The use of transvaginal ultrasonography (TVUS) for evaluation of endometrial pathology in postmenopausal women has generated interest because of its noninvasiveness. It has been assumed that endometrial thickness measured as the endometrial stripe on TVUS would help clinicians decide whether a patient should undergo endometrial biopsy. TVUS has primarily been employed in symptomatic women using hormone replacement therapy (HRT) with estrogen and progestin or with estrogen only (ERT). Little data exist on expected endometrial thickness in postmenopausal women who are using ERT or HRT. Archer and associates studied endometrial thickness in women using hormonal therapy to see if it differed from that in women not using hormone therapy. They also sought to determine whether bleeding and endometrial thickness were associated.
A total of 148 healthy postmenopausal women (43 to 70 years of age) entered the study; 121 were taking oral hormone preparations. The women were divided into five groups: (1) women who had not used any hormone therapy in the six months before the study; (2) women who had used cyclic or continuous oral ERT for at least three months; (3) women who had used oral ERT for at least three months and had irregular vaginal bleeding or spotting; (4) women using a cyclic HRT regimen for at least three months; or (5) women who had used HRT for at least three months and had irregular bleeding or spotting. TVUS was performed on the same day but before the endometrial biopsy was obtained. Endometrial biopsies were performed using routine endometrial aspirators.
No women were found to have endometrial carcinoma or adenomatous hyperplasia with atypia. Results of endometrial biopsy showed abnormal endometrial changes for 15 subjects, of whom 11 had hyperplasia without atypia. For all subjects, endometrial thickness ranged from 1 to 25 mm. The median endometrial thickness was 5.0 mm in older patients and 5.25 mm in younger patients.
The median endometrial thickness was 6.1 mm in the subjects taking hormones who were experiencing spotting or bleeding, compared with 4.0 mm in the control group. This difference was significant.
When all hormone groups were compared with the control group, the median thicknesses were 4.0 mm (control group), 5.5 mm (combined HRT groups) and 6.5 mm (combined ERT groups). The HRT and ERT groups had significantly greater endometrial thicknesses than the control group. The endometrial thickness of women using ERT was thicker than that of women using HRT, but the difference was not significant. Across all study groups, a measurement greater than 5 mm had a sensitivity of 67 percent, a specificity of 52 percent and a positive predictive value of 14 percent for determining which women had endometrial histology consistent with low estrogen stimulation. For a measurement greater than 4 mm, sensitivity was 38 percent, specificity was 73 percent and the positive predictive value was 12 percent.
Results were summarized as to the number of subjects who had abnormal and normal endometria; results of TVUS were grouped according to endometrial thickness of 5 mm or less and 4 mm or less. The results in the treated women did not show that a meaningful cutoff could be established with TVUS for correlation with endometrial pathology. However, the data for the control group support the value of using endometrial thickness as an indicator of a nonpathologic endometrium.
If a thickness greater than 5 mm had been regarded as a preliminary indicator of abnormal endometrium, 70 women would have qualified for endometrial biopsy. However, based on subsequent endometrial biopsy, only 10 of these women actually had an abnormal endometrium. A criterion of greater than 4 mm was as unreliable as 5 mm or less. Fewer women with endometrial abnormalities would have been missed, but the four who would have been missed included two with hyperplasia. In addition, 89 women would have been identified as having potentially abnormal endometrial findings although 78 of them had a normal endometrium based on endometrial biopsy.
The authors conclude that endometrial thickness assessed by TVUS is not a reliable indicator of a normal endometrium and cannot be used to rule out pathology. Thus, regardless of the thickness of the endometrium, any unscheduled bleeding during HRT should be further investigated. Even in women using HRT without abnormal bleeding, pathologic findings may be present with an endometrium that is less than 4 mm thick, although only two of 49 women in the study were affected. This study does not show a correlation adequate to recommend the use of screening TVUS as a predictor of endometrial findings.