8
Recommendations
The committee’s recommendations for strategies to reduce the toll of breast cancer fall into four categories: improve current technology and its application; develop new screening strategies that integrate biology, technology, and risk models; ensure that promising innovative technologies are adequately tested; and improve the implementation and use of new technologies. The committee expects that many of the new detection technologies currently being tested in clinical studies, as well as those in earlier stages of development, will eventually lead to an understanding of the biology and mechanisms of breast cancer and, therefore, will improve all aspects of the continuum from diagnosis to management.
One strategy that should be mentioned, but that is not included in the committee recommendations, would be to increase the use of screening mammography. The committee believes the importance of this should not be overlooked, but the focus of this report is on ways to improve technologies and systems for early detection, rather than ways to improve the utilization of those services. The two are clearly linked, but tackling the problems of service utilization would require another study.
The committee recognizes that they have set a broad and ambitious agenda—one that will require support and cooperation from a spectrum of participants, from Congress and federal agencies and regulatory agencies to physician organizations, the research community, and health care payers and providers. Perhaps most essential will be support from breast cancer and women’s health advocacy groups and from women themselves. Without the unwavering support of this vital community, little progress will be
possible. Their input must provide the impetus for change, as it has so effectively in the past.
Even with this broad base support, this effort demands strong leadership and coordination. Therefore, for practical reasons, lead responsibility for implementing many of the recommendations is assigned to the National Cancer Institute and to the relevant professional groups, who should, in turn, enlist other groups most able and qualified to assist. Where additional funding or policy changes are required, that responsibility is also designated.
A. IMPROVE CURRENT APPLICATION OF SCREENING MAMMOGRAPHY
Evidence from randomized clinical trials and from community breast cancer screening programs documents the ability of mammographic screening to reduce mortality from breast cancer. However, mammography is not a perfect technology, nor is it always applied perfectly. Wide variations in the quality of mammographic services need to be addressed, as does the serious and growing shortage of qualified mammographers.
The committee identified several key challenges to providing high-quality breast screening services to all women who would benefit: organizing breast screening services to increase their quality and efficiency (Recommendation A1), improving the overall quality of mammography interpretation and encouraging the development and dissemination of adjunct technologies that would further improve mammography (Recommendation A2), and conserving the workforce of breast imagers and support personnel and making optimal use of their skills (Recommendation A3).
A1. Health care providers and payers should consider adopting elements of successful breast cancer screening programs from other countries. Such programs involve centralized expert interpretation in regionalized programs, outcome analysis, and benchmarking.
International differences in breast cancer detection patterns and mortality are influenced by the organization of breast cancer screening programs. Comparative studies of screening programs indicate that programs with high rates of abnormal mammograms tend to have low positive predictive value for biopsies. Although these studies cannot determine the underlying causes of this trend, they highlight several characteristics of successful breast cancer screening programs in other countries that are not fully realized in the United States.
Aspects of foreign screening programs can provide models to guide the improvement of domestic programs, including the incorporation of quality
assurance measures, the integration of breast cancer screening services with treatment and support, and the organization of regionalized services in which mammograms are read by breast imaging specialists at a central location. In The Netherlands and Sweden, for example, breast cancer screening is performed at outlying facilities, while diagnosis and workup of mammograms occurs at a few dedicated centers. These countries have low rates of false-positive mammograms.
Experience with the Mammography Quality Standards Act (MQSA), which led to nationwide improvements in the technical quality of mammography, demonstrates that a national quality assurance program could be successful in the United States. Evidence indicates that high callback rates cannot be reduced by targeting individual performance, but must be addressed through the establishment of overall standards for program performance and outcomes.
A2. Breast imagers and technology developers should work in collaboration with health care providers and payers to improve the overall quality of mammographic interpretation by:
-
Adopting and further developing practices that promote self-improvement of breast imagers, but that do not jeopardize the workforce.
-
Developing technologies, such as computer-aided detection (CAD), that have the potential to improve quality, and expanding their use once they have been validated.
The key to improving mammographic interpretation is to reduce known and controllable sources of variability in quality without adding to the burden of an already overextended workforce. Although better training will significantly improve radiologists’ performance, merely increasing the amount of mandated training for breast imagers is not likely to improve mammography services. Continuing medical education is required for radiologists in the United States, but course content is not uniformly organized and few programs target recall or cancer detection rates. In the United Kingdom, however, a voluntary self-assessment program is used by more than 90 percent of radiologists who practice mammography.
Benchmarking could include setting ranges for callback rates, sensitivity measures, or predictive values for biopsies and would be adjusted for factors such as case variation that are known to influence such measures.
CAD could also be employed to provide a second reading of mammograms, following interpretation by an experienced radiologist. Although CAD is not necessarily the equivalent of double-reading by two radiolo-
gists, it can highlight areas of concern in a mammogram for further interpretation. The addition of CAD is unlikely to improve the accuracy of all breast imagers, but it has the potential to raise the performance level of general radiologists to that of breast imaging specialists.
New approaches for quality improvement should be developed in collaboration with the breast imaging community and with experts in human performance, including performance measurement. No technology has proven superior to mammography in its combination of sensitivity and specificity, nor—most importantly—in its ability to reduce breast cancer deaths. Because mammography will continue to play a central role in breast cancer detection for the foreseeable future:
A3. Mammography facilities should enlist specially trained nonphysician personnel to prescreen mammograms for abnormalities or double-read mammomograms to expand the capacity of breast imaging specialists.
The supply of radiologists (and therefore of breast imagers) is unlikely to grow as quickly as demand for their expertise, and many in the breast imaging field contend that the availability of mammography services is undergoing stagnant growth, if not decline. Significant barriers hamper the expansion of training programs for breast imaging and the immigration of well-trained radiologists to the United States.
The judicious use of physician extenders could raise the productivity of the limited number of radiologists who interpret screening mammograms. Evidence suggests that radiological technologists (RTs) can be specially trained to prescreen mammograms for abnormalities1 or double-read mammograms along with a radiologist. The committee does not suggest that RTs should interpret diagnostic mammograms or that screening mammograms should be interpreted solely by an RT; rather, the RT would expand the capacity of radiologists. Challenges to this proposal include the acceptance of the radiology profession and malpractice coverage.
The MQSA stipulates that mammograms are to be interpreted only by a physician specifically certified in mammography. The Act does not, however, preclude other personnel from examining the mammograms that are also interpreted by certified physicians. Although not widely appreciated and rarely practiced, it would in fact be permissible within the provisions of the MQSA to have nonphysician personnel examine mammograms—as long as a certified physician signed the mammogram report indicating that he or she had interpreted it. This suggestion that physician extenders could be enlisted to help read mammograms could thus offer women a more thorough examination than is currently typical of most mammography facilities where mammograms are viewed only by a single breast imager.
Physician extenders could potentially improve the overall accuracy of mammographic interpretation through double-reading, as well as alleviate the burden on the breast imaging physicians by prescreening the mammograms to allow the interpreting physician to spend more time on the more problematic mammograms.
B. INTEGRATE BIOLOGY, TECHNOLOGY, AND RISK MODELS TO DEVELOP NEW SCREENING STRATEGIES FOR BREAST CANCER
The degree of risk for breast cancer varies widely among women. A variety of breast cancer risk factors have been identified, and they continue to be discovered, but it is still not possible to predict who will develop lethal breast cancer, and who will not. This imperfect knowledge informs a spectrum of important medical decisions along the pathway from breast cancer detection to treatment, in some cases involving extreme preventive measures such as bilateral mastectomy.
B1. Researchers and technology developers should focus their efforts on developing tools to identify those women who would benefit most from breast cancer screening. Such tools should be based on individually tailored risk prediction techniques that integrate biologic and other risk factors.
The awareness that women do not have uniform risk for breast cancer suggests the possibility of identifying women who are most likely to benefit from more intensive screening for breast cancer, as well as those who could safely be screened less frequently. Risk-based screening strategies, the committee believes, are essential to improving the early detection of breast cancer. Developing such strategies will require well-designed, large-scale epidemiological studies to gain a better understanding of risk assessment in individuals.
Mammography screening guidelines already take into account two of the most significant breast cancer risk factors: gender and age. A far more comprehensive approach could be obtained through an integration of epidemiologic factors (such as those identified in the widely used Gail model of breast cancer risk prediction), genetic risk factors (of which there is only rudimentary understanding), and the consequences of adverse events such as false-positive and -negative findings. However, until sufficient knowledge and evidence accumulates to enable the individual assessment of breast cancer risk, the committee urges adherence to consensus guidelines for the minimum recommended use of mammography screening.
In addition to research needed to refine and expand knowledge of risk factors for breast cancer, mathematical models must be developed that can reliably integrate the spectrum of risk factors and predict their collective influence. Ultimately, mathematical models will relate genetic predictors, biological expression, natural course of disease, and responses to treatment. Such models will permit researchers to:
-
Elucidate the natural course of disease progression
-
Identify disease subgroups with distinctive risk profiles and treatment susceptibilities.
-
Identify aspects of the models where further research and data collection are needed.
-
Provide guidance to technology developers as to what types of technologies will be most useful, including the performance characteristics that are required for them to be useful, and the analytic techniques that would be most appropriate for evaluation.
B2. Technology innovators, including basic scientists, should work with clinicians, health systems experts, and epidemiologists from the earliest stages of development in order to increase the likelihood of creating clinically useful tools for the early detection of breast cancer.
Fulfilling the immense potential of molecular medicine for breast and other cancers will require collaboration between molecular biologists and scientists from a broad spectrum of disciplines. It will fall to epidemiologists and biostatisticians to guide the rational design of biologically based cancer diagnostics, to establish their significance and reproducibility, and, in the case of clinical epidemiologists, to adapt them for routine clinical use.2 Once these new biologically based detection and diagnostic tools have been developed, they must be tested for safety and effectiveness beyond the research setting in multicenter clinical trials. Finally, these tools will not be used in isolation, but will become part of an arsenal of tools—each with distinctive capacities and caveats. Developing evidence-based systems for integrating this new technology will require attention at all levels of our health care system—physicians, payers, and purchasers (patients).
The research engine that drives technology advances is well fueled, but the validation of those advances is another matter. Although basic research enables the development of early stage technologies, different strategies are needed to identify which technologies are truly feasible and add clinical value by improving health or the delivery of health care services. Large-
scale well-designed multicenter clinical trials provide the most definitive answers about the clinical value of new technologies.
The theme of reengineering the clinical research enterprise is particularly relevant to what the committee believes is especially needed to promote the development of more effective approaches to the early detection of breast cancer. This theme is further subdivided into three initiatives—translational research, clinical workforce training, and enhancement of clinical research networks—all of which address the committee’s conclusion that basic research should be integrated with technology development and assessment.
Scientists have become increasingly aware that the bench-to-bedside approach to translational research is really a two-way street. Not only do basic scientists provide clinicians with new tools to examine patients, but clinical researchers also make novel observations about the nature and progression of disease that can stimulate basic investigations.
There is no argument that the development of new tools and strategies for breast cancer detection is a multidisciplinary endeavor. But the truth is that research tends to be herded into disciplinary silos, not because researchers do not appreciate the value of multidisciplinary work, but because the reward systems favor this tendency in so many ways. The specialized languages and standards of different research traditions, the composition of grant review committees, and the organization of promotion and tenure within research institutions all promote disciplinary specialization. The result is that few molecular biologists—those who might be developing genomic profiles of breast cancer—understand the methodology necessary to test the validity of a new screening test. Conversely, few epidemiologists—those who might evaluate breast cancer risk factors—appreciate the uncertainties inherent in gene expression analysis. Likewise, few of the physicists or engineers who might develop advanced imaging technology understand the extent and design of clinical studies that are needed to test the technologies, or in some cases, what types of technologies would be most useful in breast cancer detection.
At every disciplinary juncture, there are pioneers who cross the divide. Nearly every report like this one calls for more multidisciplinary research and development. But this is easier said than done. This committee, like so many others before it, was impressed by how often early stage developers fail to engage the appropriate range of expertise in their endeavors.
B3. Research funders, including the National Cancer Institute and private foundations, should develop tools that facilitate communication regarding breast cancer risk to the public and to health care providers.
The likelihood that a woman will adhere to screening recommendations depends, in part, on her perceived risk of developing breast cancer. Unfortunately, women’s perception of their risk of getting breast cancer or dying from it is often distorted. Most women also misunderstand or overestimate the benefits of mammography. Thus, for example, participation in screening programs tends to decline with age, despite women’s rising risk of developing breast cancer. Women who develop breast cancer often have distorted perceptions about prognosis; for example, they may not understand that a diagnosis of ductal carcinoma in situ is far less grave than a diagnosis of invasive breast cancer.
If women are to make well-informed decisions regarding breast cancer detection, diagnosis, and treatment, they need a firm understanding of the risks they face. Not only are better methods needed to assess a woman’s risk of breast cancer, but more effective means are needed to communicate those risks.
C. IMPROVE THE ENVIRONMENT FOR RESEARCH AND DEVELOPMENT OF NEW TECHNOLOGIES FOR BREAST CANCER DETECTION
C1. The National Institutes of Health, Agency for Healthcare Research and Quality, and Centers for Medicaid and Medicare Services should collaborate to establish programs and centers (which may be virtual) that bring together expertise and funding to enable a more comprehensive approach to technology assessment and adoption.
-
These efforts should involve collaboration with technology developers, not-for-profit organizations (including professional societies), advocacy groups, private health care payers, and provider organizations.
-
Experimentation with innovative organizational structures for the centers should be encouraged.
-
Adoption of standards for collecting and sharing data should be a priority.
The National Institutes of Health is exploring development of regional translational research centers. These centers would provide sophisticated advice and resources to better enable scientists to master the many steps involved in bringing a new product from the bench to clinical use.
Despite the promise of technological advances that might significantly
impact breast cancer mortality, the committee is concerned that limited resources and outdated infrastructure will increasingly limit the pace of progress. Thus, to ensure that the promise of these advances is fulfilled, the committee believes that both research and the clinical infrastructure must be adapted in order to overcome existing barriers to research and the adoption of effective technologies.
C2. Professional societies should work together with women’s health organizations to identify barriers to participation in studies (especially those that require provision of biologic specimens) and ways in which those barriers might be overcome.
-
A public education campaign should be undertaken to inform the public, particularly underrepresented groups, of the merit of participation in research studies that require the involvement of healthy volunteers and the donation of biologic specimens for genetic analysis.
-
Advocacy groups and women’s health organizations should participate in design and execution of public education about clinical trials. This could be a collaborative effort, and might include the National Cancer Institute and the American Cancer Society.
-
The Department of Health and Human Services should join with private entities in monitoring the effect of the Health Insurance Portability and Accountability Act Privacy Rule on the pace of research progress.
Of particular concern are the barriers to public participation in clinical trials that have been raised as the unintended consequences of privacy concerns and other initiatives. Because the development of better and evidence-based methods for the early detection of breast cancer will require large-scale clinical trials and those trials depend on public participation, the committee recommends seeking ways to overcome barriers to public participation. The same barriers threaten to impede essential epidemiological research because the identification of markers depends on the availability of blood and other biologic specimens from healthy volunteers.
Another roadblock to improving the current situation is the tendency of many women to either over- or underestimate their own breast cancer risk, which can affect their decision about whether to participate in a screening program.
D. IMPROVE THE IMPLEMENTATION AND USE OF NEW TECHNOLOGIES
D1. Breast cancer research funders, such as the National Institutes of Health, Department of Defense, and private foundations, should support research on screening and detection technologies that encompasses each aspect of technology adoption from deployment to application, and should include monitoring of use in practice.
-
This will involve identification of optimal combinations and sequencing of breast cancer detection technologies.
-
Research funders and private foundations should model and assess changes in practice and organization change that would optimize the benefit of new technology (including risk assessment).
D2. The National Institutes of Health, the Agency for Healthcare Research and Quality, and other public and private research sponsors should collaborate with health systems, providers, and payers to support research that would monitor clinical use of technologies to identify potential failures, as well as opportunities for improvement, with particular attention to:
-
How appropriately the technologies are being utilized,
-
Their impact on clinical decision making, and
-
Their impact on health outcomes.
REFERENCES
1. Casey B. 2003, March 11. Breast Center Enlists Radiographers for First Look at Mammograms. Accessed February 19, 2004. Web Page. Available at: http://www.auntminnie.com/default.asp?Sec=sup&Sub=wom&Pag=dis&ItemId=57614&stm=radiographers.
2. Ransohoff DF. 2002. Challenges and opportunities in evaluating diagnostic tests. J Clin Epidemiol 55(12):1178-1182.