The performance of MRI versus x-ray mammography with varying degrees of breast density

While there is some variation in breast cancer screening programs across Canada, for women at average risk of breast cancer, x-ray mammography (XRM) is generally offered to women aged 50 to 74 every two years. One of the limitations of XRM is in the screening of women with dense breast tissue.¹ Breast density is divided into four categories from ACR density A (almost entirely fatty) to ACR density D (extremely dense). For women aged 40-74, 43% have dense breasts (categories C and D).² High breast density is a clinical challenge as the performance of XRM decreases with increasing breast density. Furthermore, women with dense breasts have an increased risk for breast cancer.³

In 2016, two large multicenter studies, Gadobutrol-enhanced MR mammography (GEMMA) 1 and GEMMA 2, investigating the use of Gadovist® for the assessment of the extent of malignant breast disease, were published.⁴ The GEMMA studies formed the basis for the regulatory approval of Gadovist® for breast MRI in Canada. In a recent additional analysis of the GEMMA 1 and GEMMA 2 data, Barkhausen et al. have compared the sensitivities of contrast-enhanced breast MRI (CE-BMRI) with XRM for detection of breast cancer in women with varying degrees of breast density.⁵ Sensitivities of the two modalities were compared both for index cancers, which were those cancers that triggered the breast cancer diagnosis, as well as for additional cancers, referring to those cancers detected during preoperative examination and/or by CE-BMRI.

A total of 774 women were included (386 women from GEMMA 1 and 388 from GEMMA 2) and there were 169 additional cancers. For the index cancers, the median reader sensitivity of CE-BMRI remained stable at 83% for GEMMA 1 and 91% for GEMMA 2, while the median reader sensitivity of XRM dropped from 79% to 62% for GEMMA 1 and from 82% to 64% for GEMMA 2 (Figure 1A). For the detection of at least one additional cancer, the median reader sensitivity of CE-BMRI increased from 50% to 73% for GEMMA 1 and from 57% to 81% for GEMMA 2, while the median reader sensitivity of XRM dropped from 34% to 20% for GEMMA 1 and remained stable at 24% to 25% for GEMMA 2 (Figure 1B). Similarly, for the detection of all cancers (index plus additional cancers), the sensitivity difference between CE-BMRI and XRM increased with increasing breast density (Figure 1C).

In summary, the sensitivity of CE-BMRI was relatively stable while the sensitivity of XRM decreased with increasing breast density. The benefits of the higher sensitivity provided by CE-BMRI in women with dense breasts has been evaluated by Bakker et al. in the DENSE trial.¹ Results from the first round of screening and two years of follow-up in the DENSE trial demonstrated a 50% reduction in interval-cancer rate from 5.0 per 1000 screenings with XRM alone to 2.5 per 1000 screenings when CE-BMRI was offered to women with extremely dense breast tissue.