To07.endcancer.ca

6. Chi JT, Wang Z, Nuyten DS, et al. Gene expression programs 10. Reya T, Morrison SJ, Clarke MF, Weissman IL. Stem cells,
in response to hypoxia: cell type specificity and prognostic sig- cancer, and cancer stem cells. Nature 2001;414:105-11.
nificance in human cancers. PLoS Med 2006;3:e47.
11. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ,
7. Minn AJ, Gupta GP, Siegel PM, et al. Genes that mediate breast Clarke MF. Prospective identification of tumorigenic breast can-
cancer metastasis to lung. Nature 2005;436:518-24.
cer cells. Proc Natl Acad Sci U S A 2003;100:3983-8. [Erratum, 8. Paik S, Shak S, Tang G, et al. A multigene assay to predict Proc Natl Acad Sci U S A 2003;100:6890.]
recurrence of tamoxifen-treated, node-negative breast cancer. 12. Fan C, Oh DS, Wessels L, et al. Concordance among gene-
expression–based predictors for breast cancer. N Engl J Med 9. Liu R, Wang X, Chen GY, et al. The prognostic role of a gene 2006;355:560-9.
signature from tumorigenic breast-cancer cells. N Engl J Med Copyright 2007 Massachusetts Medical Society. The Mammogram That Cried Wolfe
That mammographic density is an important risk detected breast cancer was 3.5 in women with factor for breast cancer was first recognized by extensive density as compared with women who Wolfe in the 1970s. His pioneering observation had density in less than 10% of the breast. The has since been confirmed in more than 42 stud- masking effect greatly increased the odds of ies, the vast majority of which have shown an a cancer detected by nonscreening methods in association between increased mammographic women with extensive density as compared with density and the risk of breast cancer.1 Women in those with density in less than 10% of the breast the highest quartile of mammographic density — odds ratio, 17.8.
have a risk of breast cancer that is approximate- The results of the study by Boyd et al. are simi- ly four to six times as high as that among women lar to those of the Breast Cancer Surveillance Con- of similar age who are in the lowest quartile. sortium, which reported that the rate of screen- Only two other factors increase the risk of breast detected breast cancer (defined as breast cancer cancer more than mammographic density: age detected within 12 months after a positive screen- and mutations in the breast cancer–susceptibil- ing examination) is 2.5 times as high in women 40 to 49 years of age whose mammographic-den- Mammographic density is a function of the sity category was extremely dense as in women abundance of epithelial and connective tissue in in the category called “almost entirely fat,” accord- the breast, but a cancer and these normal tissues ing to the American College of Radiology Breast can have a similar radiographic attenuation, which Imaging Reporting and Data System (BI-RADS) can make both appear radiodense or white on a (Table 1). The rate of breast cancer not detected mammogram. By contrast, fat is radiolucent or by screening mammography (defined as breast dark on a mammogram.2,3 Therefore, it is possi- cancer detected within 12 months of a negative ble that the risk associated with mammographic screening examination) is 15 times as high in density is due to a masking effect — extensive women 40 to 49 years of age with a BI-RADS den- sity of extremely dense as in women whose mam- In this issue of the Journal, Boyd et al.4 sug- mogram is in the almost-entirely-fat category. gest that a masking effect is likely in the short Rates of cancer not detected by screening mam- term after mammography among women with mography among women with extensive breast density in 75% or more of the breast, as measured density increase with age and are highest in by qualitative or semiquantitative methods. They women 60 to 69 years of age. A large propor- calculated the odds of screen-detected breast can- tion of women 40 to 49 years of age who receive cer (defined as a breast cancer detected at the a diagnosis of breast cancer have extensive mam- time of screening mammography) and of breast mographic density, but the absolute risk of can- cancer detected by methods other than screening cer among women in this age group who have (defined as a breast cancer that was detected with- extensive density is relatively low — even lower in 12 months after a negative screening examina- than that among older women with extensive tion) in relation to the extent of mammographic density (Table 1).
density. They found that the odds ratio of screen- Boyd et al. also report an association between n engl j med 356;3 www.nejm.org january 18, 2007 T h e n e w e ng l a n d j o u r na l o f m e dic i n e Table 1. Breast Cancer Surveillance Consortium Rates of Breast Cancer on Screening Mammography, According to
BI-RADS Density Categories and Age, 1996 to 2003.*
Rate of Cancer
Rate of Cancer
Screening
BI-RADS Density
Screening Examinations
Detected†
Not Detected‡
* Data were provided by the Breast Cancer Surveillance Consortium (BCSC). The BCSC can be accessed at http:// breastscreening.cancer.gov/. BI-RADS denotes American Col ege of Radiology Breast Imaging Reporting and Data System. The BI-RADS categories are defined as almost entirely fat (<25% fibroglandular), scattered fibroglandular densities (25 to 50% fibroglandular), heterogeneously dense (51 to 75% fibroglandular), and extremely dense (>75% fibroglandular).
† Breast cancers were detected within 12 months after a positive screening mammographic examination.
‡ Breast cancers were detected within 12 months after a negative screening mammographic examination.
breast cancer and extensive mammographic den- and adjusted for age and race or ethnic group, is sity even when the density was observed as long as accurate as the more complex Gail model of as 8 years before a diagnosis of breast cancer, the National Cancer Institute.6 The risk model of thus verifying the results of a previous study.5 the Breast Cancer Surveillance Consortium shows This finding indicates that the association be- that assessment of breast density according to tween extensive mammographic density and an BI-RADS is a statistically significant addition to increased risk of breast cancer is due not only to the prediction of breast cancer in premenopaus- a masking effect but also to a biologic connec- al and postmenopausal women.7 An updated Gail tion between breast density and breast cancer. model that includes a semiquantitative measure More research is needed to understand how breast of breast density improves the discriminatory density affects susceptibility to breast cancer; power of the model, as compared with the origi- mammographic density is known to be influenced nal model and with models that use BI-RADS cat- by genetic factors that may increase susceptibili- egories of breast density.6‑8 Yet the ability to predict ty to breast cancer more in younger women than breast cancer accurately in an individual woman remains in question; all three risk models have Extensive mammographic density is present in only moderate predictive power, perhaps in part 25% of cases of breast cancer,4 which suggests because measures of breast density are imprecise.
that measures of mammographic density could Measures of density according to BI-RADS cat- be useful in assessing the risk of breast cancer egories of density are routinely included in mam- and in guiding measures to prevent breast can- mography reports to practitioners, but they are cer. Three models incorporate breast density into only moderately reproducible between observers an assessment of risk for breast cancer. A simple and between examinations rated by the same ob- prediction model for breast-cancer risk, based on server.9,10 Semiquantitative measures that outline the BI-RADS assessment of breast density alone the breast and draw with computer guidance the n engl j med 356;3 www.nejm.org january 18, 2007 dense areas on digitized mammograms are time intensive, have an element of subjectivity, and re- sity is strongly associated with increased suscepti- quire training, but they have better reproducibil- bility to breast cancer and decreased detection ity than the qualitative measure reported by Boyd of cancer by mammography. Measures of breast et al. In its current form, the semiquantitative density can be used in combination with other method has not been integrated into routine mam- risk factors to determine a woman’s risk of breast mography, because it requires acquiring, digitiz- cancer.7,8 Routine screening mammography could ing, and analyzing mammograms in a separate include an assessment of risk factors and a mea- laboratory. Automated measures of the volume surement of breast density, which together give of density are under development and hold great a woman and her physician an estimate of her promise in improving the accuracy and precision risk of breast cancer. Including risk assessment at of measures of breast density and their applica- the time of screening mammography could sub- stantially increase the identification of high-risk Postmenopausal women rarely are assessed women and provide an opportunity to consider for the risk of breast cancer with the use of the pharmacologic and nonpharmacologic means to Gail model, and they do not usually receive pre- reduce their risk. The time has come to acknowl- ventive therapy, even if they are at high risk for edge breast density as a major risk factor for breast breast cancer.12 A model for risk assessment that cancer and to determine, develop, and test the includes breast density that is used around the best ways to measure breast density in clinical time of screening mammography would create practice and use this measurement to maximize an opportunity to discuss with a woman her risk primary and secondary prevention of breast of breast cancer during the next five years and cancer.
to counsel her about prevention if she has a high No potential conflict of interest relevant to this article was re- risk. Clinicians could refer women with an esti- ported.
mated risk of breast cancer of 2% or higher for From the Departments of Medicine and Epidemiology and Bio- detailed risk assessment, counseling about modi- statistics and the General Internal Medicine Section, Depart- fiable risk factors, and the use of selective estro- ment of Veterans Affairs, University of California, San Francisco, gen-receptor modulators to reduce the risk of San Francisco.
1. McCormack VA, dos Santos Silva I. Breast density and paren-
Should women with increased mammographic chymal patterns as markers of breast cancer risk: a meta-analy- density be screened more often or with a different sis. Cancer Epidemiol Biomarkers Prev 2006;15:1159-69.
screening method than should other women? Dig- 2. Carney PA, Miglioretti DL, Yankaskas BC, et al. Individual
and combined effects of age, breast density, and hormone re- ital mammography seems to detect more breast placement therapy use on the accuracy of screening mammogra- tumors in women with dense breasts than does phy. Ann Intern Med 2003;138:168-75. [Erratum, Ann Intern film mammography.3 There is, however, no evi- Med 2003;138:771.] 3. Pisano ED, Gatsonis C, Hendrick E, et al. Diagnostic perfor-
dence that deaths from breast cancer are reduced mance of digital versus film mammography for breast-cancer in women who undergo digital mammography. screening. N Engl J Med 2005;353:1773-83. [Erratum, N Engl Nevertheless, early detection could provide ben- J Med 2006;355:1840.] 4. Boyd NF, Guo H, Martin LJ, et al. Mammographic density
efits similar to those of film-screen mammogra- and the risk and detection of breast cancer. N Engl J Med 2007; phy. If additional studies confirm that digital 356:227-36.
mammography detects more breast tumors in 5. Byrne C, Schairer C, Wolfe J, et al. Mammographic features
and breast cancer risk: effects with time, age, and menopause women with dense breasts than does film mam- status. J Natl Cancer Inst 1995;87:1622-9.
mography, clinicians with limited access to digi- 6. Tice JA, Cummings SR, Ziv E, Kerlikowske K. Mammographic
tal machines will have to refer women with dense breast density and the Gail model for breast cancer risk predic- tion in a screening population. Breast Cancer Res Treat 2005;94: breasts for digital mammography. As noted by 115-22.
Boyd et al., increasing the frequency of screen- 7. Barlow WE, White E, Ballard-Barbash R, et al. A prospec-
ing is not likely to influence the rate of cancer de- tive breast cancer risk prediction model among women under- going screening mammography. J Natl Cancer Inst 2006;98: tection among women with extensive density, 1204-14.
because the tumors are not visible, because the 8. Chen J, Pee D, Ayyagari R, et al. Projecting absolute invasive
tumors may grow quickly between examina- breast cancer risk in white women with a model that includes mammographic density. J Natl Cancer Inst 2006;98:1215-26.
9. Kerlikowske K, Grady D, Barclay J, et al. Variability and ac-
n engl j med 356;3 www.nejm.org january 18, 2007 T h e n e w e ng l a n d j o u r na l o f m e dic i n e curacy in mammographic interpretation using the American mings SR. Novel use of single x-ray absorptiometry for measur- College of Radiology Breast Imaging Reporting and Data System ing breast density. Technol Cancer Res Treat 2005;4:173-82.
(BI-RADS). J Natl Cancer Inst 1998;90:1801-9.
12. Kaplan CP, Haas JS, Perez-Stable EJ, et al. Breast cancer risk
10. Ciatto S, Houssami N, Apruzzese A, et al. Categorizing breast reduction options: awareness, discussion, and use among women
mammographic density: intra- and interobserver reproducibil- from four ethnic groups. Cancer Epidemiol Biomarkers Prev ity of BI-RADS density categories. Breast 2005;14:269-75.
11. Shepherd JA, Herve L, Landau J, Fan B, Kerlikowske K, Cum-
Copyright 2007 Massachusetts Medical Society. The End of an Era in Otitis Research
Retrospective studies carried out in the 1950s, Pittsburgh area serving families across the socio- 1960s, and 1970s suggested an association be- economic spectrum. The middle-ear status of tween otitis media early in life and subsequent these patients was monitored by means of pneu- developmental impairments in children. Although matic otoscopy and, in most instances, tympa- these studies were not designed to establish a nometry at least monthly until 3 years of age in cause-and-effect relationship, many prominent order to document the presence of a middle-ear physicians, audiologists, and speech pathologists effusion and identify a subgroup of patients who thought that a cause-and-effect relationship did met specified criteria for the insertion of tympa- exist, because severe, bilateral sensorineural hear- nostomy tubes. Children in the randomized trial ing loss was known to cause irreversible devel- underwent extensive developmental testing at opmental impairments. Therefore, aggressive in- 3, 4, 6, and 9 to 11 years of age. During the course tervention to restore normal hearing became the of the follow-up period, the domains assessed standard of care, and the placement of tympa- included parental stress and the children’s behav- nostomy tubes became the second most frequent ior, psychosocial development, language, speech, surgical procedure performed in the United States intelligence, attention, and academic achievement. (after neonatal circumcision). In July 1994, the Among the study participants, 429 children with Agency for Health Care Policy and Research (now bilateral middle-ear effusion lasting for 90 days the Agency for Healthcare Research and Quality) or unilateral middle-ear effusion lasting for 135 published a clinical practice guideline that rec- days were randomly assigned to receive tympa- ommended the insertion of tympanostomy tubes nostomy tubes promptly or up to 9 months later when bilateral middle-ear effusion had persisted if the effusion persisted. If a causal relationship for 4 to 6 months with a hearing threshold of exists between persistent effusion and develop- 20 dB or higher in otherwise healthy children mental impairments, the test scores of the chil- dren receiving tympanostomy tubes earlier should The article by Paradise et al. in this issue of have been better than the scores of the children the Journal2 marks the final installment of reports with tubes placed later in life.
from a longitudinal study that have led to a re- In addition, a representative sample of 241 appraisal of this practice. Recognizing the defi- children selected from those who were ineligi- ciencies of previous retrospective studies, Para- ble for randomization underwent similar devel- dise and colleagues designed a prospective study opmental assessments. This observational com- to answer three key questions. First, is there an ponent of the study was included to determine association between early otitis media and later whether the association between early otitis media impairments of speech, language, and cognitive and later developmental impairments remained development? Second, if an association exists, is when the effects of potentially confounding dem- it a cause-and-effect relationship? Third, if devel- ographic variables were considered.
opmental impairments result from early otitis In previous reports on assessments at 3, 4, and 6 years of age, no significant differences were From June 1991 through December 1995, the found in the test scores for any of the outcomes study enrolled 6350 healthy infants who were between the children who received tympanos- 2 to 61 days of age from eight sites in the greater tomy tubes early in life and those who received n engl j med 356;3 www.nejm.org january 18, 2007

Source: http://to07.endcancer.ca/site/DocServer/NEJM_editorial_Jan_18_2007.pdf?docID=2082