classifying benign and suspicious lesions viewed with transvaginal hysterosonography. They applied these criteria prospectively to radiographic interpretations and correlated all cases with histopathologic findings obtained at hysterectomy or at dilatation and curettage (D&C) to evaluate the predictive values of transvaginal hysterosonography in diagnosing malignant endometrial disease.

Women between 25 and 81 years of age who sought medical attention at a Houston gynecology clinic for significant vaginal bleeding were eligible for the study. Patients with negative findings on aspiration biopsy of the endometrium were referred for transvaginal US.

Results of the transvaginal hysterosonographic examinations were interpreted prospectively, and the pathology reports of patients who had had either a D&C or a hysterectomy within one month of the examination were reviewed retrospectively. A benign endometrial appearance was defined as an endometrial thickness of less than 4 mm, the presence of a smoothly echogenic mass in the endometrial lumen or a diffusely thickened endometrium of greater than 4 mm. A suspicious endometrial appearance was defined as a focal nonhomogenous mass projecting into the lumen or focal endometrial thickening of greater than 4 mm. Sensitivity, specificity and predictive values for predicting malignant disease were then calculated.

Eighty-eight women met the study criteria and underwent transvaginal hysterosonography, followed by D&C (37 patients) or hysterectomy (51 patients). Of the 37 women who underwent D&C, 30 had benign-appearing endometria, and 28 had diffuse, smooth endometrial thickening that was benign on histology. Two patients had a thin endometrium at hysterosonography, despite an appearance on transvaginal US that suggested a thickening. D&C in these two patients revealed an inactive endometrium. Seven patients had pedunculated homogenously echogenic luminal masses identified on hysterosonography, and all of these patients underwent hysterectomy. Histology revealed six adenomatous polyps and one focus of carcinoma in situ in a polyp.

Of the 51 women who underwent hysterectomy, eight had endometrial carcinoma, 24 had adenomatous polyps, five had hyperplasia, 11 had fibroids and three had endometritis, which was subsequently cultured and found to be positive for Chlamydia.

The proposed classification system for detecting endometrial carcinoma with transvaginal hysterosonography had a sensitivity of 89 percent, a specificity of 46 percent, a positive predictive value of 16 percent and a negative predictive value of 97 percent.

The authors conclude that transvaginal hysterosonography is useful in identifying benign pathologic conditions of the endometrium and indicating when biopsy is warranted. Most endometrial abnormalities, including carcinoma, appear as a focal mass with this examination. The authors' criteria were highly predictive for benign processes, particularly endometrial thickening. However, because many pathologic conditions had a suspicious appearance, the positive predictive value of transvaginal hysterosonography for carcinoma was low. Therefore, women with multifocal or sessile lesions should undergo a guided biopsy, and any benign-appearing polyps should be removed to rule out carcinoma. Transvaginal hysterosonography should be used to identify patients who need more invasive therapy if an endometrial lesion is detected or to provide medical therapy if no lesion is present.

editor's note: One major limitation of this study was the use of D&C to obtain the endometrial sample. This procedure is now considered a blind sampling technique, not the gold standard it once was. Of note, however, is that the histologic samples obtained with D&C were consistently abnormal in patients whose transvaginal hysterosonogram showed a focal mass.—b.a.