A limitation of endometrial biopsy is the difficulty in obtaining an adequate sample in some patients. Transvaginal ultrasonography has been recommended by some clinicians as an alternative to endometrial biopsy. Langer and colleagues, as part of the Postmenopausal Estrogen/Progestin Interventions (PEPI) trial, compared transvaginal ultrasonography with endometrial biopsy for the detection of endometrial hyperplasia in asymptomatic postmenopausal women.

The 875 women in the PEPI study were between 45 and 64 years of age and were post-menopausal for no longer than 10 years. All of the women received 0.625 mg of conjugated estrogen daily. Treatment groups included estrogen alone and several different dosing regimens of progesterone. Endometrial biopsies were normal at baseline in all of the participants and were repeated annually thereafter during the three years of the study. All biopsy specimens were evaluated by at least two pathologists. Approximately halfway through the study, transvaginal ultrasonography was added to the annual evaluation. This procedure was always performed within 48 hours before the endometrial biopsy.

A total of 448 women were included in the ultrasound portion of the PEPI study, and 577 transvaginal ultrasound examinations were performed. On the basis of data from four previous studies, 4 mm was selected as the maximal normal endometrial thickness. Endometrial thickness was greater than 4 mm in 307 (53.2 percent) of the 577 ultrasound examinations. Biopsy results were abnormal in 31 instances, or 5.4 percent of the 577 examinations. Most abnormal biopsy results were simple endometrial hyperplasia.

When the results of transvaginal ultrasonography were compared with histopathologic findings, a 5-mm threshold for abnormal endometrial thickness had a 90 percent sensitivity and a 48 percent specificity for detecting endometrial disease. The positive predictive value was 9 percent, and the negative predictive value was 99 percent; 53 percent of the women with normal biopsy results had an endometrial thickness of at least 5 mm. When simple hyperplasia was excluded from the analysis of patients with abnormal biopsy results, the positive predictive value of an endometrial thickness of more than 5 mm was low (i.e., 12 percent) in these high-risk patients.

The authors conclude that transvaginal ultrasonography is useful in detecting endometrial hyperplasia in asymptomatic women receiving hormone replacement therapy, but the high false-positive rate associated with ultrasonography makes it a poor screening test for following asymptomatic women whether or not they are receiving hormone replacement therapy.

editor's note: My interpretation of this study is that transvaginal ultrasonography is not a very good diagnostic alternative to endometrial biopsy in women receiving hormone replacement therapy. Other studies have found that the endometrial thickness does not normally exceed 4 to 5 mm in post-menopausal women. However, a significant number of women in this study had normal results on endometrial biopsy and an endometrial thickness of 5 to 9 mm. In an accompanying editorial, Fleischer cites a study in which estimates of endometrial thickness in the same patient differed by up to 1.5 mm among radiologists. Interestingly, he supports the use of transvaginal ultrasonography, both for monitoring the endometrial response to hormone replacement therapy and for evaluating unexplained vaginal bleeding. Since transvaginal ultrasonography can detect disease in the absence of symptoms, he believes it deserves a place in the routine evaluation of perimenopausal and, particularly, postmenopausal women. I believe the data suggest that family physicians should continue to perform endometrial biopsy as the diagnostic procedure of choice in symptomatic patients. The data do not suggest that transvaginal ultrasonography is of benefit in screening asymptomatic women.—j.t.k.