| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Perspectives |
1 From the Department of Radiology, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA 02215. Received January 23, 2008; final version accepted January 23. Address correspondence to the author (e-mail: fhall{at}bidmc.harvard.edu).
Discuss this article online at www.rsna.org/radiology/discuss.
It's tough to make predictions, especially about the future.
—Yogi Berra
I believe that mammography is going to be replaced by magnetic resonance (MR) imaging as the standard for breast screening, not only in high-risk women but increasingly in those at average risk. Breast MR imaging is already widely used as a diagnostic examination, but its emergence as a screening study is recent. In this perspectives article I will discuss why I think this change is going to occur and many of its downstream ramifications. The latter will include a presumed reduction in breast cancer mortality and an alteration in behavior and thinking for a generation of women and their physicians for whom screening mammography has become an annual staple. Breast MR imaging is also going to reinvigorate breast imaging, much as it did for other organ-based subspecialties such as neuroradiology and musculoskeletal imaging.
Screening mammography is the modality that literally created the subspecialty of breast imaging, with reduced mortality from breast cancer of approximate 25%, ranging from 15% to 35% depending primarily on age (1). Of course, most man-made entities rise and fall, including civilizations. I was taught in medical school 50 years ago that half of what I learned and practiced would be nonapplicable, or simply incorrect, in a decade. That time interval is currently far shorter, and screening mammography, introduced a quarter of a century ago amid great controversy as to its cost, benefit, and possible harm, has endured far longer than most of us predicted.
Unfortunately, a mammogram, like other radiographs, is a two-dimensional planar image with inherent limitations in identifying three-dimensional lesions, particularly in the relatively homogeneous soft tissues of the breast. It cannot compete in sensitivity with MR imaging, which provides not only volumetric assessment of morphology by using a variety of pulse sequences, but also contrast material–enhanced kinetic measurements of the vascular perfusion of individual voxels (2–4). Tumor angiogenesis, as measured by using contrast material enhancement and washout, is closely associated with invasive cancer and with high-grade ductal carcinoma in situ (5–7). For me, interpreting a mammogram in a dense breast is painfully analogous to looking at an abdominal radiograph rather than the corresponding contrast-enhanced computed tomographic (CT) or MR image. Kudos to mammographers who, despite the inherent limitations of this modality, have accomplished so much in their efforts at earlier detection and mortality reduction of breast cancer. Imagine how much more we might be able to accomplish with an inherently more sensitive examination.
In 2007, the American Cancer Society recommended annual breast cancer screening with MR imaging for women with an "approximately 20%–25% or greater lifetime risk of breast cancer" (8,9). This report was based on a comprehensive study of the literature by a distinguished group of international experts and, "where evidence was insufficient or lacking, the final recommendations incorporated the expert opinions of the panel members" (8). Eligibility figures were based on BRCAPRO or other risk models available at the time, which were determined largely by using family history. It is noteworthy that most of these models did not incorporate such important factors as mammographic density and density relative to age, ethnicity, and the number of female relatives available for assessment (10–14).
The American Cancer Society is almost certainly correct that screening breast MR imaging in high-risk women, however that risk is defined, will help identify more breast cancers at an earlier stage and presumably reduce mortality. Results of multiple studies (15,16) have shown that breast MR imaging helps identify three to four unsuspected cancers per 100 examinations in higher-risk women, which is approximately 10 times the four to five cancers per 1000 sensitivity of screening mammography. And recall that the American Cancer Society was prescient in its similarly controversial recommendations for screening mammography at a time when mammographic false-positive findings and interobserver variation in interpretations were very high and that most of us were skeptical of both the practicality and cost of such a program. I believe that screening breast MR imaging will someday be recommended for many women at only a 14% (one in seven) lifetime risk of breast cancer. Since this is the risk of the average American woman, I am suggesting that MR imaging will gradually replace screening mammography, albeit with many exceptions. This transition will be abetted by improved risk models, with women at low risk and with fatty parenchyma probably continuing to undergo annual mammography.
Some mammographers and other physicians, particularly those without access to breast MR imaging, felt blindsided by the American Cancer Society recommendation because there currently are not enough facilities or experienced radiologists to accomplish this proposal. Many of us have been inundated by often unanswerable inquiries from patients and their physicians regarding eligibility, availability, and insurance. Some felt that a more reasonable initial American Cancer Society guideline might have been a higher 25%–30% lifetime risk of breast cancer, perhaps with MR imaging screening performed only every 2 years, with intervening mammography. Of course, if patients or their insurers pay, then economic incentives will ensure that these examinations are performed. However, given the current shortage of breast imagers, it will take years to develop a cadre of radiologists trained in breast MR imaging and MR imaging–guided interventions. This is reminiscent of the inception of screening mammography, a similarly disruptive modality requiring new areas of expertise and innovation in both interpretation as well as preoperative localization of nonpalpable abnormalities (17).
The major arguments against breast MR imaging are that its lower specificity (increased false-positive findings) and higher cost make it unacceptable as a screening modality in which there is a low pretest probability of cancer. These same issues were raised 25 years ago when newly introduced screening mammography was compared with the heretofore standard of physical examination. At that time, poor-quality mammograms (by contemporary standards) were often interpreted by inexperienced radiologists (and we were all relatively inexperienced), leading to many false-positive findings. These problems were eventually moderated by the introduction of quality standards for technical performance, interpretation and recommendations (Breast Imaging Reporting and Data System), and continued medical education, many of which were federally mandated in 1994. Currently, mammography is by far the most regulated imaging modality in radiology. This contrasts with breast MR imaging which lacks almost any standards (2), although a new American College of Radiology voluntary accreditation program for breast MR imaging will be introduced in 2008. Regulations, standards, and accreditation usually follow the introduction of new examinations, and they will, in my opinion, alleviate many of the present concerns about the implementation of cost-effective screening breast MR imaging.
When I interpret screening mammograms I see perhaps 10 "probably normal" lesions per day where I estimate the probability of cancer to be less than one in 100. I cannot recall these individuals and maintain the American College of Radiology standard of an overall callback rate of 10% or less. However, I know that a half-dozen times per year one of these findings will subsequently be shown to be a cancer, one that I had seen, debated, and dismissed. This is not surprising since approximately 50% of cancers are retrospectively identified on the previous annual mammogram and unfortunately are often evident to lay jurors. The point is that mammographers must be willing to take these "risks" in order to offer a cost-effective product. Experienced mammographers have learned that, in the words of President Harry Truman, "if you can't take the heat, get out of the kitchen," that is, stop reading mammograms. I believe the high false-positive rates currently seen at breast MR imaging will substantially ameliorate as these studies are increasingly interpreted by experienced organ-based (breast) rather than modality-based (MR imaging) radiologists and as guidelines are developed for callbacks and biopsies on the basis of a woman's risk assessment (2). In my opinion, very low suspicion MR imaging findings, such as the frequently encountered "contrast-enhancing ditzel," should be ignored.
Kuhl (2) has pointed out that "early reports on breast MR imaging (particularly from the morphology camp) found poor specificity—as low as 30%", which "established the technique's reputation for low predictive value and specificity ... more current publications suggest that breast MR imaging does not unduly increase the false-positive biopsy rate and that it provides about the same specificity as mammography." MR imaging callback rates and 6-month follow-ups have decreased in almost all published series in which women were screened for more than one round (2,8,15). This presumably reflects both the increased experience of imagers and the advantage of having previous comparison examinations. Of course, regardless of the breast screening modality, false-positive rates will remain relatively high in the United States because of our greater medical-legal concerns, lower-volume readers, and our affluent society's willingness to pay a higher price per quality life year saved. Even today, many experienced mammographers in the United States have difficulty keeping their screening callback rates at the American College of Radiology recommended level of 10% or less, while similar recommended recall rates in Europe are 5% for baseline examinations and only 3% for all subsequent annual examinations.
MR imaging is more expensive to perform and more time consuming to interpret than is mammography. However, the $500 000 cost of a digital mammographic machine performing four examinations per hour, with its requisite high-resolution image storage requirements, copiers, and 5-megapixel monochrome monitors, puts cost at least in the same ballpark as the $1.5-million magnet performing two to three screening examinations per hour. The latter costs and times will improve with the efficiencies that come with high volume and dedicated breast MR imaging equipment. Likewise, the time for interpretation of screening breast MR images should decrease with experience and improved computer-aided detection software.
It is conceivable that the higher sensitivity of MR imaging in identifying small invasive cancers might mean that screening would be necessary in some women only every 2 years. After all, the current yearly intervals for screening are relatively arbitrary, as is the commonly recommended baseline age of 40. The average doubling time of invasive breast cancer is 120 days, corresponding to 8 years of sojourn time prior to its becoming recognizable on a mammogram or becoming palpable. These doubling times are somewhat analogous to those for colon cancer where screening for average-risk persons is at far longer intervals (18), although colonic screening is benefited by these cancers developing within more discernable adenomatous polyps. Mammographic screening in Europe has often been performed biannually in the past. This presumably resulted in delayed diagnosis, although reported reductions in breast cancer mortality have been similar to those reported in the United States.
Although the high sensitivity of breast MR imaging in the diagnosis of breast cancer is well documented, to my knowledge there are no randomized controlled studies to assess whether this increased detection will affect disease-specific or overall mortality, as well as cost-effectiveness. Such trials are expensive, require large numbers and long follow-ups, and are often controversial, as illustrated by early breast cancer screening trials and by the current contentious debate regarding screening chest CT for lung cancer in high-risk smokers. It remains to be seen how feasible or meaningful such trials and/or studies of screening breast MR imaging will be, particularly given the variable risk factors in women currently undergoing these examinations and the inevitability of interval technologic and interpretive improvements in the examination. In the meantime, many assumptions regarding screening breast MR imaging will be based on the robust inferential evidence from screening mammography, the only breast screening modality with a proved mortality-reduction benefit with an estimated cost per quality life year saved in the United States of $25 000–$30 000. Less-affluent countries often further boost this cost-effectiveness by using biannual examinations, by beginning screening at later and ending at earlier ages, or by recommending fewer callbacks and biopsies.
Even without "proof" of decreased mortality, it is difficult for me to believe that screening breast MR imaging will not find its way into the standard medical armamentarium, particularly for such a common and emotionally charged disease as breast cancer. Political lobbying resulted in Congressional approval of payments for mammographic screening, digital mammography, and computer-aided detection, all with no proof and little evidence-based verification of benefit. There will be, and should be, legitimate questions about whether screening breast MR imaging is the best use of our health care dollars, but in affluent America, cost has rarely been a long-term impediment to the acceptance of a seemingly better medical mousetrap, and certainly breast screening is far more cost-effective per quality life year saved than much of what the medical profession does. I think it will be difficult for physicians, insurers, and politicians to deny this more sensitive diagnostic study to knowledgeable patients, whose fear of breast cancer often trumps their concerns about false-positive findings (2). Patient demand has become a major driving force in medical decision making, including medical imaging (19) and, as noted above, nowhere has this been more evident than in breast screening.
The higher spatial resolution of mammography, permitting better assessment of the microcalcifications associated with ductal carcinoma in situ, had been thought to be a drawback of breast MR imaging. However, results of recent work (5–7) suggest that contrast-enhanced MR imaging is actually more accurate than mammography in diagnosing ductal carcinoma in situ, particularly the less common but clinically more important high-grade ductal carcinoma in situ. This relates to ductal carcinoma in situ and other proliferative lesions, such as atypical ductal hyperplasia, apparently having increased blood flow that can be identified with contrast-enhanced MR imaging. This is important because approximately one-third of biopsies performed on the basis of screening mammographic findings are performed for calcifications, but only 20% of cancers identified at mammography are ductal carcinoma in situ, with the great majority of these being lower-grade. Thus, it appears that "the detection of DCIS [ductal carcinoma in situ] contributes in only a minor way to the mortality reduction gained from screening" (20).
Breast MR imaging should ultimately result in fewer errors of detection, versus errors of interpretation, because of its higher sensitivity compared with mammography. These interpretive errors can be expected to improve with experience. As with most newly introduced modalities, MR imaging screening will initially be performed as an adjunct to mammographic screening. The relative contribution of mammography in women undergoing screening MR imaging is unknown but probably will be far less than the 10%–15% of palpable cancers missed at screening mammography. Perhaps women will initially undergo mammography to assess breast density, both as a guide to their risk of cancer and mammographic sensitivity. Very high-risk women might undergo both periodic MR imaging and mammography, but it is very unlikely to be cost-effective to annually perform both these examinations in most women. I believe that mammography will gradually become more of a problem-solving diagnostic examination, much as breast MR imaging and ultrasonography are in current use.
Screening programs are inherently inefficient because an entire population is subject to the test, but relatively few individuals benefit, and some are actually harmed. Unfortunately, most breast cancers occur in women with few if any known risk factors. BRCA mutations account for 5%–10% of breast cancer with an estimated equal number of as of yet unidentified intrinsic genetic risk factors. Ideally, predictive genetic testing might allow us to better target who undergoes screening. However, despite the increasing commercial availability and acceptance of personalized genetic information, many experts believe it is unlikely that this will play any major role in determining breast cancer screening strategies in the foreseeable future (21,22).
Advantages of breast MR imaging include lack of radiation and painful compression. Counterbalancing this is the necessity for contrast material injection, menstrual cycle variations, and exclusion criteria such as claustrophobia, pacemakers, renal disease, and cerebral aneurysm clips. The transition from screening mammography to screening MR imaging, with breast MR imaging probably becoming the largest user of MR imaging, will have many downstream ramifications. Winners and losers would include the manufacturers of mammographic and MR imaging hardware and their respective computer-aided detection software. Computer-aided detection for breast MR imaging, unlike the more controversial mammographic computer-aided detection (23,24), is essential for the expeditious interpretation of the thousands of dynamic contrast-enhanced breast MR images. The place of MR spectroscopy, currently a fringe player in diagnostic breast imaging, might change. The development of tomosynthesis and breast stereoscopic digital mammography, which I view as transitional volume imaging modalities similar to traditional radiographic tomography, could be adversely affected. There will certainly be political ramifications regarding insurance coverage and payments.
As screening breast MR imaging becomes commonplace, one important ancillary winner will be the subspecialty of breast imaging. This is the newest and by most criteria the weakest and most "low-tech" of all radiology subspecialties (24). Breast cancer is the most common basis for lawsuits in radiology. Stress levels have further increased because virtually all breast abnormalities are examined with core-needle biopsy with imaging guidance prior to consideration of surgery. Screening mammography often loses money and is subsidized by other sections of the radiology department. It is heavily regulated with added administrative burdens and unfunded mandates. Additionally, reading screening mammograms is, for a number of us, intellectually relatively boring. An indirect sign of this lowly status is our lack of turf battles—no one covets the breast imaging business. Not surprisingly, a generation of radiology residents has avoided breast imaging. This will almost certainly change with the advent of MR imaging, which has led to clinical advances and exciting research opportunities wherever it has been introduced.
In conclusion, screening breast MR imaging, already a reality in high-risk women, is increasingly being used to screen those of intermediate risk. I believe this modality will eventually be used to screen most women. How efficiently this comes about will depend in part on the lessons we have learned from the introduction of screening mammography. I am a bit saddened by this anticipated demise of screening mammography because this modality helped launch my own academic career, and, in my waning professional years, I am spending a disproportionate amount of my time interpreting these studies. The excitement of breast MR imaging might even encourage me to curtail any plans I have for retirement. However, there will be no eulogy for screening mammography because, to paraphrase General Douglas MacArthur, old imaging modalities never die, they just fade away.
Discuss this article online at www.rsna.org/radiology/discuss.
 |
FOOTNOTES |
|---|
|
References |
|---|
 TOP References |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Chun, B. Pocock, K.-A. Joseph, M. El-Tamer, L. Klein, and F. Schnabel Breast Cancer Risk Factors in Younger and Older Women Ann. Surg. Oncol., January 1, 2009; 16(1): 96 - 99. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. B. Kopans and F. M. Hall The Science Is Needed Radiology, November 1, 2008; 249(2): 719 - 720. [Full Text] [PDF]
|
|
|
|
|
|
P. Loubeyre, P. Petignat, and F. M. Hall Is the Field of Breast Imaging Moving toward Broader Use of MR Imaging? Radiology, November 1, 2008; 249(2): 721 - 722. [Full Text] [PDF]
|
|
|
|
 |
|
 C. K. Kuhl Nonmammographic Screening for Breast Cancer--Reply JAMA, October 1, 2008; 300(13): 1516 - 1517. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
