| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Beth Israel Deaconess Medical Center Harvard Medical School Boston, MA 02215
The important article by Sumkin et al. [1] could affect the way hundreds of millions of screening mammograms throughout the world are interpreted each year. These authors conclude that sensitivity is not affected significantly, but specificity using mammograms from an examination obtained 1 year previously as a reference was significantly better (p = 0.03) than specificity using mammograms obtained 2 years previously. Sumkin et al. state that "Mammograms obtained 1 year previously, when available, should be the ones to use as a reference."
There is an increase in specificity (decrease in false-positives) using more recent comparison mammograms because many small changes occur over time, particularly small cysts in premenopausal women or those undergoing hormonal replacement therapy. Additionally, lesions that were present before may only be recognized later because of the decrease in breast density that occurs with aging.
The major rationale for choosing to compare more remote mammograms is to increase sensitivity (true-positives) by identifying subtle slowly progressing malignancies, often in dense breasts, when the changes are less conspicuous on the more recent comparison study. Sumkin et al. [1] did increase sensitivity from 71% to 73% using the 2-rather than the 1-year comparison mammograms, but this increase was not statistically significant. I am concerned that the 128 cases they review, containing 30 cancers, was underpowered and does not prove their contention that a 2-year comparison will not identify more cancers.
Many mammographers, including myself, currently compare examinations from 2 years previously and, if an interval change is suspected, then compare with the 1-year reference and usually with more remote examinations as well. Occasionally, the more remote examinations are more useful than the recent one, particularly regarding calcifications. Admittedly, making these comparisons increases the time needed for interpretation, but this should not be an undue burden in the 5–10% of screening patients normally recalled for additional imaging.
The study by Sumkin et al. [1] was necessarily performed in an artificial environment in which reviewers had the choice of only 0-, 1-, or 2-year comparison examinations. In real life, these 128 cases, all of which were selected because they had been recalled or had subtle interval changes, would have had multiple previous mammograms available for comparison. The dozen reviewers in this study expressed an overwhelming subjective preference for the 2-year comparison examination.
I commend the authors for their study. However, I think the jury is still out on the issue of optimal comparison intervals for screening mammography.
References
University of Pittsburgh Pittsburgh, PA 15213-3180
We thank Dr. Hall for his comments. It is always hard to challenge one's own long-term practices and beliefs, and it was for us, as well. We certainly hope that our small preliminary study [1] is not seen as the sole reason to change clinical practices in millions of examinations. Although limited in size and perhaps affected by selection bias, our study is what it is, and it has already caused us all to think about this possibly important aspect of our practice. Whatever the optimal comparison procedure may be, the subject has not been studied adequately, and, if nothing else, our preliminary study indicates that it should be. We fully agree that the jury is still out regarding this issue, but before our study, limited though it was, there was no "jury," and it definitely was not "out."
In our previous response, to Kopans' letter, we covered many of the points raised in Hall's letter. The fact that many of these cases had been originally recalled on the basis of an initial comparison with the 2-year-old mammograms, followed by a review of other prior examinations as needed, could have biased it against the 1-year mode in terms of sensitivity, and it did not, at least not to the extent that specificity was affected.
We emphasized that our own reviewers subjectively preferred the 2-year-old mammograms, and it is possible, if not likely, that their reaction was similar to that of Hall. Is it possible that a long-standing comfort level with one's current practice makes it hard to challenge it or accept it as possibly suboptimal? Our study was carried out in a laboratory environment and was statistically underpowered to assess small changes in sensitivity. We fully agree that the benefit of increasing sensitivity significantly outweighs comparable changes in specificity and believe a more comprehensive study is needed. If this debate leads to such a study, our article will have done more than we expected.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
