Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
News Ten Years Later Breast Density Gains Acceptance As Breast Cancer Risk Factor By Vicki Brower This is part of an occasional series that updates some of the stories reported 10 years ago in the news section of the Journal. Breast cancer patients with very dense breasts who undergo lumpectomy have a greater risk of recurrence than women with less dense breasts, according to a new 10-year Canadian study that appeared in the December 15 issue of Cancer. Led by Steven Narod, M.D., at the University of Toronto, the study is the most recent confirmation of the importance of breast density in the biology of breast cancer. Ten years ago, a JNCI news story reported that breast density 374 News | JNCI had been shown to be a risk factor for breast cancer, but that there was little agreement on how to measure density or how it contributed to risk. Today, many studies later, those questions remain, but density is more firmly established as a key risk factor and has been incorporated into widely used breast cancer risk models. “During the past decade, about three dozen studies have demonstrated that breast density may be the single most important factor for cancer risk,” said Jack Cuzick, Ph.D., professor of epidemiology at the Wolfson Institute of Preventive Medicine in London. Scientists now estimate that dense breasts raise a wom- an’s risk of cancer sixfold, said Thea Tlsty, Ph.D., professor of pathology at the University of California in San Francisco. Ten more years of research has also affected clinical practice. Although earlier studies showed that density was a risk factor along with age, parity, family history, and hormone use, clinicians have begun to accept it in just the past few years, because several seminal reports appeared in mainstream journals, said Celine Vachon, Ph.D., of the Mayo Clinic in Rochester, Minn. Researchers first turned their attention to breast density when J.N. Wolfe of Detroit noticed an association with cancer in 1976. Vol. 102, Issue 6 | March 17, 2010 N EWS Breast density is a phenomenon that shows up on mammograms because fatty tissue is relatively translucent to x-rays, whereas epithelial and stromal cells block x-rays and appear as white areas. Thus “dense” breast tissue contains less fat and more epithelial and stromal cells; it also has more collagen and acini—small, milk-producing elements in breast lobules. Many researchers refer to this concept as mammographic density. Little research occurred, however, until the 1990s. “While the field is now . . . well funded and populated, till the mid-1990s you could fit all of us into a phone booth,” said Norman Boyd, M.D., who studies breast density at the Ontario Cancer Institute in Toronto. The Gail model, a widely used tool to assess breast cancer risk, was revised to include density in 2006. Its creator, Mitchell Gail, M.D., Ph.D., at the National Cancer Institute, said that the revised model is available only for research purposes at present, because it has not yet been validated in independent data. There are now other risk models in development incorporating density, some of which are not easily used by clinicians, said Karla Kerlikowske, M.D., professor of medicine, epidemiology, and biostatistics at the University of California in San Francisco. She and colleagues have developed a model that she said is easily employed and applicable to many ethnic groups. It was published in the Annals of Clinical Acceptance Internal Medicine in 2008. The ultimate goal is In 2007 Boyd published research in the New to use risk prediction to stratify women for England Journal of Medicine topping off three primary and secondary prevention, she said. decades of study and marking a turning point for Although density is now part of some risk wider acceptance of breast density, said Gertrude models, clinicians could be doing more accordMaskarinec, M.D., Ph.D., of the University of ing to Cuzick. “Breast density is definitely Hawaii in Honolulu. Before that, many clinicians underapplied in risk assessment by clinicians, thought that mammographic density only masked and patients do not understand its importance cancer. Boyd’s analysis of three nested case– either,” he said. “Breast density should be taken control studies with more into consideration in than 2,000 women demdetermining how fre“While the field is onstrated that masking quently individuals get couldn’t account for a mammograms, and if now. . .well funded and fivefold-increased risk. they should receive adpopulated, till the mid-1990s ditional tests such as Depending on age, 16%–30% of breast you could fit all of us into a ultrasound, [computed cancers can be associtomography], or [magated with high density, netic resonance phone booth.” Boyd said. Family hisimaging].” tory and known oncogenic genes account for a One reason for the gap between risk much smaller proportion of cancer. research and clinical practice may be that no Boyd and others have shown that density is widespread accepted “treatment” exists for highly heritable; is related to tumor size, lymph mammographic density. Some studies show node status, and invasiveness; and is associated that tamoxifen and possibly raloxifene reduce with two aggressive breast cancer subtypes, lu- density, along with cancer incidence, but a minal A and triple negative. Women with causal connection has not been shown yet. mammographic density have a different cancer A study led by Cuzick, called IBIS I and profile, including larger and more interval can- published in JNCI in 2004, showed that tacers, after adjustment for age, hormone use, moxifen reduced density and risk in some family history, and mode of detection, accord- women within 12–18 months. In an update 4 ing to a study led by Carolyn Nickson of the years later, Cuzick reported that women who University of Melbourne in Australia, pub- had a moderately elevated risk of breast canlished last year in the Journal of Medical cer and who reduced breast density by 10% Screening. Steven Cummings, M.D., director of were half as likely to develop cancer as those the San Francisco Coordinating Center at the on placebo. In contrast, women who had little California Pacific Medical Center Research or no change in density had a risk about equal Institute reported last year that adding density to those on placebo. to risk models improves predictive accuracy. Other studies of this issue are under way. (see J. Natl. Cancer Inst. 2009;101:384–98). “More data is forthcoming that will show jnci.oxfordjournals.org breast density to be a potential surrogate marker of therapeutic efficacy of chemoprevention, which will also underscore the clinical significance of breast density,” Vachon said. Measuring Density Perhaps the most important reason that mammographic density research has not gained greater acceptance by clinicians is the lack of any standardized, quantitative, and automated method of density measurement available to them, Gail said. Radiologists frequently use a subjective, qualitative tool, BIRADS, which designates density from 1 to 4, from predominantly fatty to highly dense tissue. New quantitative methods of measurement are being developed. Many researchers now use a semiautomated, user-assisted measurement method, Cumulus. Vachon and collaborator John Hein, Ph.D., of the H. Lee Moffitt Cancer Center in Tampa, Fla., are developing a totally automated digitized system to estimate density. Martin Yaffe, Ph.D., senior scientist, imaging research at the Sunnybrook Health Sciences Centre in Toronto, and Cummings and Kerlikowske have developed an automated software and hardware system to measure density with volume. Their system is installed onto 26 digital mammogram machines around the country and will be used Norman Boyd, M.D., D.Sc. to study 250,000 women prospectively in a longterm density study, according to Tlsty. Researchers also question whether it is better to measure the percentage of dense area or the absolute dense area. Cuzick and Jennifer Stone, Ph.D., of the University of Melbourne in Australia recently found that percent dense area is associated with many risk factors but that dense area is not, making the latter a simpler biomarker for risk prediction modeling. Their results appeared last year in the American Journal of Epidemiology. Analyzing parenchymal patterns according to texture, measuring volume using computed tomography and tomosynthesis, and calculating collagen and tissue stiffness are other approaches being studied. © Oxford University Press 2010. DOI: 10.1093/jnci/djq080 JNCI | News 375