| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, Massachusetts General Hospital, Avon Comprehensive Breast Evaluation Center, Wang Building, Suite 240, 15 Parkman Street, Boston, MA 02114
The article by Dr Jiang and colleagues (1), published in the May 2007 issue of Radiology, is a detailed and comprehensive evaluation of what it might take to prove that a new technology finds one additional breast cancer per 1000 women. Superficially their conclusions seem discouraging, but those of us who have been in the field long enough remember that similar issues were confronted in the mammography screening trials, and they were performed successfully. More than 400 000 women participated in the trials, which proved that screening could decrease deaths from breast cancer. As a result, because of widespread screening, the death rate from breast cancer, which had been unchanged for the preceding 40 years, has decreased by 25% since 1990 and tens of thousands of lives have been saved. Several points need to be considered.
1. The authors used detection rates and reader variability for mammography screening. A new modality may not have as wide a variation. I suspect that, for example, once magnetic resonance (MR) imaging is standardized, we will find that readers are better able to see cancers that "light up" than they are able to discern shades of gray on a mammogram.
2. There are almost certainly many times the number of breast cancers "below the surface" (2) than are presently detected each year. A new modality may be able to depict these earlier cancers making it less difficult to show a benefit.
The use of surrogate endpoints to avoid large trials (eg, finding cancers at a smaller size means benefit) has been proposed since the 1970s, but, thus far, these have not been used successfully (witness the present debates about lung cancer screening). Randomized controlled trials are the only way to prove a benefit from a screening test (unless everyone is "cured"). If we do not perform the trials to prove a benefit then one of the following will happen:
1. We will follow the "slippery slope" in which the test (eg, MR imaging) will be introduced for more and more women at increasing cost (far greater than the cost of trials), and we will still have no idea whether women are benefiting.
2. Women will be denied access to a new test that may, in fact, further reduce deaths from breast cancer.
Early detection using mammography has had a major impact on breast cancer. We need to build on that success and not be deterred by what may seem daunting. Those who have preceded us have faced similar challenges and have prevailed, to the great benefit of women.
 |
References |
|---|
 TOP References References |
|---|
,
Charles E. Metz, PhD *, and
Robert A. Schmidt, MD *
* Department of Radiology, University of Chicago, 5841 South Maryland Avenue, MC2026, Chicago, IL 60637
e-mail: y-jiang{at}uchicago.edu
Group Health Cooperative, Seattle, Wash
We thank Dr Kopans for emphasizing the importance of screening mammography trials. We hope that, one day, a new imaging modality such as MR imaging will reduce interreader variability, but whether that is a realistic hope is presently unknown. We agree that if detection of "below the surface" breast cancer increases the cancer detection rate over the long term, then larger sustained increases in the cancer detection rate will lead to greater statistical power for detecting those increases. However, we assumed a gain of one additional "below the surface" cancer detected per 1000 examinations in our study (1), which is already a very large increment. In our nationally representative study population, the rate of breast cancer detected with any method (including interval cancers detected clinically during the 1-year follow-up period) was 5.09 cancers per 1000 examinations compared with 3.94 cancers per 1000 examinations detected with mammography (1). These data suggest that our assumed gain in cancer detection rate is likely already near the upper limit of what is achievable.
As Dr Kopans emphasizes, in designing imaging trials the challenges that are imposed by interreader variability and the rarity of breast cancer in screening, which we described (1), do not detract from the importance of screening trials. However, our study findings do indicate that underpowered studies can lead to the conclusion that a new technology has no benefit—or indeed may even appear harmful—merely because the statistical significance of its benefit cannot be demonstrated (2). The importance and difficulty of screening trials that we discuss here underscore a need for alternative valid and efficient methods for assessing the clinical efficacy of new technologies. For example, the opportunity in imaging trials of matching patients and readers across the trial arms may be one possibility to conduct trials more efficiently (1,3).
C.E.M. and R.A.S. are shareholders of and receive royalties from Hologic (Bedford, Mass). Y.J. and C.E.M. receive research funding from Hologic.
|
References |
|---|
|
TOPReferencesReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
