Am Fam Physician. 2020;101(11):695-696
Does supplemental magnetic resonance imaging (MRI) screening for women with very dense breasts reduce the number of interval cancers?
Supplemental MRI screening for women with very dense breasts compared with mammography alone every two years significantly reduces the likelihood of an interval cancer, from 5 per 1,000 to 2.5 per 1,000 in the intention-to-treat population and to 0.8 per 1,000 in the per-protocol population. False-positive results were common, and there were more overall and early-stage cancers detected in the MRI group, raising the concern that many of these cancers may have been present but growing slowly and indolently (so-called overdiagnosed cancers). Subsequent follow-up will hopefully determine whether mortality and not just incidence is affected. (Level of Evidence = 1b−)
Most guidelines recommend mammography every two years, typically in women 50 to 69 or 75 years of age. The only exception is the American College of Radiology, which continues to recommend annual screening. In this Dutch study, women undergoing routine digital mammography who were identified as having very dense breast tissue (grade 4/4) and who had a normal digital mammogram result were randomized in a 1:4 ratio to receive supplemental MRI screening or usual care. After randomization, the women in the MRI group were notified and invited to participate. Obtaining consent after randomization is known as a Zelen design and is thought to reduce protocol violations and anxiety in the women not randomized to MRI. Women with a Breast Imaging Reporting and Data System (BI-RADS) MRI score of 4 or 5 were recalled for further evaluation, including biopsy. Women with a BI-RADS score of 3 had a second reading of the MRI, and if the second reading was also 3, they had a follow-up MRI in six months. The primary outcome was the likelihood of interval cancer, defined as all cancers detected in the 24 months following a negative index digital or MRI mammogram and before the next scheduled mammogram (or within 24 months if aging out of the cohort). A total of 8,061 women were randomized to receive supplemental MRI screening; 4,783 (59%) agreed to the screening. The comparison group had 32,312 women.
Of the 4,783 women who underwent supplemental MRI screening, 79 (1.65%) had breast cancer detected. In the intention-to-screen analysis that included all 8,061 women randomized to MRI, the rate of interval cancers was lower than in the usual care group (2.5 vs. 5.0 per 1,000 women). This is the appropriate comparison because in other mammography trials, women who ignored the invitation to screen had worse health outcomes than those who chose to volunteer. Among women randomized to MRI, the rate of interval cancers was 0.8 per 1,000 women in those who volunteered and 5 per 1,000 women (nearly identical to the control group rate of 4.9 per 1,000) in those who did not. With regard to harms, 9.5% of women undergoing supplemental MRI screening were recalled, and 6.3% of all women had a biopsy. The false-positive rate was 8% among women undergoing supplemental MRI. Of the 79 cancers detected in the MRI group, approximately 80% were invasive, and the remainder were ductal carcinoma in situ. The characteristics of the interval cancers did not differ significantly between groups. At the second round of screening, the rate of invasive cancers was lower in the MRI group than in the digital mammography–only group (2 vs. 7 per 1,000), and they were more likely to be stage 0 or 1 cancers, suggesting that MRI advanced the time of detection. The study was not powered to detect a reduction in mortality.
Study design: Randomized controlled trial (nonblinded)
Funding source: Government
Reference: Bakker MF, de Lange SV, Pijnappel RM, et al.; DENSE Trial Study Group. Supplemental MRI screening for women with extremely dense breast tissue. N Engl J Med. 2019;381(22):2091–2102.
Editor's Note: Dr. Ebell is deputy editor for evidence-based medicine for AFP and cofounder and editor-in-chief of Essential Evidence Plus, published by Wiley-Blackwell.