| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorials |
1 From the Department of Radiology, University of Pittsburgh, Imaging Research, 3362 Fifth Avenue, Pittsburgh, PA 15213. Received December 6, 2006; revision requested January 3, 2007; revision received January 8; final version accepted February 1. Address correspondence to the author (e-mail: gurd{at}upmc.edu).
In light of the recent New England Journal of Medicine article (1) describing the results of the Digital Mammographic Imaging Screening Trial (DMIST), I, and I suspect others, have been asked more frequently whether or not there is a "need" to change from using film mammography to using digital mammography now, because not using digital mammography may carry a possible liability if a cancer is missed, in particular in the subgroups of women identified by the DMIST study (ie, women younger than 50 years of age, premenopausal women, and women with dense breast tissue). I have heard similar discussions and I have been asked this particular question in private, as well as during the 2006 Radiological Society of North America Annual Meeting. The question was posed with such sincerity that I believe a discussion of my own opinion may be of interest.
First, in a recent issue of Medical Physics, a comprehensive debate entitled "Film mammography for breast cancer screening in younger women is no longer appropriate because of the demonstrated superiority of digital mammography for this age group" was presented on this topic in a "Point/Counterpoint" section (2). Unfortunately, many of those involved in such a decision are not frequent readers of that journal, so I recommend reading this debate for those who wonder about the need for an immediate conversion. However, in my personal opinion, given here, I will try to address the issue from a somewhat different perspective and raise some points that were not addressed in the aforementioned published debate.
Despite the fact that most savvy investigators in the field could find several flaws in and disagree with at least some of the results of the DMIST that were published in the New England Journal of Medicine (as is frequently the case with most scientific publications), I wish to emphasize that we should all applaud the American College of Radiology Imaging Network and the DMIST investigators for successfully executing a very difficult large multiinstitutional prospective study and for publishing the results with due consideration of many aspects that are typically not addressed in many experimental studies. However, I mention here but one nonanalytical comment that was not addressed, to my knowledge, and that is a key to why I believe the published results of the DMIST study by themselves cannot and should not be used as a primary reason to rush to convert to digital mammography.
When using the results of any single study for decision-making purposes, even when that study is what is considered a "pivotal study," we need to adhere to some fundamental principles of consistency and plausibility. If we ignore the film-based results from the DMIST study altogether, the digital results alone seem to suggest that the diagnostic performance of digital mammography as measured by the area under the receiver operating characteristic curve (AUC) in the subsets of great interest—namely, young and premenopausal women (which populations, by the way, overlap substantially)—is substantially and potentially significantly better (ie, AUC in women younger than 50 years of age was 0.84) than the diagnostic performance of digital mammography in the general population, and definitely in the older, postmenopausal women in the DMIST study. In the latter subgroups of women, the performance of digital mammography is difficult to compute directly without the raw data, but the AUC is likely to be lower than 0.78 when one assumes that the diagnostic performance in the whole population had an AUC of 0.78 and approximately one-third of the whole population belonged in one of the subsets of younger and/or premenopausal women, in whom the performance was higher than an AUC of 0.78. This observation—namely, that the performance of full-field digital mammography (FFDM) in younger women was substantially and possibly significantly better than that in older women—is inconsistent with expectations, and, while theoretically possible, is unlikely and not very plausible. At best, one would expect the performance of digital mammography in these subgroups of women to be equal to that of the performance of digital mammography in older, postmenopausal women or in women with fatty breast tissue. This point has little to do with the comparison made with film mammography in the DMIST study, but it raises the question of whether or not, for some reason (and one can think of several), the subsets on which the publication focuses are experimental outliers that should not be used for decision making in practice without careful consideration and clear evidence of reproducibility.
I note, however, that the fact that the reported performance of film mammography decreased substantially in women younger than 50 years of age, as compared with the performance of film mammography in women older than 50 years of age, suggests that FFDM may indeed perform substantially and possibly statistically significantly better in this group of women (<50 years of age), even if the actual performance of FFDM is not as high as that reported in the DMIST study. However, such a conclusion should not be reached on the basis of the DMIST results alone because of the unusually high performance reported for FFDM in this younger group of women.
Because consistency and plausibility should be important factors in decision making on the basis of results of pivotal studies, I would argue that conversion to digital mammography should not be made primarily (or solely) on the basis of the results of the DMIST study. However, at our institution, we made the change, and the reader who wonders about this issue should seriously consider doing it too, not because one feels one is "forced" to convert as a result of one study but rather because one believes in it and one knows this is the future of mammography for a large number of reasons that have been frequently and eloquently discussed by many. Ultimately, digital mammography is better for the patient and for our future medical practice, even if the diagnostic performance does not actually improve in the subsets of younger or premenopausal women by as much as seems to be indicated from the DMIST project. Even if we do not know the exact improvement one can obtain when using digital mammography, there are enough data (including those from DMIST) to indicate that, when done right, digital mammography is unlikely to be worse than film mammography for depiction of cluster microcalcifications and is very likely to be better in almost all other diagnostic tasks and, in particular, for mass detection (3,4).
Once one comes to this realization, the question is not "Whether," but rather, "When and at what cost?" One could argue that the best approach would be to delay conversion until technology options stabilize and perhaps until the role of new approaches, such as tomosynthesis mammography and breast computed tomography, are better understood, widely accepted, and are perhaps routinely and widely used in screening mammography. The fact is that the cost of these newer, more advanced technologies is not likely to be lower, and clinical practices when using them are not going to be simpler. Therefore, conversion to digital mammography should not be postponed primarily for the eventual stabilization of future technologies or a new standard of practice. After decades of incremental improvements (eg, dose reduction and improved contrast) in film mammography, we have entered the era of rapid, first-order types of innovations in breast imaging, and, while current film practices certainly remain acceptable, we will soon approach a point when these are considered ancient, if not obsolete. Therefore, on the basis of timing and cost considerations (not only in terms of dollars but also in terms of the cost of transitioning the practice, which is not trivial), we should all begin to seriously address the issue of conversion to digital mammography. I know we, who are not afraid to be critical of ourselves (as well as others), did it extremely carefully and methodically, and have already made the transition.
Last, one aspect that cannot be ignored is the environment in which one operates. Clearly, competitiveness in the marketplace does affect strategic decisions in radiology (in terms of the desire to, for example, "be the first to convert" or definitely "not be too far behind in implementing digital mammography"), and there is no doubt that these forces will play a role in when the conversion is made. However, this issue, as important as it may be, is beyond the scope of this commentary.
 |
FOOTNOTES |
|---|
Author stated no financial relationship to disclose.
|
References |
|---|
 TOP References |
|---|
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
