Recent Controversies in Mammography Screening for Breast Cancer

Introduction

The statistic of a 1 in 8 lifetime risk of developing breast cancer is all too familiar to most American women.^[1] We know that women all over the world are at risk of developing breast cancer, yet we know little about how to prevent this disease. While we are engaged in the discovery of ways to prevent breast cancer, much of cancer control today rests on detecting breast cancer at a stage early enough that patients can be completely cured of the disease. This is the basic tenet of cancer screening. The best cancer screening tools detect abnormal cells before they become cancer cells; the abnormal cells can then be removed.

There are several aspects of a cancer screening method that affect its applicability to the general population. The screening test must detect a cancer or precancerous condition when treatment can affect outcome. There must be a low false-negative rate, so that few affected persons will be missed. The rate of false-positives must be low, in order to prevent unnecessary procedures and treatments. The test must produce reproducible results and should be inexpensive. Optimally, randomized, controlled trials testing the screening method to either clinical breast examination or no screening. should show benefit in reduction of cancer mortality. Finally, the screening test must be available for large segments of the population.

Mammography, even at its best, is an imperfect cancer screening test. The most optimistic estimates from observational data indicate a 63% reduction in death from breast cancer in women who are regularly screened with mammograms every 1-2 years.^[2] Screening trial data suggest a more modest but significant reduction, with the results of individual studies ranging from no effect to a 45% reduction in breast cancer deaths.^[3] Mammography is imperfect because it does not reliably detect lesions before they are cancerous. Mammography does pick up in situ cancers, and therefore the proportion of such early-stage cancers has increased dramatically since the beginning of clinical use of mammography in the early 1980s ^[4]; however, it often does not find tumors until they are Stage 1 or later. Although Stage 1 cancers have an excellent prognosis compared with later-stage breast cancer, many oncologists believe that cancer cells have already spread to the bloodstream or lymphatic system by the time a woman is diagnosed -- the so-called micrometastases. Crude as it is, however, mammography is all that we have at present to detect breast cancer at the earliest possible stage.

On the basis of findings and conclusions drawn from several screening clinical trials conducted in the 1980s, most clinicians and researchers: agree that for women older than age 50 years, mammogram screening every 1-2 years reduces breast cancer deaths by 30%. For women younger than 50, the efficacy of mammography in reducing cancer deaths has in past years been the subject of considerable debate, in part because of greater difficulty of detecting lesions in the more dense breasts of younger women. Nevertheless, most clinical groups and cancer advocacy groups in the United States promote regular mammography beginning at age 40, or earlier if there is a family history of breast cancer developing at a young age.^[5]

A recent Working Group was convened by the International Agency for Research on Cancer (IARC) of the World Health Organization. It concluded that trials have provided sufficient evidence for the efficacy of mammographic screenings of women aged 50-69 years (35% reduction in breast cancer mortality) but that for women aged 40-49 years, there is only limited evidence for a reduction in breast cancer mortality.

A recent review of the 8 mammography screening trials conducted in Canada, Europe, and the United States has questioned the usefulness of mammography for preventing breast cancer deaths.^[3,6] Although Olsen and Gotzsche described all of the trials as flawed, they deemed 2 as being "adequately randomized" and found no reduction in risk of breast cancer mortality in these 2 trials (pooled related risk 1.04, 95% confidence interval 0.84-1.27).^[3]

Their decision to dismiss all of the other trials as being biased was, in my opinion, poorly justified. For example, they described an age imbalance in 2 trials as indicative of inadequate randomization. The age differences amounted to an average of 3-6 months older age for women randomized to mammography vs control. The trials were then adjusted for age in the analysis. The authors claim that this small age difference invalidates the trial results and that age adjustment does not correct the problem. This is flawed thinking on the part of Olsen and Gotzsche. If anything, the difference in ages should have resulted in more cases in the mammography arm because breast cancer incidence increases with age. The finding of reduced breast cancer mortality in the screening arms (24% reduction in one study, 42% reduction in the other) strongly supports the effectiveness of mammographic screening.

One thing that should be kept in mind when reviewing the results of screening trials is that the results are what one would see in women offered screening, not in those who are actually screened. Not all women offered mammography actually had regular mammograms during the trials. Moreover, some women randomized to the control group received mammograms. The analysis of these trials as intent-to-treat, while the correct technique for methodologic reasons, produces an underestimate of benefit for those women who do undergo yearly mammography.

No large screening trial can be perfect. The challenges of recruiting, randomizing, and following thousands of women almost guarantee that there will be some methodologic flaws. For this reason, scientists usually look at the totality of evidence in deciding on intervention effectiveness. In the case of mammographic screening, the totality of evidence points to a benefit of screening.

There are other potential benefits of mammographic screenings that have not yet been capitalized upon. Although the amount of dense parenchymal tissue can affect the validity of cancer screening, it is also a marker of breast cancer risk^[7]; in a nested case-control study in a large health-maintenance organization, women with extremely dense breast tissue were 6 times more likely to have an interval-detected breast cancer than were women with extremely fatty breast tissue.^[8] The sensitivity of mammography was 30% in the women with dense breasts and 80% in women with fatty breast tissue. That breast density is a marker of risk of developing breast cancer may warrant advising women with dense breasts about ways to reduce their breast density, although more research in this area is necessary.

The controversy over the effectiveness of mammography highlights the urgent need for developing better methods of screening for breast cancer. Some new modalities under evaluation are digital mammography, magnetic resonance imaging, ultrasound, thermal detection monitors, computed tomography scans, and proteomics. Testing the effectiveness of these as screening modalities should be a high research priority.

Clinicians must be able to give confident recommendations to their patients, and public health experts must be able to make more global recommendations. The recent US Senate hearings on mammography underscore the frustrations of the public and of clinicians who are attempting to understand the scientific controversy surrounding mammography and who have to make decisions based on their understanding. Simply telling women the risks and benefits of mammography and expecting them to make their own decisions based on statistics does not approach the issue. The preponderance of evidence indicates that mammography saves lives. Until we have improved breast cancer screening methods, this 40-something woman intends to continue getting annual screening mammograms.

© 2002  Medscape

Cite this: Recent Controversies in Mammography Screening for Breast Cancer - Medscape - Apr 22, 2002.