`
`MEDICAL PROGRESS - HARRIS ET AL.
`
`319
`
`REVIEW ARTICLES
`
`(� �� -M�E_n_ic_AL�-PR-OG�-RES�S���-J
`(First of Three Parts)
`
`BREAST CANCER
`
`JAY R. HARRIS, M.D., MARC E. LIPPMAN, M.D.,
`UMBERTO VERONESI, M.D.,
`AND WALTER WILLETI, M.D., DR.P.H.
`
`fourth decade and become substantial before the age
`
`
`
`of 50, thus creating a Jong-lasting source of concern
`
`for women and a need for vigilance. After menopause,
`
`the incidence rates continue to increase with age, but
`
`Jess dramatically than before. Breast cancer is the
`leading cause of death among American women who
`are 4-0 to 55 years of age.3 In less affluent parts of the
`world and in the Far East, the same pattern of in
`
`crease with age is seen,4 but the absolute rates are
`
`much lower at each age. In Japan, for example, the
`
`overall incidence of breast cancer has been only about
`one fifth that in the United States.�
`The rates of breast cancer have been steadily in
`creasing in the United States since formal tracking of
`
`cases through registries began in the 1930s (Fig. I).
`BREAST cancer is a major public health problem
`Between 194-0 and 1982, the age-standardized
`inci
`
`
`of great interest and importance to physicians in
`dence rose by an average of 1.2 percent per year in
`
`a variety of specialties. Since this topic was last re
`
`viewed in the journal, 1 the incidence of the disease has
`
`
`Connecticut, which has the oldest cancer registry in
`
`continuous operation.6 Improvements in the thor
`
`
`increased dramatically, heightening concern among
`
`oughness of the registry, whose coverage became vir
`
`physicians and womtn in general. In addition, long
`tually complete in the early 1970s, 7 are unlikely to
`
`
`term results are now available from clinical trials initi
`ated in the 1970s and 1980s to evaluate the
`account for more than 25 percent of the increase that
`usefulness
`occurred before 1982. Between 1982 and 1986, the
`
`of early detection with mammography and physical
`examination,
`incidence in the United States rose more sharply, at
`
`breast-conserving treatment with limit
`
`
`ed breast surgery and irradiation, and adjuvant sys
`4 percent per year.6 The time trends seen in Connecti
`
`temic therapy with hormonal therapy and chemo
`
`cut appear to reflect the experience in other parts of
`
`therapy. Furthermore, in the light of newly gained
`the United States, for which only recent data are avail
`
`
`
`knowledge, new strategies for addressing this problem
`able. Increases have occurred among all age groups
`
`since 1935, although the magnitude of the increase has
`have been proposed.
`In this review, we describe the recent trends in in
`
`
`been greatest among older women.8 Age-adjusted inci
`
`
`cidence and mortality and the epidemiologic features
`
`dence rates of breast cancer have increased in parallel
`
`that may be responsible for the rise in incidence.
`among black and white women in the United States
`
`since 1975; rates among postmenopausal black women
`
`We summarize the evidence evaluating the strategies
`remain about 15 percent lower than those among post
`
`for diagnosis and therapy initiated in the 1970s and
`1980s, including their benefits and costs. Finally,
`menopausal white women, but the rates among pre
`
`we describe the prospects for prevention and for
`
`menopausal black women are now slightly higher than
`more specific treatments based on evolving biologic
`those among white women.2 As in the United States,
`
`
`long-term increases in the incidence of breast cancer
`knowledge.
`
`are being observed worldwide, in both industrialized
`1'RENDs
`
`and developing countries.9•10
`
`
`The age-adjusted mortality rates for breast cancer,
`Breast cancer is a major affliction of women in afflu
`
`ent countries. On the basis of incidence rates for 1983
`
`in contrast to the incidence rates, have been remark
`
`ably stable in the United States (Fig. 1 ). However, the
`
`through I 987 and mortality rates for 1987 in the Unit
`12 percent of all women will be given a
`ed States,2•3
`time trends appear to vary depending on the age at
`
`diagnosis of breast cancer and 3.5 percent will die of
`
`diagnosis; since 1950 mortality rates
`have increased
`by about 15 percent among women over the age of 55
`
`the disease. The impact of breast cancer is magnified
`because women are at risk from their middle to later
`and declined by about the same amount among those
`
`years. The incidence rates increase rapidly during the
`younger than 45.11 The declining mortality among
`
`younger women appears to be best characterized as
`applying to women born after about 1935 in Connecti
`From the Departments
`of Radiation Oncology, Beth Israel Hospital and the
`cut and after about I 950 nationwide.12 Since 1975 the
`
`Dana-Farber Cancer lnstitu1c, and the Joint Center for Radiation Therapy. Har
`
`mortality rates among black women have increased
`varo Medical School, Boston (J.R.H.);
`the Vincent T. Lombatdi Cancer Research
`Centtt and the Depanmcnts
`of Medicine and Pharmacology, Georgetown
`
`
`substantially and are now slightly higher than those
`Uni
`versity Medical Center, Washington, D.C. (M.E.L.}; the lstituto Nazionale per
`
`for white women.13 The relative constancy of the over
`lo Studio e la Cura dci Tunlori, Milan, Italy (U.V.); and the Departments of
`.Epidemiology and Nutrition, Harvard School of Public Heald! and the Channing
`
`
`
`all mortality rate, despite increases in incidence, could
`Laboratory, Departments of Medicine, Harvard Medical School and Brigham
`
`be the result of more complete reporting of incident
`Boston (W.W.}. Address reprint requests to Dr. Harris
`and Women's Hospital,
`
`
`cases, increases in a more benign form of disease, ear
`at the Harvaro Joint Center for Radiation Therapy, SO BiMey St., Boston,
`MA 0211S.
`
`
`lier detection, or advances in treatment. These factors,
`
`IN INCIDENCE AND MORTALITY
`
`The New England Journal of Medicine
`Downloaded from nejm.org at REPRINTS DESK INC on August 17, 2015. For personal use only. No other uses without permission.
` Copyright © 1992 Massachusetts Medical Society. All rights reserved.
`
`1 of 10
`
`BI Exhibit 1040
`
`
`
`THE i"EW ENGLA'.'ID JOURNAL OF MEDICINE
`
`] uly 30, 1992
`
`320
`
`-gs 100
`l'<lo Q) 0
`g ci 80
`Q)o
`"O,...
`�� 60
`... Q)
`�iii
`c: ..!:::.
`J� 40
`� iii (/) t:: l'O 0 20
`!!? :E
`ID
`
`ter; the incidence rate of tumors measuring 2 cm or
`more has not changed appreciably.6 I n addition, the
`proportion of cases diagnosed while the tumor is in
`situ or localized increased substantially,6 after having
`been stable during the I 970s.11 These findings as well
`as an improved two-year survival rate are compatible
`with the concomitant substantial increase in the use of
`screening mammography.6 To the extent that the re
`cent acceleration in the incidence of breast cancer rep
`resents the transient rise expected in the early stages
`of a screening program, it will eventually result in the
`prevention of deaths due to breast cancer during this
`decade. However, the incidence of larger tumors and
`those with regional or distant metastases at diagnosis
`has not decreased,6 which would be expected if a
`screening program was implemented and the true in
`cidence was constant. This indicates that the under
`lying long-term increase in the incidence of breast
`cancer has continued through the 1980s and suggests
`that no major decline in mortality rates should be ex
`pected in the near future. Stable mortality rates in the
`face of an apparent true increase in incidence suggest
`that the earlier detection of cases in more recent years,
`and possible improvements in treatment, have im
`proved survival sufficiently to offset the rising inci
`dence.
`Although the very recent surge may be due largely
`to the increased use of mammographic screening, the
`much larger increase over the past half century ap
`pears to be real. Breast cancer is clearly continuing to
`increase, especially among postmenopausal women,
`and will require even greater attention on the part of
`researchers and clinicians. In particular, specific fac
`tors that explain the long-term increase should be
`sought.
`
`RISK FACTORS
`Large variations in the rates of breast cancer among
`countries� and over time within countries10 and large
`increases in the rates of breast cancer among popu
`lations migrating from nations with a low incidence
`to those with a high incidence18 indicate the existence
`of major nongenetic determinants of breast cancer
`and the potential for prevention. The elucidation
`of specific risk factors for breast cancer is important
`to understand the observed variation among and with
`in countries, to identify women who could benefit
`from intensified surveillance or prophylactic treat
`ment, to select subjects for participation in interven
`tion studies, and to modify factors that will ultimately
`reduce risk.
`The strength of a risk factor is typically indicated by
`its relative risk - the incidence among persons pos
`sessing a characteristic in question divided by the inci
`dence among otherwise similar persons without the
`characteristic. The relation of a risk factor to the dis
`ease, however, can be complex for a number of rea
`sons. Many risk factors are measured as continuous
`variables (for example, the age at which breast cancer
`was diagnosed in a relative and the ages of women at
`
`Mortality
`
`a...._����������������....-.-.
`1�1M1�1�1�1e1m1�1�1�
`Figure 1. Age-Standardized Incidence of Breast Cancer and Mor
`tality Rates in Co nnecticut from 1940 to 1988.
`The data are from the Surveillance, Epidemiology, and End
`Results Program (Miller B: personal communication).
`
`all of which appear to be contributing tO the diver
`gence of incidence and mortality, are discussed subse
`quently.
`Whether the increase in the incidence of breast can
`cer has been the result of more widespread use of
`screening mammography has been examined in sever
`al analyses. The initiation of a screening program will
`temporarily increase the incidence by advancing the
`time of diagnosis, as was noted nationally in 1974
`through 1976 (Fig. 1). If screening is not repeated, a
`deficit of incident cases will ensue; if screening is per
`formed regularly, a new steady-state incidence will be
`achieved at a rate close to that which will occur with
`out screening. The number of breast cancers diag
`nosed in screening programs that would not eventual
`ly be recognized clinically appears to be small; there is
`minimal underdetection of breast cancer in autopsy
`series, 12 no excess incidence in a I 0-year period was
`seen in a randomized screening trial, 14 and little in
`crease was seen among women undergoing mammog
`raphy for routine screening in a national program for
`the detection of breast cancer.15 In an Oregon prepaid
`health plan, only 9 percent of cases diagnosed in 1985
`were initially detected by screening mammography,
`and it was estimated that screening could account for
`no more than 5 percent of incident cases.16 However,
`most of the increase between 1960 and 1985 was ac
`counted for by tumors with estrogen receptors, sug
`gesting a hormonal influence and the possibility that
`the increase may be due to a more benign form of
`breast cancer. In the United States as a whole, the
`annual rate of screening mammography among wom
`en over the age of 50 years did not appear to exceed 15
`percent in 1984.'7 Because screening causes at most a
`transient rise in incidence and because its use was not
`widespread at least through the early 1980s, it can
`explain little of the long-term increase in the incidence
`of breast cancer.
`The upsurge in the incidence of breast cancer that
`began in the early 1980s is almost entirely due lo an
`increase in tumors measuring less than 2 cm in diame-
`
`The New England Journal of Medicine
`Downloaded from nejm.org at REPRINTS DESK INC on August 17, 2015. For personal use only. No other uses without permission.
` Copyright © 1992 Massachusetts Medical Society. All rights reserved.
`
`2 of 10
`
`BI Exhibit 1040
`
`
`
`Vol. 327 :-<o.
`5
`
`MEDICAL PROGRESS - Hi\RRIS ET /\L.
`
`321
`
`Table 1. Established and Probable Risk Factors for Breast Cancer.
`
`COMPARJ.SON
`CATF.(X) ltY
`
`ft1$K CA TEOOKY
`
`TYPICAi.
`RtLATlVI!.
`RISK
`
`STUDY
`
`16 yr
`
`Before 20 yr
`
`45-54 yr
`
`RISK FACTOJt
`
`Family his1ory of
`breast cancer
`
`Age a1 menarche
`
`Age at birth of
`l<t child
`
`Age a1 menopause
`
`Benign breast
`disease
`
`Radia1ion
`
`Obesi1y
`
`Height
`
`menarche, thf' hirth of thf' first
`child, and menopause), and their
`relative risks can be quite arbitrary,
`depending on the segments along
`the continuum that are compared.
`To evaluate the potential causes
`of breast cancer and the reasons
`for the international differences,
`comparisons of extremes are of
`ten of interest, such as an age of
`11 years at menarche as compared
`with an age of 16 years. From a
`clinical perspective, however, the
`group with the highest risk on the
`basis of any particular factor is usu
`ally of primary interest; the relative
`risk for this group as compared with
`that for the rest of the population
`will typically be much smaller than
`when it is compared with the group
`with the lowest risk. Furthermore,
`the risk for an individual woman
`cannot be determined by multiply
`ing the relative risk by the average
`risk for the population because the
`general population includes per
`sons with and without the risk fac
`tor. In addition, the occurrence of
`an elevated risk in association with
`a given factor does not necessarily
`imply causation; however, this in
`formation may still be useful for
`prediction.
`A number of variables that pre
`dict the occurrence of breast cancer
`and their typical relative risks are
`described briefly in Table l. As can
`be appreciated, the established risk factors for breast
`cancer - a family history of breast cancer, early men
`arche, late age at first childbirth, late age at meno
`pause, history of benign breast disease, and exposure
`to ionizing radiation - are generally associated with
`only weak or moderate elevations in risk. The excep
`tions occur in uncommon subgroups of these vari
`ables; for example, a family history of breast cancer at
`a young age or a family history of bilateral disease.31•32
`A family history of breast cancer, particularly when
`the diagnosis was made in the mother or a sister at a
`young age, can be an important risk factor for breast
`cancer.33 As compared with the risk among women
`having no first-degree relatives with breast cancer,
`overall the relative risk is on the order of 1.5 to 2 for
`women who have one first-degree relative with breast
`cancer34 and may be as high as 4 to 6 for those with
`two affected first-degree relatives.19 The risks are
`heightened if the cancer was bilateral.31•32 For a wom
`an with a sister who had bilateral breast cancer before
`the age of 50, the lifetime cumulative risk of breast
`cancer appears to be greater than 50 percent, and it is
`even higher if the sister was affected before the age of
`
`No 1st-degree
`Mother affected before
`1he age of 60
`relatives
`affected
`Mo1her affected afler
`the age or 60
`rela-
`Two 1st-degree
`tives affected
`II yr
`12 yr
`13 yr
`14 yr
`15 yr
`20-24 yr
`25-29 yr
`;;.30 yr
`Nulliparous
`After 55 yr
`Before 45 yr
`Oophorectomy before
`35 yr
`No biopsy or
`Any benign disease
`1.5
`
`Proliferation only
`aspiration
`2.0
`4.0
`A1ypical hyperplasia
`
`A1omic bomb (I 00 rad) 3.0
`No special
`Repealed Huoroscopy
`exposure
`1.5-2.0
`I 0th pcreentile
`9Qth percentile:
`Age, 30-49 yr
`Age, ;;.50 yr
`I 0th percentile
`9Qth percentile:
`Age, 30-49 yr
`Age, ;.50 yr
`Current uset
`Pas1 uset
`Current use all ages
`Age, <55 yr
`Age, 50-59 yr
`Age, ;;.6Q yr
`Pas1 use
`I drink/day
`2 drinks/day
`3 drinks/day
`
`Nurses' Heallh Study•
`
`Nurses' Health Study•
`Gail ct ar.1•
`
`Kampen cl al. 20
`
`Whi1e"
`
`Trichopoulos et al.22
`
`Willen c t al."
`Dupont and Page24
`Dupont and Page,.
`Boice and Monson1s
`McGregor et al. 26
`Tretli21
`
`Tretli"
`
`Romicu ct al. 28
`Colditz c1 al."'
`
`Longnecker et al."°
`
`2.0
`
`1.4
`
`4-6
`
`1.3
`1.3
`1.3
`1.3
`I.I
`1.3
`1.6
`1.9
`1.9
`1.5
`0.7
`0.4
`
`0.8
`1.2
`
`1.3
`1.4
`1.5
`1.0
`1.4
`1.2
`1.5
`2.1
`1.0
`1.4
`1.7
`2.0
`
`Never used
`Oral contracep1ive
`use
`Postmenopau.sal
`estro ..
`Never used
`gen-replaceme111
`1hcrapy
`
`Alcohol use
`
`Nondrinker
`
`•unpublished pzospectivt ditit were obtained from Gnham Col-dill (pcl"S()nal c:ommunication).
`tRdaove rists m.a)' be hightr for \Ao'(>men �iven a d1agnosi� of brcas1 <.'an'-"Ct t-:fore the age of 40.
`
`40.31 The excess relative risk declines with the age of
`the relative at the time of diagnosis.33•35 For a woman
`whose mother had unilateral breast cancer after the
`age of 60, the excess relative risk is only about 40
`percent greater than that associated with having
`no first-degree relatives with breast cancer (Nurses'
`Health Study: unpublished data). An intensive search
`for DNA markers of familial risk is ongoing and wiU
`be described later.
`Early menarche is a well-established but weak risk
`factor. 20 The relative risk is approximately 1.2 for
`women in whom menarche occurred before the age of
`12 as compared with women in whom it occurred at
`the age of at least 14.17 However, this variable may
`account for a substantial part of the international dif
`ferences, because the contrasts are more substantial;
`in China the average age at menarche is 17 years,36 as
`compared with 12.8 years in the United States.37
`Nulliparity and a late age at first birth both increase
`the lifetime incidence of breast caacer.21•38 The risk of
`breast cancer among women who have their first child
`after the age of 30 is about twice as high as that among
`those who have their first child before the age of 20;
`
`The New England Journal of Medicine
`Downloaded from nejm.org at REPRINTS DESK INC on August 17, 2015. For personal use only. No other uses without permission.
` Copyright © 1992 Massachusetts Medical Society. All rights reserved.
`
`3 of 10
`
`BI Exhibit 1040
`
`
`
`322
`
`THE NEW ENGLAND JOURNAL OF MEDICINE
`
`July 30, 1992
`
`sociated with increases in risk irrespective of when
`
`
`women who have their first child after the age of 35
`
`they were used. The use of postmenopausal estrogen
`
`
`have a slightly higher risk than nulliparous women.33
`supplements appears to increase the risk of breast
`An earlier age at the birth of a second child further
`cancer by about 40 percent among women who are
`
`reduces the risk of breast cancer.39 After an adjust
`
`actively taking them,29 with little increase among
`ment for the ages of the women at the births of their
`those who are no longer taking them.53 This increased
`
`children, the number of births has at most a small
`risk among current users appears to be concentrated
`
`influence on the risk of breast cancer.39•40 Although
`among older women, who also tend to take them for
`pregnancy before the age of 30 reduces the lifetime
`
`longer periods. Combining progesterone with estrogen
`risk of breast cancer, recent evidence suggests a more
`
`replacement, which reduces the risk of endometrial
`complex pattern of a transiently increased risk relative
`cancer, does not appear to decrease the incidence of
`
`to that for a nulliparous woman that lasts for one to
`
`breast cancer, and may add to it.54 Alcohol consump
`
`two decades, followed by a risk that is lower than that
`tion, even at the level of about one drink per day, has
`
`for a nulliparous woman later in life.
`
`been associated with a moderate increase in risk in
`
`A late age at menopause increases the risk of breast
`
`most, but not all, case-control and cohort studies.30•55
`
`cancer; the incidence is doubled among women with
`
`
`As for the more traditionally recognized risk factors
`natural menopause after the age of 55 as compared
`with those in whom it occurs before the age of 45.22•41
`
`described previously, the magnitude of associations
`
`between these less well-established variables and the
`
`In the extreme, women with bilateral oophorectomy
`risk of breast cancer is not strong.
`before the age of 35 had one third the risk of women
`
`Other potential risk factors have been studied, but
`with natural menopause in studies conducted before
`
`the findings have been inconclusive. The fat composi
`
`hormone-replacement therapy became standard prac
`tion of the diet has been thought to influence the risk
`tice. 41
`of breast cancer, in great part because of the large
`A history of benign breast disease has long been
`
`
`differences in rates between countries.10 However,
`known to increase the risk of breast cancer slightly.
`
`
`only weak56•57 or nonexistent23·!>8·59 associations have
`However, the term "benign breast disease" covers a
`
`In anibeen seen in case-control and cohort studies.
`
`
`
`
`heterogeneous group of histopathologic entities and
`
`mals, mammary tumors appear to be most strongly
`
`needs to be defined specifically.24•42 As compared with
`
`promoted by linoleic acid (the primary dietary poly
`women without a history of breast biopsy or aspira
`
`
`unsaturated fat) and inhibited by n-3 marine oils60;
`tion, women who have lesions with any proliferative
`
`however, there is little evidence that these fats are
`
`epithelial changes have twice the risk of breast cancer
`related to breast cancer in humans. An inverse rela
`
`and those with atypical hyperplasia about four times
`
`tion between breast cancer and the total intake of vita
`
`
`the risk.24•42 Lesions without proliferative changes are
`
`min A has been observed in some studies,61•62 but the
`
`associated with little or no excess risk. Four to 10 per
`
`
`validity of this finding is far from resolved. Lactation
`
`cent of benign biopsy specimens show atypical hyper
`21,12
`has been found to reduce the risk among premeno
`plasia.
`
`pausal women in some studies,63•64 but not in other
`Exposure to ionizing radiation, particularly be
`
`
`
`
`large investigations.65•66 Participation in varsity athlet
`tween puberty and the age of 30, can substantially
`
`
`ics was associated with reduced risk in one study,67 but
`
`increase the risk of breast cancer.25•26 However, expo
`not in another.68
`
`sure to clinically important levels is rare.
`To convey the effect of various risk factors in combi
`Obesity is not an important risk factor for breast
`
`nation, Gail and colleagues 19 have compiled detailed
`
`cancer, and among premenopausal women it is actual
`tables of estimates of the cumulative incidence of
`
`
`ly associated with a reduced incidence.27•43 Among
`
`
`breast cancer among women at specific ages and ac
`
`postmenopausal women, it has a weak but clinically
`
`
`cording to the number of first-degree relatives with
`
`
`
`
`unimportant positive association with the incidence of
`breast cancer, age at menarche, age at first live birth,
`
`breast cancer, but it has a stronger association with
`
`
`and number of biopsies for benign breast disease. For
`
`
`mortality from breast cancer, due in part to delayed
`example, the cumulative 30-year incidence of breast
`among more obese women44 and to a worse
`diagnosis
`
`cancer for a 50-year-old woman would be approxi
`
`
`prognosis that is independent of the stage of cancer.45
`mately 20 percent if she had her menarche at the age
`
`Other features have been associated with breast
`
`of 11 years, had two first-degree relatives with breast
`
`
`cancer, but they are not as firmly established as those
`
`cancer, and delivered her first child after the age of 30.
`
`noted above. Tallness is associated with an increased
`
`If she had no first-degree relatives with breast cancer,
`
`risk of breast cancer internationally46 and in numer
`
`her risk would be approximately 9 percent.
`
`
`ous case-control and cohort studies.45•4749 The use of
`
`The accumulated data on risk factors for breast can
`
`oral contraceptives appears to increase the risk of
`
`cer suggest several biologic mechanisms. Genetic fac
`breast cancer by about 50 percent, but the excess risk
`
`tors clearly contribute, and a search is now in progress
`
`drops rapidly after the drug is stopped,28·� suggest
`
`for DNA mutations associated with this increased risk.
`
`
`
`ing a late-stage tumor-promoting effect. However, is
`
`
`
`
`Estrogenic stimulation increases the risk69; the elevat
`
`sues related to their use early in reproductive life re
`
`ed risk among users of estrogen supplements29 sup
`
`
`main unsettled; in several recent case-control studies
`
`
`ports this mechanism most directly, and the effects of
`
`among women younger than 45 years,51•52 the use of
`age at menarche and menopause, obesity among post-
`
`oral contraceptives for more than a few years was as-
`
`The New England Journal of Medicine
`Downloaded from nejm.org at REPRINTS DESK INC on August 17, 2015. For personal use only. No other uses without permission.
` Copyright © 1992 Massachusetts Medical Society. All rights reserved.
`
`4 of 10
`
`BI Exhibit 1040
`
`
`
`Vol. 327 :-lo. S
`
`MEDICAL PROGRESS - HARRIS ET AL.
`
`323
`
`with the disease. For example, in the Nurses' Health
`
`menopausal women, and the therapeutic effect of ta
`
`Study, women in whom menarche occurred before the
`
`moxifen therapy70 are also likely to be mediated by this
`age of 11, who had their first child after the age of 35,
`mechanism. Studies of endogenous estrogen levels in
`
`
`who had a history of benign breast disease, or who
`
`to the risk of breast cancer are currently in
`relation
`
`had a history of breast cancer in a first-degree relative
`
`
`conclusive because of the possibility that the levels
`
`composed 41 percent of the population and together
`may be influenced by the disease in case-control
`
`had only a 54 percent greater incidence of breast can
`
`
`studies,71 •72 the limited size of prospective studies,
`73•74
`cer than did the remaining women (unpublished
`
`and the poor reproducibility of many serum hormone
`
`
`data). Furthermore, the excess incidence in the study
`assays. Another mechanism is suggested by the find
`
`
`population accounted for by these variables was only
`ing that pregnancy early in life reduces the lifetime
`18 percent. A small group of women, those with a
`
`risk of breast cancer; the protective effect may result
`
`mother or sister who has had bilateral breast cancer at
`
`
`from pregnancy-induced differentiation of breast stem
`
`
`a young age or multiple first-degree relatives with
`cells.
`
`
`7; Finally, restriction of food intake early in life,
`breast cancer at a young age, may have cumulative
`
`
`which profoundly reduces the incidence of mammary
`
`lifetime risks of 30 percent or more. These women
`
`tumors in animals,60•76 may also be relevant to hu
`
`
`warrant particularly careful follow-up by physicians
`
`
`mans. This relation is reflected in the positive associ
`
`
`experienced in breast disease, but they account for a
`
`ation between height and the risk of breast cancer,4;
`small fraction of all breast cancers. Unfortunately,
`
`and may underlie many of the differences in the rates
`
`even women without identifiable risk factors have an
`among countries.46•77
`
`
`appreciable lifetime risk of breast cancer (approxi
`
`Can the established risk factors for breast cancer
`mately 6 percent through the age of 80),19 and they
`
`
`account for the substantial increase in the incidence of
`
`will benefit from regular screening for breast cancer.
`
`breast cancer over the past 40 years? The age at men
`
`
`From the standpoint of identifying risk factors to
`arche has declined from an average of about 17 years
`ago to an average of 12.8 years, but it
`prevent breast cancer, our knowledge is even more
`two centuries
`
`
`
`disappointing. It is either impossible or culturally un
`has been stable in the United States since the l 940s.37
`
`acceptable to modify some of the clearly established
`
`Adult height has increased substantially over the past
`
`risk factors. Although great strides have been made in
`
`150 years in the United States, but it also tended to
`
`
`that are risk the identification of lifestyle variables
`
`stabilize sometime about 1940 among the middle and
`
`factors for cardiovascular disease and some forms of
`
`
`upper classes.78•79 Thus, to the extent that the im
`
`cancer, this paradigm may not necessarily apply to
`
`provements in childhood nutrition reflected by the age
`breast cancer. The search for modifiable risk factors
`
`at menarche and ultimate height adversely influence
`
`has not been exhausted and must continue. However,
`the risk of breast cancer, cohorts of women born be
`
`to the extent that the high incidence of breast cancer
`fore about 1940 will continue to have successively
`
`in affluent countries is the result of rapid growth and
`
`higher age-specific rates, but those born after this time
`
`
`early maturation of children resulting from historical
`should have little further increase.
`
`
`ly unprecedented nutritional abundance and the con
`Changes in the age at which women bear children
`
`
`
`trol of infectious disease, the lifestyle changes needed
`explain little of the long-term increases in breast
`
`to reduce the risk of breast cancer substantially may
`cancer, although recent delays in the time of first
`
`not be feasible. If this is the case, prevention may
`
`pregnancies could increase future rates by about
`
`
`9 percent.11•21 Widespread use of estrogen-replace
`
`
`depend on artificial manipulation of hormones and
`
`
`growth regulators that underlie the known risk predic
`
`
`ment therapy has almost certainly contributed to
`tors, such as a woman's age at the birth of her first
`
`
`the higher incidence among postmenopausal women.
`child and at menopause.
`
`
`Some have claimed that increased fat consumption is a
`
`
`probable explanation for the rise in incidence, 10 but
`SCREENING
`
`
`this assertion is based on data for fat production rather
`
`One potentially important strategy in reducing the
`
`than intake; fat intake has actually been declining in
`
`
`mortality from breast cancer is earlier detection. Earli
`
`the United States for the past 40 years.80 Increased
`
`
`to result in treatment beer diagnosis is hypothesized
`alcohol consumption by younger women may have
`
`fore the tumor metastasizes and thus to avert death
`
`
`contributed appreciably if the observed association
`
`
`due to the disease. The main methods for earlier detec
`with incidence is causal; alcohol consumption at the
`
`tion of breast cancer have been mammography and
`age of 18 was three times higher among Nurses'
`
`physical examination performed by a trained health
`
`Health Study participants born between 1960 and
`
`
`professional. Other potential methods of screening,
`1964 than among those born 40 years earlier (unpub
`
`
`such as self-examination of the breasts, have not yet
`lished data). Although an increase in the incidence
`of
`
`
`been demonstrated to be of value,81 and some meth
`breast cancer would have been expected on the basis
`
`ods, such as thermography and CT scanning, have
`of changes in known and suspected risk factors,
`been shown not to be of value. The ability of mam
`
`whether these factors can quantitatively account for
`mography to detect cancers well before they are ap
`the observed increase remains unclear.
`
`parent on physical examination has been indisputably
`The known risk factors for breast cancer do not col
`
`
`established. The usefulness of mammography in re
`
`
`lectively allow the identification of a small high-risk
`
`cent years has been enhanced by technical advances
`
`group that accounts for a large proportion of women
`
`The New England Journal of Medicine
`Downloaded from nejm.org at REPRINTS DESK INC on August 17, 2015. For personal use only. No other uses without permission.
` Copyright © 1992 Massachusetts Medical Society. All rights reserved.
`
`5 of 10
`
`BI Exhibit 1040
`
`
`
`THE"'�:\\' E:\GLAl"D .JOCRNAI. OF �1EDICl:\E
`
`.Jul} :io. 1992
`
`that provide increased visualization of' th<' breast pa
`ly the effect of scree ning 011 breast-cancer mortality
`eliminate these sources of bias. These trials,111•811 listed
`renchyma (and reduce exposure
`to radiation), im
`in Table 2, are not simple to summarize because of
`provements in film quality and proct'ssing, refined
`techniques of imaging (compression, reduction in the
`variations in study design, the technical
`level of the
`
`size of the focal spot, magnificatio n, and the ancillary mammography, the degree to which women in the
`
`use of ultrasonography), better guidelines for the di
`screening group and controls were actually screened,
`and the length of follow-up. (),·erall,
`agnosis of cancer, and greater availability
`howt>ver, screen
`of well
`trained mammographers. As illustrated in Figure 2,
`ing appears to reduce monalit\' from breast cancer by
`about 25 percent.89 Some haw arg ued that this is like
`d1C'se newer techniques detect a larg