| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorials |
1 From the Department of Radiology, Indiana University School of Medicine, Wishard Memorial Hospital, 1001 W 10th St, Indianapolis, IN 45202. Received April 24, 2002; revision requested May 9; revision received and accepted July 15. Address correspondence to the author.
Index terms: Breast radiography, 00.112 • Breast radiography, quality • Editorials
There has been a great deal of controversy regarding the effectiveness of screening mammography over the years, particularly for women between the ages of 40 and 49 years. The controversies had subsided over the past few years, and screening for breast cancer had become well accepted. However, the calm was upset recently after publication of two articles in The Lancet by a pair of Danish investigators, Gotzsche and Olsen (1,2). They criticized the quality of a majority of randomized controlled trials for screening mammography and concluded that mammography is ineffective in reducing breast cancer mortality. The U.S. press picked up the controversy, producing a number of damaging newspaper, magazine, and television pieces on this topic. How could this situation happen? Are the criticisms valid? What has been done to answer the criticisms?
There are three methods of breast cancer screening: breast self-examination, clinical breast examination, and mammography. The American Cancer Society, or ACS, currently recommends annual mammography for all women beginning at age 40. Since the use of screening mammography became widespread in the 1990s, there has been a 1.6%–3.4% reduction in breast cancer mortality per year in the United States (3). While some of this mortality reduction has undoubtedly been due to improved treatment of breast cancer, much of the reduction appears to be due to earlier detection of the disease with screening mammography. It is well established that the 5-year survival rates for women with localized breast cancer are far better than those for women with regional disease or distant metastases. Therefore, it seems obvious that the earlier most breast cancers are detected, the better the prognosis and treatment options for the patient. While mammography does not depict all breast cancers (hence the need for clinical breast examination and breast self-examination), it is the only method proven to depict subclinical breast cancers and thus reduce mortality.
Most experts agree that randomized controlled trials are the best way to determine the effectiveness of an intervention. There have been eight major randomized controlled trials regarding mammographic screening for breast cancer (4–12). Most of these trials began decades ago, when mammographic technology and interpretation skills were less advanced than they are today. Nonetheless, a majority of the trials showed a statistically significant decrease in breast cancer mortality for women in the screened groups, and the results have been used to guide screening programs worldwide.
Gotzsche and Olsen (1,2) are part of the Nordic Center for the Cochrane Collaboration, or Cochrane Library, an organization based in Oxford, England, with several centers around the world. Members of the Cochrane Collaboration perform meta-analyses called Cochrane Reviews. The Lancet "is an enthusiastic partner of the Cochrane Collaboration" (13). While the reviews are respected in many medical disciplines, their meta-analyses usually study treatment trials rather than screening trials. Trial design and analysis are often different for these two categories of randomized controlled trials.
In the January 2000 issue of The Lancet, Gotzsche and Olsen (1) published a meta-analysis of eight screening mammographic randomized controlled trials. They claimed randomization imbalances and inconsistencies in six of the trials: the Health Insurance Plan, or HIP, study (4), the Edinburgh trial (5), the Gothenburg trial (6), the Stockholm trial (7), and the "two-county" trials in Sweden at Kopparberg and Ostergotland (8). In fact, the only trials they considered "good" were the Canadian trial (9,10) and the initial report of the Malmo trial (11).
These last two studies (9–11) showed no benefit from screening mammography. In fact, the Canadian trial showed an increase in breast cancer mortality for women in their 40s (9). Gotzsche and Olsen (1) stated that there were unacceptable differences in the ages of the women in the screened and control groups in the "discarded" studies. They criticized the methods of determination of mortality in the six "poor" trials, stating that the appropriate end point should be all-cause mortality rather than death of breast cancer. In addition, they alleged impropriety in determining the cause of mortality in the women in these studies. They also criticized the trials for failing to account for the physical and psychological morbidity associated with false-positive imaging results. Gotzsche and Olsen (1) concluded that mammographic screening for breast cancer is unjustified.
Numerous letters were published in subsequent issues of The Lancet. A common theme of the letters was that publication of such a flawed piece by a respected medical journal had the potential to do great harm. Various authors pointed out that randomization was not a major problem in most trials, particularly those discarded by Gotzsche and Olsen (1). The difference in age distribution for the screening and control groups in the poor trials was only 1–5 months, often with the screening group being slightly older than the control group. One would expect poorer results in the older women, thus resulting in underestimation of the true effect of screening. Surprisingly, the publications of the Canadian (9,10) and Malmo (11) trials did not specify the age distribution of the study and control groups. In addition, there have been numerous reminders of the reports documenting the poor quality of the mammographic technique and interpretation in the Canadian trials (9,10), as well as randomization problems in that study (13–15). There were significantly more women with advanced cancers in the study groups than in the control groups in the Canadian trials (9,10). For some reason, Gotzsche and Olsen (1,2) had completely ignored the later report of the Malmo trial (12), published in 1997, which showed a 26% reduction in breast cancer mortality for women aged 55–69 years, despite problems with high contamination in the control group and low attendance of screening in the study group.
After the flurry of letters to the editor of The Lancet, the topic disappeared temporarily. However, in the October 20, 2001 issue, Olsen and Gotzsche published a research letter titled "Cochrane Review on Screening for Breast Cancer with Mammography" (2). They stated that their Cochrane review findings "confirmed and strengthened our previous findings." Once again, they contended that breast cancer mortality is a misleading outcome measure—the more appropriate end point would be "all-cause" mortality. They also stated that screening leads to more aggressive treatment, which produced increased morbidity and mortality in women in the screened groups of the randomized controlled trials. Many of their statements merely referred back to their original study, without referencing new data.
In the same October 20, 2001 issue of The Lancet (16), Richard Horton, editor of The Lancet, stated that the Gotzsche and Olsen (1) and Olsen and Gotzsche (2) "Cochrane analysis" was clearly at odds with the Cochrane Library. The Cochrane Collaboration had sent a letter to The Lancet stating that while Gotzsche and Olsen’s work was done at the Cochrane Nordic Center, their review was not a Cochrane Collaboration systematic review and had not been reviewed and approved by the Cochrane Breast Cancer Group (17). Horton stated that the Cochrane Breast Cancer group had insisted on changes that appeared in the Cochrane Library review but not in the letter published in that issue of The Lancet (2). He further stated that Cochrane editors had added statements in favor of screening and excluded data about the effects of screening on subsequent treatment. However, I found that such changes were not readily apparent in the version of the Cochrane review published on the Cochrane Collaboration Web site (18). Somewhat surprisingly, Horton (16) concluded his editorial by stating that screening mammography has no value.
The information in the October 2001 Lancet articles (2,16) was subsequently picked up by the lay press, leading to a number of well-publicized articles. The most damaging of these were published in the New York Times. Headlines such as "Analysis: Mammograms Don’t Cut Cancer Death Rate," "Breast X-Ray Doubts Raised," "Are Mammograms Worth It," "Why Place Women’s Health in the Hands of Self-interested Professionals," and "Circling the Mammography Wagons" were seen on the front pages of newspapers and magazines almost daily in January and February of 2002. A number of prominent radiologists and epidemiologists responded or attempted to respond in favor of mammography. Unfortunately, the degree of controversy from all sides was of nearly unprecedented proportions, damaging the credibility of many involved on either side. The American College of Radiology, or ACR, issued an official opinion, stating that "The ACR strongly disputes the results of a Danish study on the efficacy of mammograms in reducing cancer death risk that were published in The Lancet. This article, and the accompanying commentary by Editor Richard Horton, spotlights a study based on flawed data." Subsequently, the ACR made the following statement on its Web site:
The ACR is a strong supporter of evidence-based guidelines and the role of science in validating medical interventions. The ACR has supported and will continue to advocate screening mammography based on accepted and rigorous scientific scrutiny. If the clearly flawed analysis by Gotzsche and Olsen, and the unfortunate press coverage it has received in recent days, deters women from the clear benefits of screening mammography, it would constitute a tragic reversal of the obvious benefits we are now beginning to realize.
In January and February 2002, the ACR and the Society of Breast Imaging, or SBI, issued press statements noting the value of mammographic screening. These statements, as well as the many letters to editors by those in favor of screening mammography, fell on deaf ears. In many cases, newspaper editors merely refused to publish letters that pointed out the value of screening mammography. In other cases, surprising statements were made regarding the self interest of the "cancer establishment" or "cancer industry" (19), as though the many health care professionals who have devoted their careers to preventing, diagnosing, and treating cancer have done it only for money. I suspect that most of my colleagues feel the way I do—nothing would please me more than to be required to find a new subspecialty because cancer had been eradicated.
In early February 2002, the tide finally began to turn with the publication of an article by Miettinen et al (20) in the February 9th issue of The Lancet. They reviewed the randomized controlled trials used by Gotzsche and Olsen and rejected the Canadian trial (9,10) because of its well-known problems. For unspecified reasons, they agreed to reject the other six randomized control trials rejected by Gotzsche and Olsen (4–8) and accepted old Malmo trial data (11) for further analysis. They strongly stated that breast cancer mortality rate, rather than all-cause mortality rate, is the correct end point. They also pointed out that population benefits for processes such as breast cancer take a long time to become evident. This is apparent in the data from the published randomized controlled trials—the longer patients are followed up after study entry, the more pronounced the effect of mammographic screening on breast cancer mortality.
There has been serious concern that U.S. health care policy might change on the basis of the Gotzsche and Olsen analysis (1,2) and the ensuing charges by the lay press. In fact, the National Cancer Institute, or NCI, screening and prevention editorial board of the Physician Data Query, or PDQ, agreed that seven large studies of mammography had serious flaws, which weakened or cast doubt on the studies’ validity. The PDQ writes information for the NCI’s on-line database and therefore has the potential to change its screening recommendations, which are referenced by many health care providers. Despite the PDQ’s position, Peter Greenwald, MD, NCI cancer prevention chief, indicated that women should continue to undergo screening for breast cancer after age 40. In further support of breast cancer screening, on February 21, 2002, Secretary of Health and Human Services Tommy G. Thompson issued new guidelines strongly recommending mammographic examinations beginning at age 40. In addition, on February 28, 2002, Senators Mikulski and Harken held congressional hearings, at which mammography received enthusiastic support.
Are there scientific data to counter the attacks of investigators such as Gotzsche and Olsen? Yes. Six of the currently published randomized control trials show significant breast cancer mortality reduction for women who underwent screening with mammography (4–8,12). Large trials are difficult to develop and implement, and there are flaws, albeit minor, in all of the trials. Should a new, more "perfect" trial be undertaken? Such an endeavor would be extremely costly, require many years to perform, require huge numbers of patients, and probably be considered unethical. Does anyone actually believe that women who would be randomized to the control group could be denied access to mammography? The "contamination" of the control group would be unacceptably large. Therefore, we will have to accept the currently published trials as the best evidence we have for the effectiveness of a test that, while not perfect, is the best weapon we have in the war against breast cancer.
What about other imaging modalities currently in use or under investigation for the detection of breast cancer? Full-field digital mammography, or FFDM, is currently under investigation, and two manufacturers now have U.S. Food and Drug Administration approval for the sale of their equipment. FFDM is undoubtedly a major technologic advance, but studies to date (21) have not shown significant improvement in screening accuracy. In addition, the cost of such systems will limit their use in the United States in the near future.
Ultrasonography (US) has been used for years in breast imaging centers as an adjunctive problem-solving examination. Early studies of US for breast cancer screening showed high false-positive rates and extremely low detection rates for nonpalpable mammographically occult malignancies. However, in the past decade, there have been marked advances in US equipment, and recent studies have shown increased breast cancer detection rates for US, as compared with studies performed in the 1970s and 1980s (22–24). However, screening with US is not the current standard of care (25), and further research will be necessary to determine the false-positive and false-negative rates by using state-of-the-art equipment, as well as the cost of US screening, including the cost of biopsy for the large number of false-positive findings that will undoubtedly occur.
Magnetic resonance (MR) imaging has been under investigation for use in the breast for a number of years. It has been shown to have high sensitivity for breast cancer but lower specificity. MR imaging is very expensive and requires contrast material injection for the detection of breast cancer. Despite improved technology and protocol refinement in recent years, MR imaging remains limited to specific problem-solving situations and patients at high risk for cancer. Recent studies (26) have shown improved detection of nonpalpable mammographically occult cancers, but biopsy of MR imaging–detected lesions remains a problem. Thus, while MR imaging may have a screening role in the future, it is certainly not the standard of care today.
The Institute of Medicine, or IOM, evaluated breast cancer screening and diagnosis in 2001 (27). The panel concluded that screen-film mammography remains the reference standard, despite its limitations. Recognizing the significant access problems for mammography in the United States today, the IOM recommended that the Health Care Financing Administration (now the Centers for Medicare and Medicaid) analyze current reimbursement rates to ensure that they cover the total cost of performing the procedure and to examine the factors that affect radiologists’ and technologists’ decisions to enter or remain in the field.
Some may argue that reexamination and discussion of the value of screening mammography is healthy; however, considerable damage has been done, since women and their health care providers are confused with regard to the need for mammographic screening. In addition, this issue has consumed major resources of the ACR, ACS, and many practicing radiologists in the past several months. The ACR, for example, cannot even begin to estimate the cost of staff and volunteer radiologist time in this effort (Zinninger M, oral and written communications, 2002). The time and financial resources could have been much better spent attacking the real problems associated with breast imaging: limited access to mammography in many parts of the country, manpower shortages, and insufficient reimbursement for mammography.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  C. I. Henschke Reply to the Letters to the Editor from Bach and Silvestri Clin. Cancer Res., April 15, 2008; 14(8): 2511 - 2511. [Full Text] [PDF]
|
|
|
|
 |
|
 D. F. Yankelevitz, W. C. Black, D. R. Aberle, C. D. Berg, and on behalf of the Executive Committee of the Nation Will the National Lung Screening Trial Be Able to Demonstrate a Mortality Reduction? Radiology, February 1, 2008; 246(2): 653 - 654. [Full Text] [PDF]
|
|
|
|
 |
|
 D. F. Yankelevitz Screening for a Cancer: Acting on Social Responsibility Am. J. Roentgenol., May 1, 2007; 188(5): 1171 - 1172. [Full Text] [PDF]
|
|
|
|
 |
|
 C. I. Henschke Medicine on Lung Cancer Screening: A Different Paradigm Am. J. Respir. Crit. Care Med., November 15, 2003; 168(10): 1143 - 1144. [Full Text] [PDF]
|
|
|
|
|
|
D. B. Kopans, B. Monsees, and S. A. Feig Screening for "Cancer": When is it Valid?--Lessons from the Mammography Experience Radiology, November 1, 2003; 229(2): 319 - 327. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. F. Yankelevitz, S. J. Swensen, J. R. Jett, T. E. Hartman, D. E. Midthun, J. A. Sloan, A.-M. Sykes, G. L. Aughenbaugh, and M. A. Clemens Randomized Controlled Trials and Cancer Screening [letter] * Dr Swensen and colleagues respond: Radiology, November 1, 2003; 229(2): 603 - 604. [Full Text] [PDF]
|
|
|
|
|
|
E. Anorbe, P. Aisa, and V. P. Jackson Screening Mammography [letter] * Dr Jackson responds: Radiology, June 1, 2003; 227(3): 903 - 904. [Full Text] [PDF]
|
|
|
|
|
|
L. Berlin Breast Cancer, Mammography, and Malpractice Litigation: The Controversies Continue Am. J. Roentgenol., May 1, 2003; 180(5): 1229 - 1237. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
