Media misleads on MRI screening for breast cancer

December 6th, 2019

Earlier this year, the U.S. Food and Drug Administration (FDA) proposed adding breast density to the information patients receive after a mammogram. Over half of women over age 40 have dense breasts, which raises one’s risk of developing cancer and makes tumors more difficult to spot on mammograms, so it makes sense that women would want to know their breast density.

However, the question remains–what are women supposed to do with this information? The American College of Obstetricians and Gynecology does not recommend routine use of other scans like ultrasound or MRI to screen for breast cancer in women with dense breasts. Supplementing mammograms with ultrasound has not been shown to improve cancer detection or survival. MRIs too show high false positive rates and are not recommended for women with an average risk of breast cancer.

But a recent study in the New England Journal of Medicine and associated media coverage has generated more hype about supplemental screening for cancer for women with dense breasts. The NEJM study was a randomized trial of 40,373 women with dense breasts who had previous normal (negative) mammogram results. About 8000 were randomly assigned to be invited to get a supplemental MRI screening as well as a mammogram and 4,783 women accepted the invite.

The study showed that, over two years, MRIs detected more cancers than mammograms alone for women with dense breasts. Women in the MRI-invite group were 50% less likely to be diagnosed with an “interval cancer,” or a cancer that was detected in between imaging tests, compared to those in the group who only received a mammogram. This finding indicates that MRIs found more cancers that mammograms missed. However, the rate of cancer detection among all of the trial participants was low. The rate of interval cancers detected in the mammogram-only group was 0.5%, compared to 0.25% in the MRI-plus-mammogram group. 

While MRIs may detect more cancers, it is unknown whether finding these cancers actually improves breast cancer survival. Many cancers we find in screening tests would never grow or become harmful, but we cannot distinguish between harmless cancers and dangerous ones. As NEJM editor Dan Longo wrote in an accompanying editorial, “We do not know whether the cancers that were detected needed to be found or treated.” Looking at the rate of “node-positive” cancer detected (cancer that has spread to the lymph nodes) in the study, both the MRI and mammogram groups had similar detection rates. However, MRIs found ten times more cases of ductal carcinoma in situ (DCIS), a form of breast cancer that has a low risk of becoming invasive. 

The study also found several downsides to MRIs for breast cancer screening. MRIs are more sensitive tests, but less specific–they pick up more potential cancer, but also find many false positives. Out of the women who received an MRI as well as a mammogram, nearly 10% were found to have an “abnormal scan.” Three hundred women had a biopsy to test for cancer after the MRI, and 74% of these women were found not to have cancer. Among the women who underwent MRI, 0.1% had either an adverse event or a serious adverse event during or immediately after the screening.

To sum up: For every 1000 women with dense breasts screened with MRI, compared to those screened with mammogram only, there were

  • 2.5 fewer interval cancers
  • 0 more cases of node-positive cancer found
  • 2.8 more cases of DCIS found
  • 80 false positive results
  • 1 adverse event during or immediately after the MRI
  • No more saved from breast cancer (that we know of)

Not exactly a resounding success. This isn’t so surprising, because breast density alone is not a perfect predictor of breast cancer risk. In fact, definitions of breast density are variable depending on radiologists’ readings; at least 13% of women will have their breast density classification change over a two-to-three year period. Using breast density as a proxy for risk will result in many false positives, because the rate of breast cancer incidence among women with dense breasts (without other risk factors) is low. Add in the high sensitivity and low specificity of MRI scans, and you get an even higher rate of false positives. 

However, as Longo correctly foretold, “the findings of this trial are likely to reinforce the idea that MRI screening is important in women with dense breast tissue.” Media coverage of this study is already reinforcing the idea that women with dense breast tissue benefit from getting additional screening. For example, look at these headlines:

MRIs could help detect breast cancer, new study finds – Becker’s Hospital Review

Women With Dense Breast Tissue May Benefit From Regular MRIs, New Report Suggests – Moms.com

MRI screening for women with extremely dense breast tissue reduces interval cancer – CNN

The New York Times piece is especially unbalanced. The article quotes Dr. Wendie Berg, a professor of radiology at University of Pittsburgh School of Medicine and chief scientific advisory for DenseBreast-info.org, as saying the study’s finding are “very important.”

“The standard practice of mammography alone is not adequate for women with extremely dense breasts,” Dr. Berg said. “Women should be advocating for increased access to screening M.R.I.”

However, the New York Times piece does not mention that DenseBreast-info.org is funded in part by medical device companies Hologic, Volpara Solutions, Siemens, Hitachi, and GE Healthcare, who manufacture imaging machines. The omission of financial conflict of interest information for this source is a serious oversight. Further, the reader must get through the first nine paragraphs of the piece before seeing any information on lack of survival benefit for MRI breast screening or the high false positive rate for MRI.

When media sources emphasize the perceived benefits of cancer screening and downplay the harms, it perpetuates the unsupported view that early detection is always better. In reality, offering MRIs to all women with dense breasts without taking into account their overall cancer risk will undoubtedly result in more overdiagnosis and harm.