| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Avenue, Boston, MA 02215. e-mail: fhall@bidmc.harvard.edu
Editor:
Kudos to Dr Kolb and colleagues (1) who, along with Dr Kaplan (2), have proven to my satisfaction that in a select group of women, screening breast ultrasonography (US) can help identify breast cancer at a relatively early stage, which is comparable to that achieved with screening mammography. Are large randomized controlled trials necessary to document a corresponding survival benefit? The controversial screening mammography trials remind us of how difficult and expensive this might be.
Assuming a benefit from screening breast US, the major determinant of its use is cost, including the downstream costs of false-positive findings. Whatever the benefits, the use of screening US in the foreseeable future will be curtailed severely by manpower limitations. There are simply not enough radiologists and technologists to perform US in even a small percentage of the 60 million women undergoing annual screening mammography in the United States.
Who should undergo screening breast US? Dr Kolb and colleagues (1) did so in 51% of women with category 2–4 mammographic findings (according to the American College of Radiology Breast Imaging Reporting and Data System [BI-RADS] [3]), whereas Kaplan (2) chose only patients with category 3 and 4 findings. There should be less benefit in women with category 2 findings than in those with findings in the more dense categories 3 and 4, because dense breasts have a higher incidence of breast cancer (4,5) and are not as well assessed with mammography. Indeed, Dr Kolb and colleagues (1) found that cancer detection in patients with category 2 findings was greater with the use of mammography and was less with the use of US when compared with detection in patients with category 3 and 4 findings.
Dr Kolb and colleagues (1) rated 49% of their mammographic findings as BI-RADS category 1, defined as fatty tissues. Investigators in other large mammographic series placed only 4% (6) or 6% (7) of mammographic findings into category 1. Clearly, more uniform criteria for categories are needed because they are increasingly used in clinical decision making (1,2,4,5). Objective assessments of breast density will be made more readily available by using digital screening mammography and automated software programs (4).
A 10% false-positive call-back rate is acceptable in screening mammography, although 5% is probably preferable (8). In my experience, false-positive findings are more common with US than with mammography, and the time to perform whole-breast screening is longer than the 4 minutes described by Dr Kolb and colleagues (1). These problems were minimized in their study because of the experience of Dr Kolb, who personally performed all 13,547 US examinations. It is noteworthy, however, that cancer was found in 10.3% (33 of 320) of biopsies performed for US findings but in 44.2% (100 of 228) of biopsies performed for mammographic findings (1). Unfortunately, the authors provide little data on the potentially large number of false-positive US findings that required follow-up or additional imaging but did not require biopsy.
Dr Kolb and colleagues (1) chose to define sensitivity as "the percentage of cancers detected (with a specific modality) among all cancers detected with any modality" (mammography, US, and physical examination). By using this definition, the sensitivity of mammography alone was 78%, and that of mammography and US together was an impressive 97%. However, the denominator of sensitivity is usually defined as the disease that is detected initially or subsequently within some time interval, such as a year. That definition is more difficult to ascertain but is also meaningful.
I assume that the authors did not charge patients for US screening. I am aware of at least one radiology practice in which screening US is performed routinely and insurers are billed for examinations in women with dense parenchyma. Patients, clinicians, and radiologists occasionally make use of screening breast US under the guise of "lumpy breasts" or other pseudohistory. This is particularly understandable in women with a previous palpable cancer not identified with mammography but documented or sampled for biopsy with use of US. Insurers would be extremely reluctant to pay for this examination because of its applicability to such a large number of women.
The media attention generated by the article of Dr Kolb and colleagues (1) prompted the American College of Radiology to issue a statement that US is not a replacement for screening mammography. It should be stressed that these examinations are complementary: Dense tissue is hyperechoic at US, and a hypoechoic invasive cancer is more conspicuous against this background, with the opposite being true for fatty parenchyma. Is it possible that 40–50-year-old women with dense parenchyma, a contentious group with regard to the relative value of screening mammography, might someday merit the use of screening US without mammography? I also wonder if screening breast US will rekindle an interest in automated whole-breast US screening with computer-aided diagnosis systems.
Finally, when yet another type of screening examination is discussed, it might be helpful to do so in the context of the current controversy over body computed tomographic (CT) screening that is being marketed directly to consumers. If insurance companies are unwilling pay for screening breast US, then entrepreneurs will almost certainly offer this examination on a pay-up-front basis. Health officials and providers will again weigh the potential individual benefits of such screening with the finite health care resources of society. Those of us who have difficulty accepting this commercialization of medicine might find it helpful to think of screening in radiology as analogous to the time-honored general physical examination in clinical medicine: CT screening is far more sensitive and expensive than physical examination, but the costs and benefits of each are without scientific proof.
REFERENCES
,
Jacob Lichy, MD,* and
Jeffery H. Newhouse, MD
222 East 68th Street, New York, NY 10021* Department of Radiology, Columbia-Presbyterian Medical Center, New York, NY e-mail: tkolb@panix.com
We appreciate the opportunity to respond to the letters of Drs Kopans, Hall, and Lewin.
Dr Kopans claims that little in the article did not appear in our earlier publication. In fact, there is a considerable amount of new information. Data are included from a much larger group of patients than we described earlier. The patient population we analyzed, unlike the earlier one, consists only of women undergoing screening. New analyses include not only the performance of mammography, physical examination, and screening US in subgroups divided by age, mammographic breast density, and hormonal status, but also the identification of false-negative findings for each modality from findings in the combination of all studies rather than from analysis of interval cancers. New findings include establishing that mammographic breast density, rather than age, is the best predictor of mammographic sensitivity and that US is a much more sensitive test than is physical examination.
Drs Hall and Lewin propose that our accuracy data might be different if different analyses were used. Dr Hall claims that use of the number of interval cancers would permit a more meaningful estimate of the false-positive rate. It is certainly the more traditional technique, but whether it is a more accurate measure of the number of tumors missed at the instant of screening remains to be proved. Dr Lewin feels that comparison of the accuracies of mammography and screening US in a group that had not had undergone mammography in earlier years would be more valid. He is correct, but while this analysis may provide an indication of tumor growth rate, the assumption that the analysis may be a surrogate indicator for lives saved remains to be proven.
Dr Kopans appears to think that we set out to determine what the performance of screening US might be if it were performed without knowledge of other recently performed screening examinations. We did not; it has never been our contention that US might be used as the sole or first examination method for screening. Since we felt that US would likely assist in detection of cancers missed at mammography only in women with dense breasts, it was necessary to perform mammography first to see which patients should undergo US. Our major purpose, however, as we stated clearly in our article, was to evaluate the performance of screening US when it is performed with the information provided by the other examination methods. If screening US ever becomes widely used, this is the way it will most likely be performed, and our data describe what is likely to occur.
Dr Hall wonders whether screening US might ever be used as the first screening method in women with dense breasts (those presumably being examined with mammography in prior annual screening examinations). We are inclined to doubt this: Our data show a significant number of cancers detectable with mammography but not US in women with dense breasts, and it has not yet been shown that US is able to determine when mammographically dense breasts evolve to fatty ones, at which time, screening probably should revert to mammography alone. Dr Hall is correct in stating that it is crucial to develop breast density criteria that yield results that are reproducible among assessors; the expense of an extra screening study should be avoided for any woman whose breasts are fatty enough not to need it.
Dr Hall’s concern about the shortage of qualified breast imagers is well founded; only innovative training and qualification initiatives are likely to solve this problem.
We cannot agree with Dr Kopans’ assertion that screening US will demonstrate findings that "look like cancer" in nearly all patients. If that were true, our biopsy rate would have been much higher than it was, and the published lesion detection rates of other investigators would not have agreed so closely with our own. We would welcome seeing the case series that led Dr Kopans to this conclusion.
We have never claimed that our data suggest that screening US should become the standard of care; we agree with Dr Kopans that, ultimately, demonstration of cost-effective reduction of morbidity or mortality must remain the standard for such policy prescriptions. We differ with Dr Hall’s suggestion that randomized trials might be avoided; his understandable concern for the considerable expense of screening US and the biopsy and follow-up required makes it even more important that any survival benefit be quantitated to determine the cost-benefit ratio of screening US. However, we defend the utility of our more preliminary clinical experiment: It would be inappropriate to consider searching for morbidity or mortality improvements until it had been demonstrated that additional cancers could be found with US, and we have now done so.
We hope that any mortality benefit that screening mammography might provide could be magnified with the use of screening US, since the sizes and stages of tumors found with US so closely resemble those of tumors found with mammography. It remains possible, however, that the biology of tumors detectable with US differ in some as-yet-unknown way from those detectable with mammography, which might make their diagnosis either more or less important.
Finally, we are concerned that Dr Kopans objects to the publication of studies that might lead to public confusion or concern. Some information is confusing—it needs critical attention and often leads to further valuable work. It is unfortunate that news media presentations may be misleading and even cause distress for some, but the proper response is to try to assure the accuracy of such presentations and to augment them with carefully considered professional announcements. To suppress publication of studies would be a big mistake.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  O. Graf, T. H. Helbich, G. Hopf, C. Graf, and E. A. Sickles Probably Benign Breast Masses at US: Is Follow-up an Acceptable Alternative to Biopsy? Radiology, July 1, 2007; 244(1): 87 - 93. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. B. Kopans Sonography Should Not Be Used for Breast Cancer Screening Until Its Efficacy Has Been Proven Scientifically Am. J. Roentgenol., February 1, 2004; 182(2): 489 - 491. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
