| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Commentary |
1 Department of Radiology, Box 1667, University of California, San Francisco School of Medicine, San Francisco, CA 94143-1667.
Received May 12, 2006; accepted after revision July 17, 2006.
Each month the American Journal of Roentgenology will republish
online one of the 100 most-cited articles from its first century. A
corresponding commentary in the print journal by a contemporary radiologist
will provide a current perspective. For a full list of these articles, see
page 3 of the January 2006 issue of the
AJR or go to
www.ajronline.org.
Keywords: breast cancer • mammography
John Wolfe's AJR publication in 1976
[1] of data on the association
of mammographic parenchymal patterns with future breast cancer risk
represented the culmination of almost 10 years of personal analysis
[2-4].
In that 1976 article, Wolfe classified 6 years of mammography examinations
(n = 7,214; case accrual, January 1967-January 1973) into four
parenchymal-pattern categories that he devised empirically: N1, primarily
fatty; P1, 
25% prominent ducts; P2, > 25% prominent ducts; and DY,
dense fibroglandular tissue.
Subsequent cancer diagnosis was determined by linking studied mammography examinations with the data stored in a regional tumor registry, reviewing institutional medical records, and using a questionnaire mailed to patients. Data analysis revealed a progressive increase in future cancer risk from N1 (lowest risk) to P1 (low risk) to P2 (high risk) to DY (highest risk) examinations. For the majority of studied examinations (n = 5,284), the highest risk (DY) category was associated with a 37 times higher future cancer risk than the lowest risk (N1) category.
Wolfe [1] also extrapolated the lifetime cancer risk for parenchymal pattern categories among women 40-59 years old, indicating that this risk was only 2% for the N1 category and 45% for the DY category (82% for the DY category in women 50-59 years old). Based on these calculations, he suggested that screening mammography programs might involve only women in the higher risk (P2 and DY) categories and that symptomatic women in the DY category should consider prophylactic subcutaneous mastectomy simply on the basis of this parenchymal pattern.
To place the 1976 Wolfe article [1] in proper perspective: It was published before the effectiveness of periodic mammographic screening had been widely accepted and at a time when there was concern that the presumed oncogenic risk of ionizing radiation imparted by mammography [5] might exceed the presumed benefit suggested by the early results of the first randomized controlled mammographic screening trial [6]. Consequently, in the Wolfe study most of the mammography involved patients who had signs or symptoms of breast disease (median age, approximately 47 years). These diagnostic examinations were limited to 90° lateral and craniocaudal views of each breast. Neither spot-compression nor magnification mammography techniques had been developed, scatter-reduction grids were not yet in use, what we now consider to be modern breast positioning and compression were not used, and breast sonography was not generally available. The technical advances inherent in modern mammography and sonography primarily augment the radiologist's ability to detect cancer in dense-breasted women [7-9].
Despite these technical advances, the sensitivity of mammography is still significantly reduced by increased breast density [10, 11] because of the masking of some tumors by dense fibroglandular tissue. The inclusion of such masked cancers among the future cancers in a risk assessment study inflates the estimate of risk, specifically the risk attributable to breast density. An effective method of reducing this masking bias is to exclude from study those cancers diagnosed within 2 years of mammography. Wolfe [1] excluded from study only those cancers diagnosed within 6 months of mammography. He did not have full access to clinical outcomes to report in detail on the interval between initial mammography and future cancer diagnosis, except that for a subset of 1,930 cases the mean and median intervals were 30 months and 32 months, respectively, with a range of 6-41 months. Thus, it is likely that approximately 40% of study cases involved intervals between 6 months and 2 years.
Therefore, the combination of a young, primarily symptomatic patient population, the choice of a 6-month exclusion interval, and the limited mammographic capability of the late 1960s and early 1970s made it likely that as many as 40% of the future cancers identified in the Wolfe study [1] actually represented undetected (masked) cancers, disproportionately so among the more densebreasted women. The net effect is that Wolfe overestimated future cancer risk to a substantial degree, leading him to what we now judge to be erroneous and unrealistic conclusions.
Even in 1976, clinical experience suggested to many radiologists that the Wolfe parenchymal patterns had a smaller effect on future breast cancer risk than Wolfe [1] reported. However, because John Wolfe was one of the most influential breast radiologists of his era, his article was taken seriously, it generated considerable controversy, and many follow-up studies were published to more precisely determine the magnitude of future cancer risk. Subsequently, as Wolfe's preferred technique of xeromammography was gradually replaced by screen-film mammography (which emphasized density differences in breast parenchyma rather than the visibility of prominent ducts), four categories of overall breast density replaced the four Wolfe parenchymal patterns as an index of parenchymal content.
The discovery of a significant relationship between increased breast density and decreased mammographic sensitivity in detecting cancer [10, 11] led to publication of another large series of articles, most of which appropriately cited the 1976 Wolfe article [1] as seminal to the understanding of the density-sensitivity association. Indeed, this association is why the American College of Radiology's Breast Imaging Reporting and Data System (BI-RADS) [12] recommends that every mammography report start with a description of breast density to inform the referring clinician about how the mammographic density of the patient may affect the sensitivity of the examination. More recently, another large series of published articles has now established a significant independent association between overall breast density and future breast cancer risk, albeit at a much lower magnitude (4-5 times increase) than the 37 times increase initially proposed by Wolfe for his parenchymal patterns [13, 14]. These several lines of scientific investigation (parenchymal patterns vs risk, overall density vs sensitivity, overall density vs risk) have accounted for 337 citations of the 1976 Wolfe article, placing it seventh among the 100 most frequently cited AJR articles [15].
Future work on the density-risk association is ongoing, including, but not limited to, development of quantitative methods to measure breast density as opposed to the subjective assessment of density currently provided in clinical practice [16-18] and attempts to identify specific genes responsible for dense breasts given the recent discovery of a heritable component to mammographic density [19]. Therefore, it is likely that the 1976 Wolfe article [1] will continue to be cited in many future publications.
The major purpose of the AJR in highlighting its previously published classic articles is to provide historical perspective to the important role radiology has played in medicine and in the management of specific diseases [15]; my previous comments should serve this purpose. However, a second stated purpose is to honor the authors of these classic articles [15]. So, even though the Wolfe mammographic parenchymal pattern classification has been replaced by breast density as an indicator of future cancer risk and we now understand that the magnitude of this risk is insufficient to justify Wolfe's proposal that screening mammography be limited to women in the higher-density categories, we should recognize the important contribution that John Wolfe [1] made in 1976 in leading the way, albeit indirectly, to our current knowledge on the subject. Eventually, mammographic density may be used to guide decisions concerning the supplementary use of nonmammographic imaging (sonography, MRI) for breast cancer screening, and mammographic density also may be used to guide decisions concerning primary chemoprevention of breast cancer.
Were John Wolfe still alive, I doubt that he would have selected the 1976 article on future breast cancer risk [1] as his most important contribution to medicine. Such are the vagaries of a selection process based solely on most frequent citation in subsequent articles. I believe that Wolfe's most important contribution to breast imaging was his major role in the development and popularization of the (now obsolete) technique of xeromammography at a pivotal time when its improved image quality and ease of use encouraged a critical mass of radiologists to provide and successfully promote the widespread use of mammography in the face of considerable and determined opposition. The many millions of women who already have undergone and who will continue to undergo mammography are the principal beneficiaries of Wolfe's work.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Related articles in AJR:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
