Vol. 4. 567-571. Julv/Augi�.st /995 Cancer Epidemiology, Biomarkers & Prevention 567

Review

Models of Breast Cancer Show That Risk Is Set by Events of Early

Life: Prevention Efforts Must Shift Focus1

Graham A. Colditz2 and A. Lindsay Frazier

Channing Laboratory. Brigham and Women’s Hospital 1G. A. Cl; Harvard

Medical School [G. A. C., A. L. Fl: and Center for Cancer Prevention,harvard School of Public Health [G. A. C.]; and Pediatric Oncology, Dana-

Farber (ancer Institute IA. L. Fl. Boston, Massachusetts 021 15

Abstract

We have recently published a mathematical model of theetiology of breast cancer based on the data from theNurses Health Study that extends the Pike model ofbreast cancer (see Appendix). The most salient feature ofthe model is that it identifies the years before the firstbirth of a child as the most crucial in establishing futurerisk of breast cancer. The extended model includesseveral additional details of reproductive risk factors,allowing us to quantify the relative importance of theeach of the reproductive risk factors and to estimate theeffect of changes in key determinants of breast cancer. Inthis review, we present the evidence from both animalstudies and epidemiological research that corroborate thecritical importance of the exposures that occur beforefirst birth. We argue that research and preventiveinterventions should now focus on youth. Population-wideprevention strategies are necessary because the inheritedgenetic risk for breast cancer accounts for no more than10-15% of all breast cancer cases, leaving 85% of casesdiagnosed among women who are not in this high-risksubgroup of the population. An example of a population-based intervention would be the promotion of increasedphysical activity among young girls that could result inthe delay of menarche. An example of additional researchthat focuses on the importance of early life exposureswould be an analysis of the relation between diet andother lifestyle factors during adolescence and thesubsequent risk of breast cancer and studies of precursorlesions (atypical hyperplasia). Shifting the focus of breast

cancer prevention to this age group is urgently needed.

Important Aspects of the NHS Extended Modelof Breast Cancer

Key determinants of the risk of breast cancer for women are thetiming of the reproductive events in her life: age at menarche,age at first birth, number of children, and age at menopause.The relation between reproductive risk factors have been syn-

thesized into an hypothesized sequence for breast carcinogen-

esis by de Waard and Trichopoulos (1). The mathematical

Received I 0/7/94; revised 4/1 2/95: accepted 4/20/95.

I Supported by NIH Grants CA 40356 and K07CA62252-OlAl. G. A. C. has aFaculty Research award from the American Cancer Society.2 To whom requests for reprints should he addressed, at Channing Laboratory,

180 Longwood Avenue, Boston, MA 02115-5899.

model of breast cancer published by Pike et a!. (2) was basedon the observed age-incidence curve and the known relations

among age at menarche, first birth, and menopause; parity; and

the risk of breast cancer. However, it did not include terms for

the spacing of pregnancies, nor did it easily accommodate

pregnancies after age 40 years. To analyze the importance of

these events, we fitted an extension of the Pike et a!. (2-4)

model of breast cancer incidence to the Nurses’ Health Studydata from follow-up of 1976 to 1988 (5; see Appendix).

When we fit the Pike model to an independent prospective

data set (the prospective Nurses’ Health Study) and added aterm to summarize the spacing of births, we observed that

closer spacing of births was significantly related to reduced risk

of breast cancer. Another feature of our extended Pike ci a!. (2)

model of breast cancer incidence is a term for an increase in risk

associated with the first pregnancy. This term was significantonly for the first pregnancy not for subsequent pregnancies (5).

Thus, the breast is presumably protected against the effects ofcell proliferation (and the high hormone levels) during second

and subsequent pregnancies by the terminal differentiation that

occurs during first pregnancy. The increase in risk with first

pregnancy has been observed in a subsequent analysis of pro-

spective data from Sweden (6) and analysis from an interna-

tional case-control study (7).

Pike coined the phrase “breast tissue aging” to signify the

effects of the different variables in his model in determining

risk of breast cancer. In molecular terms, breast tissue aging is

the accumulation of molecular damage in the pathway to breast

cancer. In our extension of the Pike model, the time betweenmenarche and first birth had the highest rate of breast tissue

aging, consistent with our hypothesis that this time period wasthe time when the breast tissue was most vulnerable to mu-

tagenesis. We also observed in our model that the increase inrisk of breast cancer associated with first pregnancy is followed

by a 20% decrease in the rate of breast tissue aging. This

transient increase followed by a decrease in risk of breast

cancer explains the “cross-over” effect seen in certain sub-groups of women where initially a group has higher rates but,

usually around the age of menopause, the rates drop below thatof the other subgroup. For instance, using incidence data from

New York state, Janenich and Hoff (8) showed a cross-over in

breast cancer mortality between single and married women at

age 42, such that married women had higher incidence than did

unmarried women before this age and lower incidence of breast

cancer mortality after this age. A similar cross-over of mci-dence has been reported between black and white women in the

United States (9, 10), consistent with the distribution of age atfirst birth by race. Over many decades, black women in the

United States have had higher rates of pregnancy and an earlierage at first birth than white women (11). Whereas others have

argued that the cross-over effect is a result of differential

exposure to carcinogens (10), the model predicts these patterns

of incidence based solely on the reproductive variables.

Association for Cancer Research. by guest on October 26, 2020. Copyright 1995 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

568 Review: Breast Cancer Risk Set by Early Life Events

Support from Animal Models of the Importance of thePrepregnancy Period to Risk of Breast Cancer

The susceptibility of the rat mammary gland to carcinogenesisis entirely dependent on the age of the rat and the corresponding

stage of mammary gland development (12). In the rat, themammary gland results from progressive branching of the mainlactiferous duct. The bulbous ends of these branches are called

TEBs.3 The TEBs are formed by day 21, which marks thebeginning of menarche for the rodent. The TEBs are stimulatedwith each estrous cycle of the rat to divide and differentiate intoABs. Administration of a carcinogen during the time that TEBsare differentiating into ABs induces the greatest number ofintraductal carcinomas. Once this process of differentiation hasbeen completed, the incidence of tumors induced by the samedose of carcinogen decreases by >50%. The ABs coalesce into

a collection called a lobule. During pregnancy, these lobulesalso undergo a massive amount of cell proliferation. The majormorphological change is the sprouting of new ABs in each

lobule. In the human, the mean number of alveolar buds per

lobule is 1 1 before pregnancy, increasing to 80 AB/lobule bythe end of pregnancy.

Each step in the progression from TEB to AB to virginallobule to lactational lobule is accompanied by a lengthening of

the average cell-cycling time. For instance, the length ofthe cellcycle ofthe TEB is 13 h; once the cell has differentiated into anAB, the cell cycle is prolonged to 34 h. The difference is dueto longer time spent in G1. More time in G1 means more timefor repair of DNA and less likelihood of propagation of trans-formed DNA. In the rat, the amount of repair can be measured

by the removal of DNA adducts. TEBs remove fewer adductsthan do ABs. The lack of adequate time for DNA repair is a

biological explanation of the enhanced vulnerability of the TEBto carcinogenesis as compared to the AB or lobule.

Differentiation within the mammary gland is neither uni-form nor synchronous. Even in the parous rat, it is possible tofind TEBs that have not differentiated into ABs and lobules thathave a low density of ABs comparable to the density found inthe virginal breast. Each pregnancy “recruits” more of theremaining undifferentiated cells. In rats, administration of car-cinogen immediately before pregnancy produces a high number

of tumors; the presumption is that the short interval does notallow for adequate repair in the remaining undifferentiated cellsthat will undergo proliferation and differentiation due to thehormonal stimulus. However, administration of carcinogen af-

ter pregnancy and lactation induces very few tumors, presum-ably because of the protective effect of terminal differentiation.The implications for prevention are that the timing of the

exposure is key to its carcinogenic potential. Furthermore,closer spacing of children may be protective because of de-creasing the possible amount of time available for exposure tocarcinogen before the proliferative effect of next pregnancy.

Exposures during Adolescence and Early AdulthoodDetermine the Risk of Breast Cancer for Women

Age at the time of exposure has been shown to be pivotal indetermining the subsequent risk of breast cancer. In studieslinking breast cancer with irradiation, alcohol consumption, and

cigarette smoking, the younger the age at exposure, the greater

the risk.

3 The abbreviations used are: TEB. terminal end bud: AB. alveolar bud; RR,

relative risk.

Irradiation and Breast Cancer. Girls who undergo irradia-tion of the chest at a young age have an increased risk of breastcancer. This observation has been replicated in a number of

settings. Women who survived the atomic bomb in Hiroshima

or Nagasaki, Japan, have an increased risk of breast cancer,

dependent on both age and dose (13). Elevation in risk is seenwith doses as low as 0.5 Gy. The excess relative risk increases

with decreasing age; the excess relative risk was 5.3, 2.6, 1.2,

0.0, and 0.4, respectively, for women age 0-9, 10-19, 20-39,40-49, and over 50 years at the time of the bombing. The

dramatic relation of risk to age at exposure highlights thevulnerability of the developing breast to carcinogenic exposure.

Although risk of breast cancer among survivors of theatomic bomb is obviously the most extreme example of therelation between radiation and risk, several other examples of

excessive radiation and increased risk of breast cancer havebeen do�’ mented. The recurring theme is that younger age at

exposure confers greater risk of breast cancer. Girls with tu-

berculosis treated with repeated fluoroscopy, (14) girls diag-

nosed with Hodgkin’s disease treated at a young age with

ionizing radiation (15), and infants with “enlarged” thymuseswho underwent irradiation (16) all have an increased risk of

breast cancer dependent on age at exposure.

These examples of the risk of breast cancer after radiationunderscore unequivocally the importance of the age at which

the exposure occurs. However, the amount of radiation in eachcase far exceeds the amount received by a normal individual

during a lifetime. These examples, therefore, although illustra-

tive, do not provide insight into preventive measures available

at the population level.

Alcohol Consumption at an Early Age and Risk of Breast

Cancer. Multiple studies have linked the consumption of al-cohol to breast cancer (17). The risk shows a dose-response

relationship. The Nurses’ Health Study has examined the effect

of alcohol consumption among women who were otherwise

without risk factors for the development of breast cancer (hada full-term pregnancy by age 26 years and no history of benign

breast disease or family history of breast cancer). For womenwho drank 15 g of alcohol/day (>1 drink) the relative risk of

breast cancer was 2.5 when compared to women who neverdrank (18). The degree of risk conferred by this amount of

moderate alcohol consumption is actually greater than the riskconferred by having a family history of breast cancer (RR =

1.8; Ref. 19).Several studies have explored the effect of past drinking

habits on the risk of breast cancer. All have found a significantimpact of consumption at a younger age. Harvey et a!. (20)

conducted a case-control study in which they asked participants

to estimate consumption at three different age periods: <30,30-49, and >50 years. The entire elevation in risk associatedwith drinking was confined to the earliest time period. Anotherstudy observed that women who started to drink before age 25

years had a relative risk of breast cancer of 1.8 (adjusted forreproductive history, family history, and age) compared to

women whose first exposure to alcohol occurred after age 25

years (21). Hiatt et a!. (22) used the alcohol data collected on68,674 women during routine health care to study the 303women who subsequently developed breast cancer. The women

were asked to estimate the time in their lives when they drank

“the most.” Women who reported drinking the most before age30 years had nearly twice the relative risk (2.6) when compared

to women who reported drinking more in later periods of theirlives (1.4). In a third case-control study, consumption between

ages 18-35 years conferred a greater relative risk (RR = 2.2)

Association for Cancer Research. by guest on October 26, 2020. Copyright 1995 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Cancer EpidemIologj�, Biomarkers & Prevention 569

than drinking after age 35 years (RR = 1.8; Ref. 23). Obvi-ously, these findings need to be corroborated, but the consistent

implication that drinking at a young age adversely affects therisk of breast cancer warrants serious attention by those inter-

ested in prevention. If onset of drinking could be delayed, riskof breast cancer could potentially be markedly reduced. Addi-tionally, if we could identify factors that counteracted the effectof alcohol (e.g., a high consumption of a certain vitamin), we

might also be able to affect incidence. Only if there is recog-nition of the importance of this “window” of vulnerabilitybefore first pregnancy will these suggestions become part of theresearch agenda.

Smoking and Breast Cancer. Current smoking as an adult hasnot been related to the incidence of breast cancer in numerous

studies. Most studies show either no association or, at most, asmall increase in relative risk. However, several studies that

have examined age at commencement of smoking have dem-onstrated an elevation in risk for early initiators. In three stud-

ies, the elevation in risk was modest (RR = 1.3; Refs. 24-26);however, in a fourth study, the association between early smok-

ing and risk of breast cancer was more pronounced. Amongwomen who were heavy smokers (>25 cigarettes/day), those

who started to smoke before the age of 16 years had a 70-80%greater risk of breast cancer (odds ratios of 1 .7 among cases inCanada and 1.8 among cases in the United States; Ref. 27).

Why Are Age at Menarche and Age at First Birth SoImportant in Determining Risk?

A plausible explanation for the effect of age at menanche andother reproductive risk factors is that genetic errors before firstpregnancy are propagated by the proliferation of pregnancy and

that the differentiation of breast duct cells during pregnancy and

altered rates of cell turnover after the first pregnancy are,thereafter, protective against additional genetic insult. Thetime-dependent susceptibility of the breast to carcinogens andthe impact of pregnancy on risk are consistent with thisconcept.

Implications for Prevention

From menarche to first birth, the log incidence of breast cancerincreases at a rate of approximately 7%/year. The rate decreases

after each birth except the first. With first birth, there is anincrease in risk, followed by a decrease after 10 or more years.This adverse effect of first pregnancy is related to the delaybetween menarche and first birth, with an increase in logincidence of approximately 1 .5%/year elapsed between men-arche and first birth, a finding consistent with animal models ofmammary carcinogenesis. The main effects of reproductive

events are accumulated up to the age at menopause, whereuponthe rate of increase in breast cancer drops to a significantlylower rate.

Options for prevention that offer the greatest potential toreduce the societal burden of breast cancer thus may focus onone of two approaches: (a) decreasing the period from men-

arche to first birth; or (b) preventing the accumulation of DNAdamage during this interval, perhaps through dietary factors.The alternatives to dietary and lifestyle changes [i.e., manipu-lation of hormone levels to reduce cell division (28) or admin-istration of human chorionic gonadotrophin or other hormones

to induce differentiation of mammary epithelium (29)] do notappear to be acceptable to women or feasible at the populationlevel. We emphasize population level interventions because85% of breast cancer occurs among women without a strong

genetic predisposition. Screening for carriers of the brcal genethat predisposes to breast cancer will only identify 15% ofwomen who will develop breast cancer (46). The cost of screen-

ing will be enormous; a conservative estimate of the cost of thetest alone, not accounting for the requisite counseling, teaching,

and treatment, put a price tag of $90 million on a program toscreen all potential carriers (30). Technology at this price maynot be affordable in the United States, much less in other lessdeveloped countries that are just beginning to witness the risein incidence of breast cancer due to delayed childbearing anddecreases in parity.

A reduced interval from menarche to first birth may beobtained either by delaying menarche or by reducing the age atfirst birth. Both ofthese events are strongly influenced by social

factors. Age at menarche has decreased from an average of 16years of age to <13 years of age over the past 130 years (31),

a trend reflecting improved nutrition, decreased childhood in-fections, and decreased levels of physical activity during this

century. Concurrent with this change, median age at first birth

among white women has risen in the United States and otherdeveloped countries since the second World War from approx-imately 23 to 27 years of age. As age at first birth has risen, theaverage number of children has decreased (32).

In contrast, we note that China and many developingcountries have birth rate estimates from 30 years ago (i.e., 1960estimate) of 6.5 births/woman (33). Such a reproductive rate isnot associated with late age at first birth. We also know that theaverage age at menarche in China was about 17 even through

the 1960s (34). If we use the model to fit menarche at 16, ageat first birth at 19, and 6 births spaced a year apart and hold age

at menopause constant at 50 years, we estimate a rate of breastcancer for a 65-year-old Chinese woman that is 93.6/100,000women. For the birth cohort of United States women born

during 1921-1925, the median age at first birth was 23 years,with a mean of 3 children spaced 3 years apart (35). On the

basis of these characteristics, and holding age at menopause at50 years, we predict the rate of breast cancer for a 65-year-old

United States woman to be 279/100,000 women, approximatelythree times the rate for Chinese women.

The importance of reproductive risk factors is emphasizedby the ability of the model to describe a 3-fold differencebetween developed and developing countries. However, this isan underestimate of the true difference because we do not

account for height in the model. Height is known to correlatewith the international rates of breast cancer (36) and is a strong

predictor of risk among postmenopausal women. Height hasincreased over the past century. Childhood nutrition and re-

duced frequency of childhood infections may play an importantrole in these secular changes.

Our model overpredicts rates for Chinese women by as-suming a height comparable to that of United States women forwhom the model was derived. However, the mean height of

United States women is 10 cm greater than that of Chinesewomen. Prospective data from postmenopausal women in the

Nurses’ Health study show that a 10-cm decrease in meanheight is associated with a relative risk of 0.84 (with control forage at menarche, parity, and menopause). Therefore, we mayexpect that the rates for Chinese women should be reduced bysome 15% or more to reflect the lower mean height. Suchadjustment would give a relative rate of “Western” women toChinese women of approximately 3.5 (279 per 100,000/79.6

per 100,000).To prevent breast cancer in our postindustrial society,

social forces must now be brought to bear on the structures and

social norms that support the current incidence rates. Epidemi-

Association for Cancer Research. by guest on October 26, 2020. Copyright 1995 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

570 Review: Breast Cancer Risk Set by Early Life Events

ological data indicate that increased physical activity among

preadolescent and adolescent girls is associated with delay inonset of menses (37) and also may be related to reduced risk ofpremenopausal breast cancer (38). On a positive note, recent

trends in the United Kingdom (39) and the United States (40)suggest a plateauing and a slight increase in the age at men-

arche, perhaps reflecting greater participation by girls in sportsover the past 40 years.

We must now consider seriously the influences that de-

termine age at menarche and explore options for interventionsthat will delay onset of menses. A 1-year increase in age atmenarche would lead to a 9% decrease in a population-widerisk of breast cancer at age 70 years.

Alternatives to delay in age at menarche may includeintroduction of dietary factors such as antioxidants and folatethat would protect the breast against accumulation of molecular

damage during the interval of high cell turnover between men-arche and first birth. Factors, such as vitamin A, that appearprotective in older age (Ref. 40; perhaps by promoting tumor

differentiation) may also help at this younger age. Alterna-tively, folate, which protects against hypomethylation of DNAand is important in the pathway to colon cancer (41), may alsoprotect the mammary glands during cell replication. However,data to address this protective role of diet are not currentlyavailable, and retrospective recall of adolescent dietary habitsin a case-control study would greatly increase the risk of recallbias (42). Perhaps prospective studies could use atypical hy-

perplasia as an outcome measure for evidence that diet andother lifestyle factors inhibit the initiation of carcinogenesis.Research must postulate or identify other dietary factors that

may protect the breast at this time of greatest vulnerability.

The age of initiation of smoking and alcohol consumptionmay play a key role in breast carcinogenesis if indeed adoles-cent exposures are the key determinants of risk. Interventionsthat delay uptake of these behaviors could greatly reduce risk ofbreast cancer (and many other diseases). Interestingly, becausesmoking lowers blood levels of antioxidants (43), an associa-

tion between cigarette smoking and risk of breast cancer may

indirectly support the role of dietary interventions in this agegroup.

Our postindustrial Western society has evolved such thatthe political economy of health now measures breast cancer as

a major social burden. A high proportion of women work infull-time jobs and many delay childbirth while pursuing pro-fessional education. To effect prevention at a population level,

we require three key components: (a) a scientific basis for theintervention; (b) the political will; and (c) a social strategy (44).Research on the etiology and prevention of breast cancer isurgently needed to refine our understanding of childhood andadolescent exposures. These exposures, the most important interms of lifetime breast cancer risk, are the least well described.

Although a growing level of political will to address theprimary prevention of breast cancer is in place and the knowl-edge of the etiology is well advanced, social strategies have not

yet been developed and tested. As the scientific base of under-standing of childhood and adolescent exposures grows, we

must explore the options that these suggest for both individualand population level interventions. Earlier childbearing cannot

be advocated without changes in the social and economicspheres that would allow women to take the time also necessaryfor child rearing without penalty in their professional lives.

Changes that would support a return to earlier childbearingwould be the presence of affordable childcare at the work site,

more flexible work schedules, and promotion of extended lac-tation, which has also been shown to decrease risk of breast

cancer (45). Other interventions may focus on increasing phys-

ical activity during childhood to delay menarche or reducing

alcohol intake and smoking uptake among adolescents andyoung adults. Only with a focused effort will we move towardthe prevention of breast cancer in Western society. This must beour highest priority in breast cancer prevention research.

Appendix

Revised model of breast cancer incidence for multiple births(5).

In (incidence)

= a + b1(t�, - t) + b4(t - ta,) + b17[sum(t - ti)b]

+ b6(t1 - t0) + b19[sum t - t,](t - tm)m

where t0 = age at menarche, t,, = age at first birth, t, = age at

ith birth, t,,, = age at menopause, t = current age, and ,n = Iif postmenopausal and 0 if premenopausal.

Values for the model are: a = -8.7911; bI 0.0723;

b4 = 0.0425; b6 = 0.015468; b17 = -0.003033; and b19 =

-0.00031435.

References

1 . de Waard, F., and Trichopoulos, D. A unifying concept of the aetiology of

breast cancer. Int. J. Cancer, 41: 666-669, 1988.

2. Pike, M., Krailo, M., Henderson, B., Casagrande, J., and Hod, D. ‘ Hormonal’

risk factors, ‘breast tissue age’ and the age-incidence of breast cancer. Nature

(Land.), 303: 767-770, 1983.

3. Russo, I., and Russo, J. Physiological bases of breast cancer prevention. Eur.

J. Cancer Prey., 2 (Suppl.): 101-111. 1993.

4. Trichopoulos, D., Hsieh, C., Lin, M., et a!. Age at any birth and breast cancer

risk. Int. J. Cancer, 31: 701-704, 1983.

5. Rosner, B., Colditz, G., and Willett, W. Reproductive risk factors in a pro-

spective study ofbreast cancer: the Nurses’ Health Study. Am. J. Epidemiol., 139:

819-835, 1995.

6. Lambe, M., Hsieh, C., Trichopoulos, D., Ekbom, A., Pavia, M., and Adami, H.Transient increase in risk of breast cancer after giving birth. N. EngI. J. Med., 331:

5-9, 1994.

7. Hsieh, C., Pavia, M.. Lambe, M., Lan, S., Colditz, G., Ekbom, A., Adami, H.,Trichopoulos, D., and Willeti, W. Dual effect of parity on breast cancer risk. Eur.

J. Cancer, 30A: 969-973, 1994.

8. Janerich, D., and Hoff, M. Evidence for a crossover in breast cancer risk

factors. Am. J. Epidemiol., 116: 737-742, 1982.

9. Gray, G., Henderson, B., and Pike, M. Changing ratio of breast cancerincidence rates with age of black females compared with white females in the

United States. J. Nail. Cancer Inst., 64: 461-463, 1980.

10. Kreiger, N. Social class and the black/white crossover in the age-specific

incidence of breast cancer: a study linking census-derived data to population-

based registry records. Am. J. Epidemiol., /3/: 804-814, 1990.

1 1. U.S. Dept. of Commerce Bureau of the Census. Fertility of American

Women: June 1983, Series P-2t), No. 395. Washington, D.C., 1983.

12. Russo, J., Gusterson, B. A., Rogers, A. E., Russo, I. H., Wellings, S. R., and

van Zwieten, M. J. Biology of disease: comparative study of human and ratmammary tumorigenesis. Lab. Invest., 62: 244-278, 1991).

13. Tokunaga, M., Land, C., Tokuoka, S., Nishimori, I., Soda, M., and Akiba, S.

Incidence of female breast cancer among atomic bomb survivors. 1950-1985.Radial. Res., 138: 209-223, 1994.

14. Miller, A. B., Howe, G. R., Sherman, G. J., Lindsay, J. P., Yaffe, M. J.,

Dinner, P. J., Risch, H. A., and Preston, D. L. Mortality from breast cancer after

irradiation during fluorscopic examinations in patients being treated for tubercu-

losis. N. EngI. J. Med., 321: 1285-1289, 1989.

15. Hancock, S. L., Tucker, M. A., and Hoppe, R. T. Breast cancer after treatment

of Hodgkin’s disease. J. NatI. Cancer Inst., 85: 25-31, 1993.

16. Hildreth, N. G.. Shore, R. E., and Dvortetsky. P. M. The risk of breast cancer

after irradiation of the thymus in infancy. N. EngI. J. Med., 321: 1281-1284,

1989.

17. Rosenberg, L., Metzger, L., and Palmer, J. Alcohol consumption and risk of

breast cancer: a review of the epidemiologic evidence. Epidemiol. Rev., 15:

133-144, 1993.

Association for Cancer Research. by guest on October 26, 2020. Copyright 1995 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Cancer Epidemlologj�, Biomarkers & Prevention 571

46. King, M., Rowell, S., and Love, S. Inherited breast and ovarian cancer. What

are the risks? What are the choices? JAMA, 269: 1975-1980, 1993.

18. Willeti, W. C., Stampfer. M. J., Colditz, G. A., Rosner, B. A., Hennekens, C.

H., and Speizer, F. E. Moderate alcohol consumption and the risk ofbreast cancer.

N. EngI. J. Med., 316: 1174-1180, 1987.

19. Coldit.z, G., Willett, W., and Hunter, D. Family history, age and risk of breast

cancer: prospective data from the Nurses’ Health Study. JAMA, 270: 338-343,

1993.

20. Harvey, E. B., Schairer, C., Brinton, L. A., Hoover, R. N., and Fraumeni,

J. F., Jr. Alcohol consumption and breast cancer. J. NatI. Cancer Inst., 78:

657-661, 1987.

21. Van’t Veer, P., Kok, F. J., Hermus, R. J. J., and Sturmans, A. Alcohol dose,

frequency and age at first exposure in relation to the risk of breast cancer. Intl. J.

Epidemiol., 18: 511-517, 1989.

22. Hiatt. R. A., Klatsky, A. L., and Armstrong, M. A. Alcohol consumption and

the risk of breast cancer in a prepaid health plan. Cancer Res., 48: 2284-2287,

1988.

23. Young, T. B. A case-control study of breast cancer and alcohol consumption

habits. Cancer (Phila.), 64: 522-558, 1989.

24. O’Connell, D. L., Hulka, B. S., Chambless, L. E., Wilkinson, W. E., and

Deubner, D. C. Cigarette smoking, alcohol consumption, and breast cancer risk.

J. NatI. Cancer Inst., 78: 229-234, 1987.

25. Adami. H-O., Lund, E., Bergstrom, R., and Meirik, 0. Cigarette smoking,

alcohol consumption and risk of breast cancer in young women. Br. J. Cancer, 58:

823-837, 1988.

26. Brinton, L., Schairer, C., Stanford, J., and Hoover, R. Cigarette smoking andbreast cancer. Am. J. Epidemiol., 123: 614-622, 1986.

27. Palmer, J. R., Rosenberg, L., Clarke, E. A., Stolley, P. D., Warshauer, M. E.,

Zauber, A. G., and Shapiro, S. Breast cancer and cigarette smoking: a hypothesis.

Am. J. Epidemiol.. 134: 1-13, 1991.

28. Spicer, D., Pike, M., Pike, A., et a!. Pilot trial of a gonadotrophin hormone

agonist with replacement hormones as a prototype contraceptive to prevent breast

cancer. Contraception, 47: 427-444, 1993.

29. Russo, I., Koszalka, M., and Russo, J. Human chorionic gonadotropin and rat

mammary cancer prevention. J. Nail. Cancer Inst., 82: 1286-1289, 1990.

30. Hiatt, R. Ethics and economics of genetic testing for breast cancer. In: B.

Stoll (ed), Reducing Breast Cancer Risk in Women., pp. 55-65. Dordrecht, The

Netherlands: Kluwer Academic Publishers, 1995.

31. van Wieringen, J. Secular growth changes. In: F. Faulkner and J. Tanner

(eds.), Human Growth: A Comprehensive Treatise, pp. 307-331. New York:Plenum Publishing Corp., 1986.

32. Bumpass, L., Rindfuss, R., and Janosik, R. Age and marital status at first birth

and the pace of subsequent fertility. Demography. 15: 75-86, 1978.

33. World Bank. Social Indicators of Development 1993. Baltimore: Johns

Hopkins University Press, 1993.

34. Chen, J., Campbell, T., Junyao, L., and Peto, R. Diet, lifestyle and mortalityin China. Oxford: Oxford University Press, 1990.

35. U.S. Dept. of Commerce Bureau of the Census. Fertility of AmericanWomen: June 1982, Series P-20, No. 387. Washington, D.C., 1982.

36. Micozzi, M. Nutrition, body size, and breast cancer. Yearb. Phys. Anthropol.,

28: 175-206, 1985.

37. Merzenich, H., Boeing, H., and Wahrendorf, J. Dietary fat and sports activityas determinants for age at menarche. Am. J. Epidemiol., 138: 217-224, 1993.

38. Bernstein, L., Henderson, B., and Ross, R. Physical activity and reduction in

risk of breast cancer in young women. The Lancet Conference 1994: The

Challenge of Breast Cancer, Brugge, Belgium, 1994.

39. Dann, T., and Roberts, D. Menarcheal age in University of Warwick youngwomen. J. Biosoc. Sci., 25: 531-538, 1993.

40. Hunter, D., Manson, J., Colditz, G., et a!. A prospective study of intake ofvitamins C, E, and A and risk of breast cancer. N. Engl. J. Med., 329: 234-249,

I 993.

41. Giovannucci, E., Stampfer, M. J., Colditz, G. A., Rimm, E. B., Trichopoulos,

D., Rosner, B. A., Speizer, F. E., and Willett, W. C. Folate, methionine, andalcohol intake and risk of colorectal adenoma. J. NaIl. Cancer Inst., 85: 875-884,

1993.

42. Giovannucci, E., Stampfer, M. J., Colditz, G. A., Manson, J. E., Rosner,B. A., Longnecker, M., Speizer, F. E., and Willett, W. C. A comparison of

prospective and retrospective assessments of diet in the study of breast cancer.

Am. J. Epidemiol., 137: 502-5 1 1,1993.

43. Rimm, E., and Colditz, G. Smoking, alcohol, and plasma levels of carotenes

and vitamin E. In: J. Diana and W. Pryor (eds.), Tobacco, Smoking and Nutrition,

pp. 323-334. New York: New York Academy of Sciences, 1993.

44. Richmond, J., and Kotelchuck, M. Co-ordination and development of strat-

egies and policy-the United States example. In: W. Holland (ed), Textbook ofPublic Health, pp. 195-205. Oxford, United Kingdom: Oxford University Press,

1984.

45. Newcomb, P., Storer, B., Longnecker, M., Mittendorl, R., Greenberg, R.,

Clapp, R., Burke, K., Willeit, W., and MacMahon, B. Lactation and a reduced riskof premenopausal breast cancer. N. EngI. J. Med., 330: 81-87, 1994.

Association for Cancer Research. by guest on October 26, 2020. Copyright 1995 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

1995;4:567-571. Cancer Epidemiol Biomarkers Prev   G A Colditz and A L Frazier  life: prevention efforts must shift focus.Models of breast cancer show that risk is set by events of early

  Updated version

  http://cebp.aacrjournals.org/content/4/5/567

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cebp.aacrjournals.org/content/4/5/567To request permission to re-use all or part of this article, use this link

Association for Cancer Research. by guest on October 26, 2020. Copyright 1995 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from