MRI Improves Breast Cancer Detection in Women at Average Risk

Get Permission

Christiane K. Kuhl, MD

A. Marilyn Leitch, MD

Breast MRI improves the detection of small, high-grade cancers in women at average risk to an extent that the interval cancer rate is zero. In averagerisk women, breast MRI–screening alone every 3 years may therefore be sufficient.

—Christiane K. Kuhl, MD

Magnetic resonance imaging (MRI) screening of women at average risk for breast cancer achieved a mean additional cancer yield of 15.8 cases per 1,000 women, greatly surpassing yields for supplemental digital breast tomosynthesis (1.25 per 1,000) or supplemental ultrasound (4.1 per 1,000). The study, presented at the 2015 Breast Cancer Symposium, also showed that in average-risk women, the contribution of mammography to early diagnosis will be limited.1

“Breast MRI improves the detection of small, node-negative, high-grade cancers in women at average risk to an extent that the interval cancer rate is zero,” said Christiane K. Kuhl, MD, Department of Diagnostic and Interventional Radiology at RWTH Aachen University in Germany. “In average-risk women, breast MRI–screening alone every 3 years may therefore be sufficient.”

Over- and Underdiagnosis

“Between 30% and 50% of breast cancers that are diagnosed in women who participate in mammography screening programs will not be diagnosed by mammography, but as so-called ‘interval cancers’ in between screening rounds,” said Dr. Kuhl. “In other words, mammography fails to detect up to half of all breast ­cancers.”

At the same time, she noted, because detection of cancer in mammography is based on pathophysiologic processes that reflect regressive changes, overdiagnosis may be an unavoidable effect of the procedure.

Dynamic contrast-enhanced MRI, on the other hand, is a biomarker; detection of breast cancer in dynamic contrast-enhanced MRI is based on tissue alterations that directly correlate with carcinogenesis, cell proliferation, and even metastatic potential.

“MRI will preferably detect cancers that are biologically active and prognostically relevant, whereas it is relatively ‘blind’ for inactive disease or ‘dormant’ DCIS [ductal carcinoma in situ],” said Dr. Kuhl.” And yet, MRI screening has only been used so far in high-risk women—there’s no existing evidence for its use in women with average risk of breast cancer.

Study Design

To explore this issue, researchers recruited 2,120 women who underwent a prospective, two-center, comparative, diagnostic-accuracy study. Participants (age 40 to 70) had normal screening mammograms and, in dense breasts, normal screening ultrasound at the time of study inclusion. “Prior MRI screening studies had included highly selected women at elevated or high risk of breast cancer. In this study, we looked at the much larger group of regular women without specific risk factors other than possibly dense breast tissue,” Dr. Kuhl explained. 

The women were randomly assigned to undergo dynamic contrast-enhanced breast MRI and mammography every 12, 24, or 36 months, plus follow-up of 2 years. A total of 3,861 MRI studies covering 7,007 women-years were conducted over the course of the study.

Key Results

The vast majority of MRIs (91.0%) were read as normal or benign. Women with suspicious findings (4.4%) underwent biopsy. From this group, 35.7% of lesions were ultimately found to be malignant, 24.6% revealed high-risk changes, and 39.8% revealed benign changes.

“This translates into a positive predictive value of 35.7% for cancer,  which compares favorably to the respective value for digital breast tomosynthesis (29.3%) or ultrasound screening (8.0%–16.0%).” said Dr. Kuhl.

Ultimately, 61 cancers were identified in 2,120 women for an overall detection rate of 28.8 per 1,000 women. Of these 61 cancers, none was only mammography-detected, one was found by both mammography and MRI, and 60 were only MRI-detected, giving a supplemental cancer yield of 15.8 per 1,000 screening rounds. This compares to an average supplemental yield of 1.25 per 1,000 for tomosynthesis, or 4.1 per 1,000 for ultrasound screening.

Most cancers (22.6 per 1,000) were found in the initial screening round, substantially less (2.7 per 1,000 women-years) during the 4,887 subsequent screening rounds  No interval cancers were observed, not even in women undergoing MRI screening only every 3 years.

Dr. Kuhl also noted that, in accordance with dynamic contrast-enhanced MRI being a biomarker, the added cancers diagnosed by MRI had pathologic features of biologic ­importance.

“The cancers detected skewed toward a higher-than-normal rate of high-grade and estrogen receptor/progesterone receptor–negative cancers,” said Dr. Kuhl. “Thus, cancers detected by MRI had similar biologic profiles as the usual interval cancers—interval cancers that were not observed in our study but that would have been expected to be seen in this cohort if women had undergone mammographic screening alone.”

Clinical Trial Needed

Discussant A. Marilyn Leitch, MD, Professor of Surgery at The University of Texas Southwestern Medical Center at Dallas, emphasized that this was a highly selected population with negative mammograms.

“One of the concerns we hear about MRIs,” said Dr. Leitch, “is that false-positive rates are higher than were reported in this study. MRI detects lesions that might drive patients to avoid breast-conserving surgery, and there can be unreasonable costs of screening and workup of false-positives.… Even the most liberal screening guidelines in the United States reserve MRI for high-risk patients.”

What is needed, said Dr. Leitch, is a clinical trial. “It would be useful to compare screening MRI at 3-year intervals to an annual tomosynthesis mammogram in all average-risk women,” she concluded.

While agreeing with the need for further clinical trials, Dr. Kuhl replied that the population was everything but highly selected. “These were regular women without specific risk ­factors.”

She also took exception to the notion of high false-positive rates. “The positive predictive value for MRI in our study was higher than that of digital breast tomosynthesis in the screening situation,” said Dr. Kuhl. “A high positive predictive value and an interval cancer rate of zero make up a perfect screening test.”

She concluded: “As long as breast cancer continues to represent one of the main causes of cancer death in women, the search for improved screening methods should continue. Such methods should combine a maximum sensitivity for cancers that have the potential to kill, with a desirable lack of sensitivity for disease that is prognostically unimportant. This is what MRI screening may achieve.”  ■

Disclosure: Drs. Kuhl and Leitch reported no potential conflicts of interest.


1. Kuhl C, Strobel K, Bieling H: Breast MRI screening of women at average risk of breast cancer: An observational cohort study. 2015 Breast Cancer Symposium. Abstract 1. Presented September 25, 2015.