Improving Mammography Quality, Expanding Screening Research
In passing the Mammography Quality Standards Act (MQSA) of 1992, Congress mandated efforts by the medical practice, research, and regulatory communities to improve the performance, quality assurance, and oversight of screening mammography in the United States. The Act authorized the Secretary of Health and Human Services to fund research establishing a breast cancer surveillance system that could assess more extensively mammography performance in clinical practice. Dr. Rachel Ballard-Barbash, associate director of DCCPS’ Applied Research Program, explains that “NCI was assigned the mandate for supporting this research and in response to the Act established the Breast Cancer Surveillance Consortium.”
Founded in 1994, BCSC originally consisted of independent centers studying the practice of breast cancer screening in their individual communities. However, it proved difficult to draw conclusive results from comparisons of similar but heterogenous data. NCI realized the potential of establishing an ongoing centralized database on women undergoing mammography, and increased the standardization of data collection and created a central pooled data resource from all of the centers. BCSC currently consists of five main research sites, two affiliated sites, and a statistical coordinating center located in Seattle, Washington.
“Because of the effort to create a pooled central research data resource, we now have data on over 5.5 million mammograms, representing over 2 million women. More than 52,000 cases of breast cancer have been diagnosed [at participating sites] over the 10 years that BCSC has been in operation,” says Dr. Ballard-Barbash.
This large, standardized dataset presents a unique opportunity for investigators throughout the country to study how mammography screening performance may be improved and how breast cancer screening relates to changes in disease stage at diagnosis, survival, and mortality. In addition, investigators have used BCSC to study disparities in screening and risk factors for breast cancer. It has also been used as a resource for new investigator-initiated studies. Researchers from any organization can apply to use BCSC data for their projects.
“This work produced to date includes more than 235 peer-reviewed publications on a variety of issues, including factors that affect the quality of mammography interpretation and breast characteristics, like density, that affect the likelihood of both cancer occurrence and detection,” says Dr. Stephen Taplin, project director of BCSC.
That breast density is a risk factor for breast cancer has become more widely known over the past several years, and this knowledge has been incorporated into several models for estimating an individual woman’s risk. A new study using BCSC data published in the March 7 Journal of the National Cancer Institute now adds an important piece of information on the use of breast density in calculating risk - that two or more measurements of density over time may be better at predicting risk than a single measurement.
The investigators, led by Dr. Karla Kerlikowske from the University of California, San Francisco, used prospectively collected data from 301,955 women aged 30 or older, who were not taking hormone-replacement therapy and had undergone at least two mammograms at a BCSC center. Breast density was scored on a scale of 1 to 4, in order of increasing density, by the American College of Radiology Breast Imaging Reporting and Data Systems (BI-RADS) criteria.
By linking BCSC records to cancer registry data, a capability built into BCSC data collection methods, the investigators identified 2,639 incidences of breast cancer in the group. Women diagnosed with breast cancer were more likely to have received a breast density score of 3 or 4 on their first and last mammogram than women without cancer. Overall, a high breast density score assigned on the first or last mammogram was associated with an increased likelihood of breast cancer.
Importantly, the rate of breast cancer diagnosis increased for women whose breast density score increased from first to last mammogram, and conversely decreased for most women whose breast density score decreased from first to last mammogram, with the exception that risk remained high for women with a breast density score of 4 on their first mammogram.
These results show “that breast density is a very important risk factor for breast cancer, and that because density can change over time, one measure at one point in time may not accurately reflect how breast density affects a woman’s risk of breast cancer,” explains Dr. Kerlikowske. “If you have two measurements over time, those two measurements together are more likely to give you a better idea of how density increases your risk.”
Dr. Kerlikowske and colleagues are now working on identifying the best time to measure a woman’s breast density for use in a risk model. This work and other BCSC collaborations, including projects in partnership with NCI’s Cancer Intervention and Surveillance Modeling Network and the American Cancer Society, will further leverage BCSC data to answer pressing questions about how best to use breast screening to help predict risk and reduce mortality, and how to improve mammography practices nationwide.