| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Boston, MA 02215
e-mail: fhall{at}bidmc.harvard.edu
I commend Dr Martin and colleagues (1) on their article, published in the September 2006 issue of Radiology, comparing a fully automated computer-aided mammographic density estimation (MDEST) program with estimates by radiologists and Breast Imaging Reporting and Data System (BI-RADS) density categories (2). These authors correctly point out that there is "no reference standard for determining breast density." Therefore, however it is measured, breast density is currently a subjective assessment. By using the four traditional BI-RADS density categories (2), authors of several articles (3–5) have classified over 90% of their mammograms into the scattered fibroglandular densities (25%–50% dense) and heterogeneously dense (50%–75% dense) groups, leaving as little as 4% in the fatty (0%–25% dense) and 2% in the dense (75%–100% dense) groups. More recently, some authors have used only two categories, combining fatty and scattered fibroglandular densities into a predominantly fatty category and combining heterogeneously dense and dense categories into a predominantly dense category. I believe that attempts to pigeonhole mammograms into groups on the basis of their appearance, or ratio of dense to fat tissue, potentially defeats the original purpose for their development—which is, or was, to convey how sensitive the mammogram is believed to be in detecting cancer.
Digital software programs can provide a reproducible and continuous spectrum of breast densities, as shown by Dr Martin and colleagues (1). However, this may not correlate with the sensitivity of the mammogram in depicting a mass. How does one compare a 50% dense with a 25% very dense breast? Rather than use density categories, I would prefer suitability categories such as: well suited, moderately well suited, not well suited, and poorly suited to mammographic interpretation for masses (calcifications are minimally affected by density). Suitability categories might be less reproducible across all practices, but they would be presumably more meaningful in regard to the individual mammographer's assessment of his or her own interpretive accuracy.
I teach my residents that breast density (or suitability) categories, like most things biologic, have bell-shaped curves. Therefore, if they use the standard four BI-RADS categories they should, over the long run, have approximately 20% fatty, 30% scattered fibroglandular, 30% heterogeneously dense, and 20% dense findings. An automated software system, using percentage density, could easily be programmed to do this. While the bell-shaped curve model might lead to absolute variations depending on whether one's practice consists more of thin or buxom individuals, there would be more relative consistency than exists under most current practices (2–4).
In conclusion, in our efforts to quantify mammographic density, it is important not to lose sight of the forest because of the trees. The current clinical value of assessing breast density is minimal and, outside of research protocols, it rarely affects decisions to undergo further imaging with ultrasonography or magnetic resonance imaging. The major purpose of these categories is, or should be, to convey to our clinical colleagues what the interpreting mammographer believes is his or her relative sensitivity in detecting carcinoma.
 |
References |
|---|
 TOP References References |
|---|
11881 East Carol Avenue, Scottsdale, AZ 85259
e-mail: yango{at}cox.net
We thank Dr Hall (1) for his insight into mammographic density and agree with him that the primary intent of the BI-RADS density categories (2) is to convey the relative sensitivity of mammograms in the detection of breast carcinoma. In this regard, as Dr Hall suggests, perhaps the use of density categories is not the best method of indicating the suitability of the mammogram in the detection of breast cancer, since a mammogram labeled as heterogeneously dense (51%–75% dense) may not be as difficult to interpret as another classified as containing scattered fibroglandular densities (25%–50% dense). Thus, even for its primary purpose, BI-RADS density classifications have limitations and interobserver variability.
However, mammographic density is also assuming a separate and renewed importance since it has been identified as a predictor of the risk of developing breast cancer (3). The new version of the Gail model (4) and another recently developed model of determining breast cancer risk (5) both use breast density as an independent risk factor, with increasing density correlating with increasing risk. To this end, an automated computer-aided mammographic density program such as our (MDEST) program is a tool to more reproducibly assess density than the more subjective BI-RADS categories. As Dr Hall notes, these tools are being used in current research protocols (4). Serial mammography measurements of density may prove important to quantify the success of risk reduction strategies. Thus, we believe that measurement of mammographic density may take on more clinical importance in the future than its current role of being used primarily as an indicator of the relative sensitivity of mammography.
|
References |
|---|
|
TOPReferencesReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. M. Hall and D. B. Kopans, MD Mammographically Determined Breast Density and Cancer Risk Radiology, September 1, 2008; 248(3): 1083 - 1084. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
