Current Status of Supplemental Screening in Dense Breasts.

Abstract

Currently, 24 American states have laws requiring that women receive some level of notification about breast density with their mammography results. Dense breast tissue can hide cancer on mammography, especially when the cancer lacks calcifications, resulting in delayed diagnosis and worse outcomes. Moreover, dense breast tissue is an independent risk factor for developing breast cancer. Particularly in dense (heterogeneously dense or extremely dense) breasts, a negative result on mammography does not reliably exclude the presence of cancer. Advocates pushed for legislation requiring that notification of breast density be given with mammography results so that women with dense breasts would be aware of the implications, and could pursue supplemental screening beyond mammography. The most widely available supplemental screening options for women with dense breasts are ultrasound (US) and tomosynthesis (three-dimensional [3D] mammography), but there has been a lack of information to guide the decision to have one or the other versus both. Preliminary results from the Adjunct Screening With Tomosynthesis or Ultrasound in Women With Mammography-Negative Dense Breasts (ASTOUND) trial, an important study of adjunct screening with US and tomosynthesis (3D mammography) in women with dense breasts, are presented in the accompanying article in Journal of Clinical Oncology. Even with modern treatments for breast cancer, stage at diagnosis, and especially node status, remain the most important prognostic factors. Across 11 randomized trials of mammography, only those that reduced the rate of advanced breast cancer (and increased detection of node-negative invasive cancer) produced breast-cancer mortality reduction. In women with dense breasts, invasive cancers may be masked and missed on mammography; as a result, there is an excess of late-stage disease (stages II and III). In women with extremely dense breasts, cancers are nearly 18 times more likely to be found because of clinical symptoms soon after a normal screening mammogram than in women with fatty breasts. Such interval cancers (ie, diagnosed before the next routine screening examination) tend to be larger, more aggressive, and have worse prognoses than those found on screening. Thus, another measure of screening effectiveness is a low interval cancer rate (typically less than one per 1,000 women screened per year). Results from screening US have been reported in more than 200,000 women. The vast majority of US studies have examined its addition to mammography for women with dense breasts, and the vast majority of cancers seen only on US prospectively are not evident on mammography, even in retrospect. In a review of 335 cancers seen only on screening US, Bae et al reported that 263 (78%) were obscured by dense tissue on mammography, 63 (19%) were interpretive errors on mammography, and nine (3%) were not included mammographically because of difficult location or positioning. When performed by physicians, US produces consistent increased detection of an average of four cancers per 1,000 women screened. More than 85% of cancers detected only on screening US are invasive and node negative. Technologists can also be trained to perform whole-breast handheld screening US. On video review of results from training courses in Japan, 415 technologists showed significantly better performance in lesion detection than 422 physicians (85.9% v 84.0%, respectively; P5 .037) and higher specificity on still images (86.6% v 85.1%, respectively; P 5 .026), with indistinguishable performance on other tasks. In recently published results from the Japan Strategic Anti-Cancer Randomized Trial (J-START), there was a reduction in interval cancers (0.05% v 0.10%; P 5 .034) among 32,105 women receiving supplemental US compared with 32,812 women in the control group receiving only mammography; notably, only 57.7% of women in J-START had dense breasts, although results by breast density were not detailed. Compared with results from physician-performed US, studies of technologist-performed screeningUS have shown slightly lower added cancer detection, averaging 2.5 per 1,000 women in the first year. Several different approaches to automated whole-breast US have been studied. One method uses an automated arm with standard handheld equipment and, in a multicenter experience, showed detection of an additional 3.6 cancers per 1,000 women screened after mammography. Another approach uses a transducer with a wide footprint (typically 15 cm) and produced a supplemental yield of 1.9 per 1,000 women in a prospective multicenter experience, although 13% of women required recall for additional targeted US to complete initial screening. Results are expected to improve with incidence screening. There are several barriers to implementing screening US in practice. One of these has been high rates of false positives due to US. Importantly, in preliminary results from the ASTOUND trial, false-positive recalls (2.0%) and biopsies (0.7%) were