The endometrial echo revisited: have we created a monster?

Transvaginal ultrasound has been explored as an inexpensive, noninvasive, convenient way to indirectly visualize the endometrial cavity. For more than a decade numerous studies have indicated that a thin, distinct, well-visualized echo (<4-5 mm) in postmenopausal women with bleeding is as effective as any diagnostic modality in excluding endometrial cancer (99% negative predictive value). Unfortunately, this is not the same as saying that a thick endometrial echo is pathologic. In fact, the positive predictive value of an echo greater than 5 mm is less than 10% for any disease and only 4% for serious disease (cancer or hyperplasia). No studies validating the clinical significance of a nonthin endometrial echo observed in an incidental imaging study have ever been performed. Because 5 mm has been a "cutoff" for excluding endometrial cancers in women with bleeding, many clinicians have assumed that any findings greater than 5 mm need endometrial sampling to exclude disease. The number of postmenopausal women with quiescent fibroids, or polyps, or heterogeneous uterine echoes for technical reasons (previous scarring, axial uterus) is unknown but not insignificant. Furthermore, if transvaginal ultrasound is to be used, it must be performed appropriately, further recognizing that in a substantial number of patients it may not be possible to obtain technically adequate endometrial assessment. So, although transvaginal ultrasound can be a reliable method of excluding disease in many postmenopausal women with bleeding, the incidental finding of a non-thin endometrial echo has not been investigated and should not automatically trigger a need for formal tissue sampling.