Breast imaging has two distinct roles: to screen asymptomatic women in an effort to detect cancer that is likely to be cured by therapeutic interventions, and to assist in the selection of local therapy. For many years, mammography was the only modality available for these purposes. The local therapy of breast cancer has changed dramatically over the past 30 years, and breast-conserving therapy (BCT) and neoadjuvant therapy for operable cancers have become part of routine clinical practice. Our understanding of breast cancer risk has also evolved with the recognition of the BRCA1 and BRCA2 genetic mutations, allowing the identification of a subset of women at extremely high risk of breast cancer development.
In parallel with these changes, imaging techniques have also evolved. Ultrasound is now routinely used in the evaluation of the patient with known cancer, although its role in screening remains controversial. Magnetic resonance imaging (MRI) of the breast was added to the diagnostic armamentarium of the physician in 1984. Increasing experience with this technology has raised important questions about how it should be used in screening, and for presurgical evaluation and posttherapy follow-up of women with breast cancer. This article will review the use of MRI in these settings, with an emphasis on the clinical outcomes that have been observed to date.
Mammography remains the standard of care for screening and detection of breast cancer. It is the only breast screening tool that has been proven in randomized trials to reduce breast cancer mortality, and approximately 50% of the recently observed reduction in breast cancer mortality in the United States is attributed to the use of screening mammography. Mammography, however, cannot distinguish between solid and cystic masses, misses approximately 10% to 15% of cancers, and has a lower sensitivity for the detection of lobular cancers, cancers in dense breasts, and cancers occurring in the presence of breast implants.
Screening Known or Suspected BRCA1 and BRCA2 Carriers
The detection of cancer with MRI is dependent upon the increased vascularity of neoplasms, which results in enhancement after the injection of contrast material, and is not limited by the density of the breast tissue. Because benign lesions also enhance, the combination of the presence of enhancement, the kinetics of enhancement, and the morphology of the lesion are used to distinguish benign from malignant lesions. Large-scale trials of screening with MRI in unselected populations have not yet been carried out. Screening studies to date have focused on high-risk populations, particularly women known or suspected to have mutations of BRCA1 or 2. There is a clear biologic rationale for seeking improved screening techniques for this group of women since studies have demonstrated a high rate of interval cancers in mutation carriers screened with annual mammography alone.
Five prospective, nonrandomized screening trials comparing the use of mammography and MRI in high-risk women are summarized in Table 1.[3-7] Women were eligible for these trials if they had a lifetime risk of breast cancer development of at least 15%, and the proportion of proven BRCA1/2 mutation carriers ranged from 8% in the study by Kuhl et al to 100% in the study by Warner et al. In spite of the heterogeneity in study designs, the results of these trials are remarkably similar. The sensitivity of MRI for the detection of cancer ranged from 77% to 100% and was substantially higher than that observed for mammography (25%–40%). In all of the studies, the specificity of MRI was lower than that of mammography, although in the majority of the reports the difference was less than 5% (Table 1).
The exception to this trend was the study by Leach et al, which included 22 centers in the United Kingdom, and reported a specificity of 81% for MRI and 93% for mammography. These results are probably more representative than those reported from single-institution studies, where high volumes of MRI were performed and interpreted by individuals with great levels of expertise.
It is important to recognize that the lack of specificity of MRI potentially results in additional imaging in a significant number of women. In the study by Kreige et al, the recall rate for additional imaging based on MRI findings was 10.7% compared to 3.9% for mammography, and the biopsy rates for these modalities were 3.1% and 1.3% respectively. While this may be an acceptable trade-off for increased cancer detection rates in populations at extremely high risk of breast cancer development (such as mutation carriers), it represents a significant drawback to the more widespread use of MRI screening in women at a lower level of risk.
Although a reduction in breast cancer mortality has not been demonstrated with MRI screening in high-risk women, the higher detection rate of cancerparticularly invasive cancerwith MRI, coupled with the earlier stage at detection led the American Cancer Society to conclude that sufficient evidence exists to recommend annual MRI screening for women proven to be mutation carriers, untested first-degree relatives of mutation carriers, and women with a lifetime risk of breast cancer development of 20% or greater as determined by models based on a family history of breast cancer.
Screening Other High-Risk Women
Very little information is available to assess the benefits of MRI screening in women at risk for reasons other than family history. Port and coauthors performed a retrospective study of 252 women with lobular carcinoma in situ (LCIS) and 126 with atypical hyperplasia followed with and without MRI. Patients selected for MRI screening were younger and had stronger family histories of breast cancer than their counterparts screened with mammography. Cancer was identified in 1% of the 478 MRIs performed, and 25% of the patients undergoing MRI received a biopsy recommendation during the study period, the majority of which were generated by findings seen on MRI alone. In comparison, abnormalities on physical exam or mammogram prompted biopsy in 11% of patients. In addition, 48% of patients undergoing MRI had at least one study requiring short-interval follow-up. A total of 8 cancers were found in the MRI group, 6 by MRI alone. The sensitivity of MRI was 75%, the specificity 92%, and the positive predictive value was 13%.
These data do not provide clear evidence of benefit for MRI screening in the population of women with LCIS and atypical hyperplasia, but they clearly illustrate the costs of screening in terms of follow-up examinations and biopsies. Further evaluation of the impact of recall, short-interval follow-up, and benign biopsies on compliance with screening is warranted before widespread adoption of MRI screening in populations at lower risk than women with known or suspected BRCA1/2 mutations is undertaken.
In a study of a subset of 611 women included in the United Kingdom MRI screening study, 89% reported that they would definitely return for further screening, but 4% found MRI "extremely distressing" and 47% reported intrusive thoughts about the MRI exam 6 weeks later.
The current American Cancer Society recommendations for breast MRI screening are summarized in Table 2. It is important to note that breast MRI is an adjunct to mammography, not a replacement, and that clinical breast exam remains an important part of the surveillance of women at high risk. Cancers seen on mammography but not MRI continue to be observed, and although the problem of interval cancers is reduced by MRI screening, it is not eliminated.