Hormone Levels In Vegetarian and Nonvegetarian Teenage Girls: Potential Implications for Breast Cancer Risk

[CANCER RESEARCH 52. 578-583. February 1, 1992]

Hormone Levels in Vegetarian and Nonvegetarian Teenage Girls: PotentialImplications for Breast Cancer Risk1

Victoria W. Persky,2 Robert T. Chatterton, Linda V. Van Horn, Mark D. Grant, Patricia Langenberg, and Joy Marvin

Epidemiology/Biostatistics Program, School of Public Health, University of Illinois at Chicago, Chicago, Illinois 60680 fV. W. P., M. D. G.J; Departments of Obstetricsand Gynecology ¡R.T. C.J and Community Health [L. V. V. H.J, Northwestern University, Chicago, Illinois 60611; Department of Geriatrics, La Grange MemorialHospital and Department of Family Practice, Rush Medical College, Chicago, Illinois 60612 [M. D. G.J; Department of Epidemiology and Preventive Medicine,University of Maryland at Baltimore, Baltimore, Maryland 21201 ¡P.L.]; and Department of Surgery, College of Medicine, University of Illinois at Chicago, Chicago,Illinois 60612 [J. M.]

ABSTRACT

Between September 1984 and June 1985, a total of 75 adolescent girls,35 vegetarians residing in a Seventh-Day Adventist school and 40 non-vegetarians residing in a private non-Adventist boarding school, underwent measurement of their plasma hormone levels in the follicular andluteal phases of their menstrual cycles as well as dietary intake measuredby 3-day food records, medical history, height, and weight. There wereno significant differences between vegetarians and nonvegetarians inaverage age of the girls, weight, body mass index, age at menarche, yearssince the onset of menstruation, or percentage of girls with ovulatorycycles. Vegetarian girls had significantly higher levels of log follicularestradiol |2.00 ±0.27 (SD) versus 1.85 ±0.27 pg/ml, P < 0.05) andluteal dehydroepiandrosterone sulfate (DHS) (1.88 ±0.71 versus 1.45 ±0.80 Mg/ml, /' s 0.05) than nonvegetarian girls. Follicular DHS was

higher in vegetarians than in nonvegetarians (1.72 ±0.79 versus 1.45 ±0.95 Mg/ml), but the difference was not significant. The differences infollicular and luteal DHS, but not the difference in log estradiol, weresignificant (/' < 0.05) after controlling for ovulation, smoking, and alcohol

intake with multivariable regression analysis. There were no significantdifferences in testosterone or in percentage free estradiol levels betweenvegetarians and nonvegetarians. Smoking was significantly associatedwith follicular and luteal DHS and with percentage free follicular estradiol, while alcohol use was significantly and inversely associated withpercentage free follicular estradiol after controlling for other variables.The implications for breast cancer risk are discussed.

INTRODUCTION

Previous studies have suggested that hormonal factors associated with early age at menarche, late age at menopause, andlate age at first full-term birth increase the risk of breast cancerin women (1-3). Cross-cultural studies have indicated that high-fat westernized diets may also be associated with increasedbreast cancer risk (4, 5). However, both case-control and cross-cultural studies have failed to reveal consistent differences inendogenous hormone levels either between women with andwithout breast cancer or between women at different risk ofdeveloping the disease (1, 6). Migration and animal studieshave suggested that breast cancer risk may be established earlyin life (4), with subsequent risk being determined during acritical period in teenage or early adult years (7). The purposeof the current study was to explore hormonal differences amongteenage girls whose nutritional intakes indicate varying risk ofbreast cancer, specifically girls who ingest vegetarian or non-

vegetarian diets.

Received 5/10/91; accepted 11/14/91.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

' This study was supported by Grant RO1CA37297 from the National CancerInstitute.

2To whom requests for reprints should be addressed, at Epidemiology/Biosta-tistics Program, School of Public Health, University of Illinois at Chicago. Box6998, Chicago, IL 60680.

MATERIALS AND METHODS

Recruitment and Data Collection. Between September 1984 and June1985, a total of 80 adolescent girls at two boarding schools in theChicago metropolitan area underwent measurements of their plasmahormonal levels timed to their menstrual cycles as well as diet intakemeasured by 3-day food records, medical history, height, and weight.Thirty-five of the girls resided in a Seventh-Day Adventist boardingschool, where a lacto-ovovegetarian diet was served. Forty-five of thegirls resided in a private boarding school in a Chicago suburb, wherethe usual omnivorous American diet was served. Girls and their familieswere recruited for the study at the time of registration for the schoolyear and through brochures included in routine school mailings. Afterinformed consent was obtained from the girls and their parents, thegirls completed a detailed medical history, including questions concerning previous hormone ingestion; menstrual history; medication use;current intake of alcohol, cigarettes, and drugs; vegetarian status; andillnesses which might alter endocrine balance. Girls were coded assmokers if they answered yes to the question, "Do you currently smokecigarettes?" Girls were coded as drinkers if they stated that they had

ingested any alcoholic beverages in the previous week. A 7-day activityquestionnaire developed at Stanford (8) was administered, and heightand weight were measured. The girls agreed to complete 3-day foodrecords and to undergo two venipunctures timed to their menstrualcycles. Following review of the questionnaires, girls were excluded whohad not yet menstruated, who had been pregnant, or who were takingbirth control pills.

Student coordinators assisted the participants in recording the datesof their menstrual cycles. Blood samples were drawn by school nursesbetween days 11 and 13 and between days 21 and 23 of the cycle.Nonfasting blood samples were taken in a standard fashion in the lateafternoon. Plasma was separated and stored at -20°C until transferwithin 1 month to a —¿�70°Cfreezer pending hormonal analyses. Usual

physical activity was converted to total metabolic equivalents by multiplying hours per week spent in moderate, hard, and very hard activityby 4, 6, and 9, respectively, and summing the values. Diet records werecoded by certified dietitians and analyzed by the University of Minnesota Nutrition Coordinating Center database version 6 (9). Althoughmost students at the non-Adventist school were omnivorous, 5 girls atthat school stated that they were vegetarians and were excluded fromthe final analyses.

A total of 35 girls in the vegetarian and 40 girls in the nonvegetarianschool were therefore included in the final study. These represented30% and 32% of all the girls in the vegetarian and nonvegetarianschools, respectively. Girls who were in the current study were of similarage and race (white/nonwhite) distribution as the populations fromwhich they were drawn, with the exception that girls from the vegetarianschool in the study were somewhat older than the overall student groupin the school. A total of 8 girls did not complete the food records, and11 girls in the two schools had only one of the two blood samplesdrawn. Since the final analyses separated follicular and luteal phasemeasurements, these girls were retained in the study and were includedonly in those analyses for which they completed the data collection.

Laboratory Analyses. Estradiol, progesterone, and testosterone weremeasured in diethyl ether extracts of plasma, and dehydroepiandrosterone sulfate was measured in unextracted plasma, by specific radioimmunoassays (10-12). Tritiated tracers with specific activity between45 and 55 Ci/mmol were obtained from New England Nuclear Cor-

578

Research. on January 19, 2016. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

HORMONES IN VEGETARIAN AND NONVEGETAR1AN GIRLS

poration (Boston, MA). Antisera were obtained from the followingsources: estradiol antiserum (AS029) from Dr. Fred Auletta (Universityof Vermont Medical School, Burlington, VT) (cross-reactions: 10%estriol; 60% estrone; <1 % with androgens and corticoids); progesteroneantiserum (ST9) from Dr. Robert Chatterton (Northwestern UniversityMedical School, Chicago, IL) (cross-reactions: 50% 5«-pregnanedione;12% 17-hydroxyprogesterone; 2% 20/}-hydroxy-4-pregnene-3-one;<1% with cortisol, corticosterone, estradiol, and testosterone): testosterone antiserum (T345) from Dr. Robert Rebar (University of California, San Diego, CA) (cross-reactions: 47% dihydrotestosterone, <1%with androstenedione, dehydroepiandrosterone, and estradiol); dehy-droepiandrosterone sulfate antiserum from Radioassay Systems Laboratories (Carson, CA) (cross-reactions: 36% androsterone, 12% 5o-androstane-3,17-dione, 3% androst-4-enedione). Bound steroids wereseparated from free in the assay by addition of dextran-coated charcoal,and the bound form was counted in a liquid scintillation counter.Unbound estradiol was measured in serum by equilibrium dialysis at37°Cas described previously (13).

Coefficients of variation on split samples for measured hormoneswere 0.26 for estradiol (n = 13), 0.21 for testosterone (n = 13), 0.20for DHS' (n = 11), and 0.13 for percentage free estradiol (n = 10).

Statistical Analysis. Distributions of the rollii-nini and luteal phase

estradiol, testosterone, DHS, and percentage free estradiol were examined. Testosterone, DHS, and percentage free estradiol were normally distributed, while the estradiol distributions were skewed positively. Subsequent analyses, therefore, used log,o of estradiol. Means ofeach hormone, as well as age, weight, height, body mass index (kg/m2),

years since menarche, age at menarche, usual activity, and selectednutrients were determined for vegetarian and nonvegetarian girls, anddifferences were examined with Student's t tests. Ovulatory status was

defined by luteal progesterone >5 ng/ml. Differences in racial distribution and percentage of girls with ovulatory cycles were examined byX2 analysis. Differences in hormone levels were also examined, withStudent's i tests, between smokers and nonsmokers and between girls

who did and did not drink alcohol. Nine of the 10 smokers and 11 ofthe 12 drinkers, including 4 girls who were both smokers and drinkers,resided in the nonvegetarian school. Univariate comparisons of hormones with smoking and drinking status, therefore, were limited tononvegetarians. Relationships between hormones and usual activity,age, age at menarche, body mass index, years since menarche, lengthof menstrual cycle, and selected nutrients were determined by correlation analysis. Associations of hormones with vegetarian status, selectednutrients, ovulatory status, body mass index, age, usual activity, smoking history, and alcohol use were determined by multiple linear regression analysis. All statistical analyses were performed using the SASstatistical software package on the UIC IBM mainframe computer.

RESULTS

Demographic and anthropométrievariables as well as plasmahormone levels by school are reported in Table 1. Vegetariangirls were an average of 2 inches shorter (P < 0.05) and 5 Iblighter (P > 0.10) than nonvegetarian girls. The racial compositions of the schools were significantly different, with thevegetarian school containing a greater percentage of Hispanicsand a lower percentage of non-Hispanic whites than the non-vegetarian school. Within the two schools combined, percentages of blacks and Asians were low (6.7% and 5.3%, respectively). There were no significant differences between vegetarians and nonvegetarians in average age of the girls, weight, bodymass index, age at menarche, years after the onset of theirmenstruation, weekly activity, or percentage of the girls withovulatory cycles. Mean log estradiol levels in the follicular phaseof the cycle were significantly higher in vegetarian than innonvegetarian girls. Mean DHS levels were higher in vegetari-

3The abbreviations used are: DHS. dehydroepiandrosterone sulfate; BMI,body mass index: DHEA, dehydroepiandrosterone; SHBG, sex hormone bindingglobulin.

Table I Demographic, anthropométrie,and hormonal levels by school

VegetarianVariableAge

(years)Weight(Ib)Height(inches)Body

mass index(kg/m2)Age

at menarche(years)YearsaftermenarcheActivity

(mets/week)PercentageHispanicPercentage

withovulatorycycleLog

estradiol(follicular)Logestradiol(luteal)DHS

(follicular)(Mg/ml)DHS(lutcal)(Mg/ml)Testosterone

(follicular)(ng/ml)Testosterone

(luteal)(ng/ml)Freeestradiol (follicular)(%)Freeestradiol (luteal) (%)n3535353533333535333433343335333230Mean16.2127.162.922.812.43.973.5±SD1.216.03.43.81.31.641.534.29%27.27%2.00:2.05

:1.72:1.88:bO.27t0.26t0.79b0.71399.4

±121.4455.5

:1.78:1.78:b201.3b0.30t

0.39Nonvegetariann4040383839394040313931393140313526Mean

±SD16.7

±1.1132.2±22.264.9±2.6°22.2

±4.212.4±1.34.2±1.759.6±40.92.50%*22.58%1.85

±0.27°2.15

±0.301.45±0.951.45

±0.80"398.0

±151.5445.5

±159.71.71±0.351.73

±0.33°P < 0.05 for Student's ( test of difference between the mean levels.*P £0.05 for x2 of difference in ethnic composition of the schools.

ans than in nonvegetarians, with a significant difference in theluteal phase. There were no differences in either follicular orluteal phase testosterone levels, nor were there differences ineither follicular or luteal phase percentage free estradiol. Follicular log estradiol, follicular DHS, and luteal DHS (notshown) remained higher in vegetarian than in nonvegetariangirls when analyses were limited to the subset of girls (half theoriginal group) with normal menstrual cycle lengths (26-35days). The numbers of girls with normal cycle lengths, however,were small, and the differences were no longer significant.

Mean daily intakes of selected nutrients by vegetarian statusare presented in Table 2. Vegetarian girls reported consumingsignificantly lower percentages of their calories as total fat (33.7versus 39.7%), saturated fat (12.3 versus 15.5%), and protein(13.0 versus 15.1%). Vegetarians girls also ingested significantlylower quantities of cholesterol (203.8 versus 302.0 mg) andsignificantly higher amounts of unsaturated fat (13.6 versus10.9 g). Other nutrient differences included the ingestion of

Table 2 Dietary factor/nutrient intake, 3-day means by school

Vegetarian(n =32)NutrientTotal

caloriesTotalfat(g)Totalfat (%calories)Saturated

fat(g)Saturatedfat (%calories)Unsaturated

fat(g)t'nsaturatedfat (%calories)Cholesterol

(mg)Polyunsaturated/saturatedfatratioProtein

(g)Protein(%calories)Sucrose

(g)Starch(g)Calcium(mg)Iron(mg)Vitamin

A(1U)Thiaminc(mg)Riboflavin(mg)Niacin(mg)Vitamin

C(mg)Fiber(gm)Caffeine

(mg)Mean189570.233.725.612.313.66.5203.80.6559.113.030.0103.4881.011.443941.622.0513.8131.83.185.55SD58125.97.110.43.35.41.51

13.00.2818.82.821.740.5396.23.918480.490.684.9130.01.3916.6Nonvegetarian

(n =35)Mean174276.039.729.915.510.95.9302.00.4365.015.147.185.3699.89.539221.241.5915.9105.22.1747.9SD54125.66.9°I0.5r2.6°4.5»2.7152.4*0.22°22.13.0*33.8*26.0*348.3r3.0*34680.51*0.64*6.375.40.77°35.9"

°P «0.001 for the difference between the two schools on Student's ( test.* P < 0.05 for the difference between the two schools on Student's t test.'0.05 < P < 0.10 for the difference between the two schools on Student's t

test.

579

Research. on January 19, 2016. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

HORMONES IN VEGETARIAN AND NONVEGETARIAN GIRLS

Table 3 Hormone level by smoking and alcohol use in nonvegetarian school

Smoker Alcohol use

Yes No Yes No

HormoneLog

estradici(follicular)Logestradici(luteal)DHS

(follicular)(wg/ml)DHS(luteal)(pg/ml)Testosterone

(follicular)(ng/ml)Testosterone

(luteal)(ng/ml)Free

estradiol(follicular)TOFree

estradiol (luteal) (%)No.95959585Mean1.742.292.022.104775121.711.88No.3026302631262721Mean1.882.121.281.32"3754331.711.69No.10101010111099Mean1.922.191.821.794184681.471.58No.2921292129212617Mean1.822.131.321.28*3904351.80*1.81*°P< 0.05 for difference between the means on Student's I test.* 0.05 < PS 0.10 for difference between the means on Student's I test.

significantly less sucrose and caffeine, and significantly morethiamine, riboflavin, starch, iron, and fiber among vegetarianthan among nonvegetarian girls.

Levels of hormones were also compared between girls whoadmitted to smoking or drinking and girls who did not admitto using these substances among the nonvegetarians (Table 3).Among the girls who stated that they were drinkers, meanalcohol ingestion in the previous week was 2.0 ±0.9 (SD) oz/week. Luteal phase DHS levels were significantly higher insmokers than in nonsmokers. Luteal phase DHS levels werehigher in drinkers than in nondrinkers, and percentage freeestradiol levels were lower in drinkers than in nondrinkers inboth the follicular and luteal phases of the menstrual cycle, butthe differences were of only borderline significance. Levels ofhormones were also compared among the entire study groupbetween girls whose luteal phase progesterone indicated anovulatory cycle (>5 ng/ml) and those whose levels did not (<5ng/ml) (not shown). The log of follicular estradiol was significantly higher in girls with ovulatory cycles than in girls whosecycles were not ovulatory. This difference remained significantafter control for cycle length. The percentage free estradiol inthe luteal, but not the follicular, phase of the cycle was lowerin girls with ovulatory cycles, with a difference that was of onlyborderline significance. Ovulatory status, however, did not affect either testosterone or DHS. Levels of hormones were alsocompared (not shown) between Hispanics and non-Hispanicwhites. There were no differences between the two ethnic groupsin any of the hormones measured.

Correlations of each measured hormone with level of usualactivity, age, age at menarche, body mass index, length ofmenstrual cycle, and years since menarche were examined. BMIwas inversely and significantly correlated with the log of follicular estradiol (-0.33). Height was significantly and positively

associated with log of follicular estradiol (0.26) and log of lutealestradiol (0.29). There were no other significant associationsbetween any of the hormones and height, weight, BMI, usualactivity, age, age at menarche, length of menstrual cycle, oryears since menarche.

Correlations of hormone levels with selected nutrients (dietary calories, total fat, saturated fat, unsaturated fat cholesterol,protein, polyunsaturated/saturated fat ratio, sucrose, starch,calcium, iron, vitamin A, vitamin C, thiamine, riboflavin, nia-cin, caffeine, and fiber) were also explored. The log of follicularestradiol was significantly correlated with total caloric (0.21),vitamin C (0.30), grams of unsaturated fat (0.26), grams ofprotein (0.25), calcium (0.28), iron (0.26), and fiber (0.32)intake. The percentage free follicular estradiol was significantlyand inversely correlated with grams of total fat (—0.27),gramsof saturated fat (-0.25), grams of protein (-0.30), vitamin A(-0.27), and vitamin C (-0.29) intake. The percentage free

luteal estradiol was significantly and inversely associated withpercentage of calories as protein (-0.31). There were no other

significant associations between nutrients and plasmahormones.

Multiple linear regressions with each hormone as the dependent variable and ovulatory status, school, smoking status,and drinking status as the independent variables are presentedin Table 4. Ovulation remained positively and significantlyassociated with the log of follicular phase estradiol. Vegetarianstatus was significantly associated with both follicular and lutealDHS. Smoking was positively associated with follicular DHS,luteal DHS, and percentage free follicular estradiol but not withother hormones. Percentage free estradiol remained significantly and inversely associated with alcohol use in the follicularphase. Similar regressions were also performed with age, BMI,and usual activity included as additional independent variables(not shown). Age, BMI, and usual activity were not significantlyassociated with any of the hormones, and associations of thehormones with the other independent variables were not substantially changed.

When individual nutrients were substituted for school in theregressions (not shown), the log of follicular estradiol was nolonger significantly associated with total calories, grams ofprotein, grams of unsaturated fat, calcium, vitamin C, iron, orfiber intake. The percentage free follicular estradiol remainedsignificantly and inversely associated with vitamin A intake,while the associations of percentage free follicular estradiolwith grams of fat, grams of saturated fat, grams of protein, andvitamin C intake and the association of percentage free lutealestradiol with percentage of calories as protein were no longersignificant.

Table 4 Multiple linear regression coefficients of hormones against school, ovulatory status, smoking, and alcohol intake

Dependent variable

DHS Testosterone Free estradiolIndependentvariableIntercept

Ovulation'School''Smoking'Alcohol usi-'Foil1.94"

0.16*-0.07-0.17

0.07Lut2.02"

0.110.070.160.02Foil1.70°

0.02-0.61*

0.89*

0.53Lut1.85°

0.02-0.65*

0.67*

0.45Foil387*

21.0-26.5

91.815.9Lut454*

-3.6-26.6

71.524.3Foil1.80"

-0.01-0.04

0.31*-0.36*Lut1.82°

-0.16-0.03

0.15-0.20

" P < 0.001 for significance of regression coefficient.' P < 0.05 for significance of regression coefficient.' Ovulation was coded 0 for nonovulatory cycle and 1 for ovulatory cycle.d School was coded 0 for not vegetarian and I for nonvegetarian school.' Smoking was coded 0 for not current smoker and 1 for current smoker.^Alcohol use was coded 0 for not user and 1 for user.

580

Research. on January 19, 2016. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

HORMONES IN VEGETARIAN AND NONVEGETARIAN GIRLS

DISCUSSION

The results of the current study suggest that adolescentvegetarian girls have significantly higher levels of DHS thanadolescent nonvegetarian girls. We are, to the best of ourknowledge, the first group to note this difference in teenagers.Previous studies comparing DHS levels between vegetarian andnonvegetarian adults (14-16) have found no significant differences, but the numbers of persons examined in each study weresmall. The vegetarian girls in this study ingested significantlyless total fat, saturated fat, sucrose, cholesterol, protein, andcaffeine and significantly more unsaturated fat, fiber, and selected vitamins as measured by 3-day food records. However,the lack of association of any of these nutrients with DHSsuggests that DHS may be sensitive to a characteristic ofvegetarian status other than the nutrients measured in thecurrent study. It is possible that factors specific to meat or tothe soy products substituted for meat in the vegetarian schoolmay have affected DHS. Alternatively, 3-day food records maynot be adequate to characterize individual nutrient intake inthis population, or differences in life style other than dietbetween vegetarians and nonvegetarians might explain the differences in hormone levels. Girls attending the Seventh-Day

Adventist school were more likely to be Hispanic and reportedsmoking and drinking less frequently than girls attending thenon-Adventist school. However, there were no differences inhormones between ethnic groups. Furthermore, smoking wasassociated with higher rather than lower DHS levels, and neither of these variables accounted for the differences in DHSwhen they were controlled for on multivariable regressionanalysis. Similarly, differences in age, BMI, activity, and ovulation did not explain the findings.

A positive association between vegetarian diet and DHSwould provide a mechanism by which low-fat diets in adolescence may protect against subsequent breast cancer risk. Previous studies have noted lower DHEA and DHS blood levelsand urinary production rates in women with breast cancer (17-22). Investigators who have examined premenopausal and post-menopausal women separately have suggested that the inverseassociation between DHS and breast cancer may be limited topremenopausal women (21, 23-25), although even among premenopausal women there have been studies (26) in which noassociation has been found.

In addition to comparisons between women with and withoutcancer, previous studies have also compared adrenal androgensin women at high versus low risk of the disease. Wang et al.(27) noted in 1975 that unaffected sisters of women with breastcancer had significantly lower plasma DHS than women withno family history of breast cancer. Hill et al. (28) found significantly higher levels of DHEA in young menstruating Bantuwomen, who ingested low-fat diets with predominantly vegetable protein, than in Caucasian and Japanese women. Otherresults, however, have not been consistent. The Guernsey studynoted no difference in dehydroepiandrosterone, but lower levelsof the adrenal androgens etiocholanolone and androstendione,in persons who subsequently develop cancer (29). Hayward etal. (30) found higher rather than lower levels of adrenal androgens in British compared with Japanese women. Two additionalstudies have found long-term decreases in DHS after the firstfull-term pregnancy, even though pregnancy is generallythought to be protective against subsequent breast cancer development (31-32).

A protective role for DHS is supported by an increasing body

of laboratory and animal literature. Thijssen et al. (22) havereported in vitro studies with receptors from human myometrialand mammary tumor tissue, indicating that 5-androstene-3/3,17/3-diol, a metabolite of DHEA, inhibits estrogen bindingto the receptors. Maclndoe et al. (33) have presented evidencethat DHS inhibits estrone sulfatase in cancer tissue. Others(34) have suggested that DHEA acts by inhibiting glucose-6-phosphate dehydrogenase, which in turn decreases the availability of NADPH, a cofactor necessary for the activation ofprocarcinogens by cytochrome P-450 mixed-function oxidases.Animal studies have demonstrated inhibition of tumors byadministration of exogenous DHEA. Schwartz (35) has noteda significant reduction in breast cancer incidence in mice thathave been fed DHEA. The same group has further noted thatDHEA inhibits the rates of superperoxide anión radical production by human polymorphonuclear leukocytes stimulatedwith 12-O-tetradecanoylphorbol-13-acetate (36). Other studieshave also found that DHEA inhibits lung, cervix, and colontumors in mice and have suggested that DHEA may be usefulas a chemopreventive agent (37-39).

Previous studies have noted striking age relationships withDHS, with values declining between ages 1 and 5 months,relatively stable between 6 months and 7 years, rising from 8years to late adolescence (40-42), and undergoing a markeddecline in adult years (43). Within the narrow age range of thecurrent study, DHS remained fairly stable. Previous studieshave also suggested that DHS may be associated with obesityin adolescence (44-45), although DHEA and DHS have notconsistently been found to be related to obesity in adults (43,46), and DHS was not related to BMI in the current study. Thepositive association in the current study of follicular DHS withsmoking is consistent with previous reports (47-49) and mayreflect smoking-induced release of adrenocorticotropic hormone or inhibition of aromatase activity in human granulosacells.

The current study does not find a consistent relationshipbetween diet and estradiol levels. Vegetarian girls have higherlog total follicular estradiol levels than nonvegetarian girls, butthe difference does not remain significant on multivariableanalysis. Previous studies have been inconclusive. Althoughseveral investigators have found higher serum estradiol levelsin women on high-fat diets (15, 16, 50, 51), in general mostinvestigators have not found consistent cross-cultural differences in serum estrone or estradiol levels among teenagers (52-54) or adults (14, 30, 55-59) at different risk for breast cancer.

The current study also fails to find an association betweenvegetarian status and percentage free estradiol. Results fromprevious case-control (59-63) and nested case-control (64) studies have suggested that the percentage of unbound estradiolmay be the portion of estradiol which is associated with breastcancer risk. This hypothesis is supported by data indicating thatrisk factors for breast cancer, such as obesity (65-67) andnulliparity (67, 68), may decrease levels of the protein, SHBG,to which estradiol is bound. Several previous studies haveexamined the relationships of dietary factors with SHBG levelsand with estradiol binding. Most of these studies (15, 51, 59,69,70) have found no differences in SHBG levels among womenwith different levels of fat intake. Studies of estradiol bindinghave been inconsistent, with a positive (51), a negative (69),and no (70) association between dietary fat intake and percentage unbound estradiol being noted.

The current study does not show an association betweenpercentage unbound estradiol and BMI, in contrast with pre-

581

Research. on January 19, 2016. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

HORMONES IN VEGETARIAN AND NONVEGETAR1AN GIRLS

vious reports (62, 66). This study similarly does not find anassociation between unbound estradici and smoking on univar-iate analysis, although smokers did have higher levels of percentage unbound follicular estradici than nonsmokers aftercontrolling for other variables. The current study does show anassociation between ovulation and percentage unbound estra-diol in the luteal phase of the menstrual cycle of borderlinesignificance on univariate, but not on multivariable, analysis, afinding which is consistent with previous literature (71, 72).There is, in addition, an inverse association between alcoholuse and percentage free follicular estradiol, an association whichis significant in blood from the follicular phase of the cycle andis independent of ovulation, vegetarian pattern, and smokingstatus on multivariable analysis. The few previous studies whichhave examined the relationship between alcohol intake andestradiol binding in women have been inconclusive. Two studies(73, 74) found no difference in SHBG levels between alcoholicwomen and controls, while one study (75) noted a decrease inSHBG levels after exclusion of drinkers, a change consistentwith the inverse association with percentage free estradiol seenin the current study.

The major finding in the current study, that vegetarian dietsare associated with increased levels of DHS, is present in bothphases of the menstrual cycle, independent of other risk factors,and consistent with an increasing body of literature suggestingthat DHS may be protective against cancer risk. However, thepaucity of other epidemiological data on the effects of diet onadrenal androgens and the lack of an association of DHS withspecific nutrient intake suggest that these findings should beviewed with caution. The association between vegetarian dietand DHS should be replicated in other populations, and thebiological basis for the relationship needs to be establishedbefore an inference of an etiological association can be drawn.

ACKNOWLEDGMENTS

The authors wish to express their deep appreciation to the studentsand staff of the two boarding schools, whose patient and generous helpfacilitated all aspects of data collection. The authors are also gratefulfor the scientific and financial support of Drs. Nathaniel Berlin, PaulLevy, and Jeremiah Stamler; for the insightful comments of Dr. William Haenszel and Dr. Oscar Hechter; and for the programmingsupport of Tom Tokich and Ron Hoeksema.

REFERENCES

1. Kelsey, J. L. A review of the epidemiology of human breast cancer. Epidemic!.Rev., /.-74-109, 1979.

2. Helmrich, S. P.. Shapiro. S.. Rosenberg, L., Kaufman. D. W., Slone. D..Bain. C.. Miettinen. O., Stolley. P. D., Rosenshein. N. B.. Knapp. R. C,Leavitt. T.. Schotlenfeld. D.. Engle. R. L., and Levy. M. Risk factors forbreast cancer. Am. J. Epidemici.. 117: 35-45, 1983.

3. Lubin. J. H.. Burns. P. E.. Blot. W. J.. Lees. A. W., May. C.. Morris, L. E.,and Fraumeni, J.F. Risk factors for breast cancer in women in northernAlberta. Canada, as related to age at diagnosis. J. Nati. Cancer Inst.. 68:211-217. 1982.

4. Haenszel. W., and Kurihara. M. Studies of Japanese migrants. I. Mortalityfrom cancer and other diseases among Japanese in the United States. J. Nati.Cancer Inst.. 40: 43-68. 1968.

5. Gray. G. E.. Pike. M. C.. and Henderson. B. E. Breast-cancer incidence andmortality rates in different countries in relation to known risk factors anddietary practices. Br. J. Cancer. 39: 1-7. 1979.

6. Zumoff. B. Hormonal profiles in women with breast cancer. Anticancer Res..*: 627-636. 1988.

7. Korenman. S. G. Oestrogen window hypothesis of the aetiology of breastcancer. Lancet, /: 700-701. 1980.

8. Taylor. C. B., Coffey. T.. Berra. K., [affaldano. R.. Casey. K.. and Haskeil.W. L. Seven-day activity and self-report compared to a direct measure ofphysical activity. Am. J. Epidemiol.. 120: 818-824. 1984.

9. Sievert, Y. A.. Schäkel,S. F.. and Buzzard. I. M. Maintenance of a nutrient

10.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

database for clinical trials. Controlled Clin. Trials, 10:416-425. 1989.Haning, R.. Orczyk, G. P., Caldwell, B. V., and Behrman. H. R. Plasmaestradiol. estrone, estriol, and urinary estriol glucuronide. In: B. M. Jaffe andH. R. Belii in.MI(eds.). Methods of Hormone Radioimmunoassay. Ed. 2. pp.675-700. New York: Academic Press, 1979.Orczyk, G. P., Hickens, M., Arth. G., and Behrman, H. R. Progesterone and20<(-dihydroprogesterone. In: B. M. Jaffe and H. R. Behrman (eds.). Methodsof Hormone Radioimmunoassay. Ed. 2. pp. 701-714. New York: AcademicPress, 1979.Auletta, F. J., Caldwell. B. V., and Hamilton. G. L. Androgens: testosterone,and dihydrolestosterone. In: B. M. Jaffe and H. R. Behrman (eds.). Methodsof Hormone Radioimmunoassay. Ed. 2, pp. 715-726. New York: AcademicPress, 1979.Hickson, R. C., Galassi, T. M., Kurowski, T. T., Daniels, D. G., andChatterton. R. T.. Jr. Androgen and glucocorticoid mechanisms in exercise-induced cardiac hypertrophy. J. Physiol.. 246 (Heart Cire. Physiol. 15):H761-H767. 1984.Bélanger,A.. Locong. A.. Noel. C.. Cusan. L.. Dupont. A., Prévost,J., Carón,S., and Sevigny, J. Influence of diet on plasma steroid and sex plasma bindingglobulin in adult men. J. Steroid Biochem., 32: 829-833, 1989.Barbosa. J. C.. Shultz. T. D.. Filley. S. J., and Nieman, D. C. The relationshipamong adiposity, diet and hormone concentrations in vegetarian and non-vegetarian postmenopausal women. Am. J. Clin. Nutr., 51: 798-803, 1990.Shultz. T. D.. and Leklem, J. E. Nutrient intake and hormonal status ofpremenopausal vegetarian Seventh-Day Adventists and premenopausal non-vegetarians. Nutr. Cancer. 4: 247-259, 1983.Wang. D. Y.. Bulbrook. R. D.. Herían.M., and Hayward. J. L. Studies onthe sulfate esters of dehydroepiandrosterone and androsterone in the bloodof women with breast cancer. Eur. J. Cancer, 10: 477-482, 1974.Brownsey. B.. Cameron, E. H. D., Griffiths, K., Gleave, E. N., Forrest, A.P. M., and Campbell, H. Plasma dehydroepiandrosterone sulfate levels inpatients with benign and malignant breast disease. Eur. J. Cancer, 8: 131-137. 1972.Thomas. B. S.. Kirby, P., Symes. E. K., and Wang. D. Y. Plasma dehydroepiandrosterone concentrations in normal women and in patients with benignand malignant breast disease. Eur. J. Cancer. 72:405-409, 1976.Rose, D. P.. Slaubcr, P., Thiel, A., Crowley, J. J., and Milbrath, J. R. Plasmadehydroepiandrosterone sulfate, androstenedione and Cortisol, and urinaryfree cortisol excretion in breast cancer. Eur. J. Cancer, 13: 43-47, 1977.Zumoff. B.. Levin. J.. Rosenfeld. R. S., Markham, M., Strain, G. W., andFukushima. D. K. Abnormal 24-hr mean plasma concentrations of dehydro-isoandrosterone and dehydroisoandrosterone sulfate in women with primaryoperable breast cancer. Cancer Res., 41: 3360-3363. 1981.Thijssen, J. H. H., Poortman. J.. and Schwarz, F. Androgens in postmenopausal breast cancer: excretion, production and interaction with estrogens.J. Steroid Biochem.. 6: 729-734. 1975.Gordon. G. B.. Bush, T. L., Helzlsouer, K. J., Miller, S. R.. and Comstock.G. W. Relationship of serum levels of dehydroepiandrosterone and dehydroepiandrosterone sulfate to the risk of developing postmenopausal breastcancer. Cancer Res.. 50: 3859-3862. 1990.Barrett-Connor. E., Friedlander, N. J., and Khaw. K. T. Dehydroepiandrosterone sulfate and breast cancer risk. Cancer Res., 50: 6571-6574, 1990.Bernstein. L., Ross, R. K., Pike, M. C., Brown. J. B.. and Henderson, B. E.Hormone levels in older women: a study of post-menopausal breast cancerpatients and healthy population controls. Br. J. Cancer, 61: 298-302, 1990.Secreto. G., Toniolo, P.. Pisani. P., Recchione, C.. Cavalieri. A., Fariselli,G., Totis, A., DiPietro, S.. and Bcrrino, F. Androgens and breast cancer inpremenopausal women. Cancer Res.. 49: 471-476. 1989.Wang, D. Y., Bulbrook. R. D.. and Hayward, J. L. Urinary and plasmaandrogens and their relation to familial risk of breast cancer. Eur. J. Cancer,//: 873-877. 1975.Hill. P.. Wynder, E. L., Helman. P.. Hickman, R., Roña,G., and Kuno, K.Plasma hormone levels in different ethnic populations of women. CancerRes., 36:2297-2301. 1976.Bulbrook. R. D.. Hayward. J. L., and Spicer, C. C. Relation between urinaryandrogen and corticoid excretion and subsequent breast cancer. Lancet, 2:395-398. 1971.Hayward. J. L.. Greenwood. F. C.. Glober. G.. Stemmernian. G., Bulbrook.R. D., Wang. D. Y., and Kumaokas. S. Endocrine status in normal British,Japanese and Hawaiian Japanese women. Eur. J. Cancer. 14: 1221-1228,1978.Musey, V. C.. Collins, D. C., Brogan. D. R.. Santos, V. R., Musey, P. I.,Martino-Salzman. D.. and Preedy, J. R. K. Long term effects of a firstpregnancy on the hormonal environment: estrogens and androgens. J. Clin.Endocrinol. Metab.. 64: 111-118, 1987.Key, T. J. A.. Pike, M. C., Wang, D. Y., and Moore. J. W. Long term effectsof a first pregnancy on serum concentrations of dehydroepiandroteronesulfate and dehydroepiandrosterone. J. Clin. Endocrinol. Metab., 70: 1651-1653. 1990.Maclndoe. J. H.. Woods. G.. Jeffries, L.. and Hinkhouse. M. The hydrolysisof estrone sulfate and dehydroepiandrosterone sulfate by MCF-7 humanbreast cancer cells. Endocrinology, 123: 1281-1287. 1988.Boone, C. W., Kelloff, G. J., and Malone, W. E. Identification of candidatecancer chemopreventive agents and their evaluation in animal models andhuman clinical trials: a review. Cancer Res., 50: 2-9, 1990.Schwartz, A. G. Inhibition of spontaneous breast cancer formation in femaleC3H (A'Ya) mice by long-term treatment with dehydroepiandrosterone.Cancer Res.. 39: 1129-1132, 1979.

582

Research. on January 19, 2016. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

HORMONES IN VEGETARIAN AND NONVEGETARIAN GIRLS

36. Whitcomb, J. M., and Schwartz, A. G. Dehydroepiandrosterone and 16a-Br-epiandrosterone inhibit 12-O-tetradecanoylphorbol-n-acetate stimulation of superperoxide radical production by human polymorphonuclear leukocytes. Carcinogenesis (Lond.), 6: 333-335, 1985.

37. Nyce, J. W., Magee, P. N., Hard, G. C., and Schwartz, A. G. Inhibition of1,2-dimethylhydrazine-induced colon tumorigenesis in Balb/c mice by dehy-droepiandrosterone. Carcinogenesis (Lond.), 5: 57-62, 1984.

38. Schwartz, A. G., and Tannen. R. H. Inhibition of 7,12-dimethyl-benz(a)anthracene- and urethan-induced lung tumor formation in A/J miceby long-term treatment with dehydroepiandrosterone. Carcinogenesis(Lond.), 2: 1335-1337, 1981.

39. Rao, A. R. Inhibitory action of dehyroepiandrosterone on methylchooan-threne-induced Carcinogenesis in the uterine cervix of mouse. Cancer Lett.,45:1-5, 1989.

40. Babalola, A. A., and Ellis, G. Serum dehyroepiandrosterone sulfate in anormal pediatrie population. Clin. Biochem., 18: 184-189, 1985.

41. Mavoungou, D., Gass, R., Emane, N., Cooper, R. W., and Roth-Meyer, C.Plasma dehydroepiandrosterone, its sulfate, testosterone and FSH duringpuberty of African children in Gabon. J. Steroid Biochem., 24: 645-651,1986.

42. Sizonenko, P. C., Paunier, L., and Carmignac, D. Hormonal changes duringpuberty. IV. Longitudinal study of adrenal androgen secretions. HormoneRes. (Basel), 7: 288-302, 1976.

43. Barrett-Connor, E., Khaw, K. T., and Yen, S. S. C. A prospective study ofdehydroepiandrosterone sulfate, mortality and cardiovascular disease. N.Engl. J. Med., 315: 1519-1524, 1986.

44. Katz, S. H., Hediger, M. L., Zemel, B. S., and Parks, J. S. Blood pressure,body fat, and dehydroepiandrosterone sulfate variation in adolescence. Hypertension (Dallas), *: 277-284, 1986.

45. Katz, S. H., Hediger, M. L., Zemel, B. S., and Parks, J. S. Adrenal androgens,body fat and advanced skeletal age in puberty: new evidence for the relationsof adrenarche and gonarche in males. Hum. Biol., 57:401-413, 1985.

46. Kurtz, B. R., Givens, J. R., Komindr, S., Stevens, M. D., Karas, J. G.. Bittle,J. B., Judge, D., and Kitabchi, A. E. Maintenance of normal circulating levelsof A"-androstenedione and dehydroepiandrosterone in simple obesity despiteincreased metabolic clearance rates: evidence for a servo-control mechanism.J. Clin. Endocrinol. Metab., 64: 1261-1267, 1987.

47. Khaw, K. T., Chir, M. B. B., Tazuke, S., and Barrett-Connor, E. Cigarettesmoking and levels of adrenal androgens in postmenopausal women. N. Engl.J. Med., 318: 1705-1709, 1988.

48. Friedman, A. J., Ravnikar, V. A., and Barbieri, R. L. Serum steroid hormoneprofiles in postmenopausal smokers and nonsmokers. Fértil.Steril., 47: 398-401, 1987.

49. Dai, W. S., Culai, J. P., Kuller, L. H., and Cauley, J. A., for the MRFITResearch Group, Cigarette smoking and sex hormoens in men. Am. J.Epidemici., 128: 796-805, 1988.

50. Rose, D. P.. Boyar, A. P., Cohen, C., and Strong, L. E. Effect of a low-fatdiet on hormone levels in women with cystic breast disease. I. Serum steroidsand gonadotropins. J. Nati. Cancer Inst., 78:623-626, 1987.

51. Prentice, R., Thompson, D., Clifford, C, Gorbach, S., Goldin. B., and Byar,D. Dietary fat reduction and plasma estradici concentration in healthypostmenopausal women. J. Nati. Cancer Inst., 82: 129-134, 1990.

52. Gray, G. E., Pike, M. C., Hirayama, T., Tellez, J., Gerkins, V., Brown, J.B., Casagrande, J. T., and Henderson, B. E. Diet and hormone profiles inteenage girls in four countries at different risk for breast cancer. Prev. Med.,//: 108-113, 1982.

53. Hill, P.. Wynder, E. L., Garbaczewski, L., Helman, P., Hill, M., Sporangisa,J., and Huskisson, J. Diet and menarche in different ethnic groups. Eur. J.Cancer, 70:519-525, 1980.

54. Gray, G. E., Williams, P., Gerkins, V., Brown, J. B., Armstrong, B., Phillips,R., Casagrande, J. T., Pike, M. C., and Henderson, B. E. Diet and hormonelevels in Seventh-Day Adventist teenage girls. Prev. Med., //: 103-107,1982.

55. Goldin, B. R., Adlercreutz, H.. Gorbach, S. L., Warram, J. H., Dwyer, J. T.,Swenson, L., and Woods, M. N. Estrogen excretion patterns and plasmalevels in vegetarian and omnivorous women. N. Engl. J. Med., 307: 1542-1547, 1982.

56. Wysowski, D. K., Comstock, G. W., Helsing, K. J., and Lau, H.L. Sexhormone levels in serum in relation to the development of breast cancer.Am. J. Epidemiol., 125: 791-799, 1987.

57. Hagerty, M. A., Howie, B. J., Tan, S., and Shultz, T. D. Effect of low- andhigh-fat intakes on the hormonal milieu of premenopausal women. Am. J.Clin. Nutr., 47: 653-659, 1988.

58. Woods, M. N., Gorbach, S. L., Longcope, C., Goldin. B. R., Dwyer, J. T.,and Morrill-LaBrode, A. Low-fat, high-fiber diet and serum estrone sulfatein premenopausal women. Am. J. Clin. Nutr., 49: 1179-1183, 1989.

59. Reed, M. J., Cheng, R. W., Noel, C. T., Dudley. H. A. F., and James, V. H.T. Plasma levels of estrone, estrone sulfate, and estradiol and the percentageof unbound estradiol in postmenopausal women with and without breastdisease. Cancer Res., 43: 3940-3943, 1983.

60. Langley, M. S., Hammond, G. L., Bardsley, A., Sellwood, R. A., andAnderson, D. C. Serum steroid binding proteins and the bioavailability ofestradiol in relation to breast diseases. J. Nati. Cancer Inst., 75: 823-829,1985.

61. Moore, J. W.. Clark, G. M. G., Bulbrook, R. D., Hayward, J. L., Murai, J.T., Hammond, G. L., and Siiteri, P. K. Serum concentrations of total andnon-protein oestradiol in patients with breast cancer and normal controls.Int. J. Cancer, 29: 17-21, 1982.

62. Ota, D. M., Jones, L. A., Jackson, G. L., Jackson, P. M., Kemp, K., andBauman, D. Obesity, non-protein-bound estradiol levels, and distribution ofestradiol in the sera of breast cancer patients. Cancer (Phila.), 57: 558-562,1986.

63. Jones, L. A., Ota, D. M., Jackson, G. A., Jackson, P. M., Kemp, K.,Anderson, D. E., McCamant, S. K., and Bauman, D. H. Bioavailability ofestradiol as a marker for breast cancer risk assessment. Cancer Res., 47:5224-5229, 1987.

64. Moore, J. W., Clark, G. M. G., Hoare, S. A., Millis, R. R., Hayward, J. L.,Quinlan, M. K., Wang, D. Y., and Bulbrook, R. D. Binding of oestradiol toblood proteins and aetiology of breast cancer. Int. J. Cancer, 38: 625-630,1986.

65. von Schoultz, B., and Carlstrom, K. On the regulation of sex-hormone-binding globulin—a challenge of an old dogma and outlines of an alternativemechanism. J. Steroid Biochem., 32: 327-334, 1989.

66. Ingram, D., Nottage, E., Siobhan, N. G., Sparrow, L., Roberts, D., andWilcox, D. Obesity and breast disease. Cancer (Phila.), 64:1049-1053,1989.

67. Bernstein, L.. Pike, M. C, Ross, R. K., Judd, H. L., Brown, J. B., andHenderson, B. E. Estrogen and sex hormone-binding globulin levels innulliparous and parous women. J. Nati. Cancer Inst., 74: 741-745, 1985.

68. Moore, J. W., Key, T. J. A., Clark, G. M. G., Hoare, S. A., Allen, D. S., andWang, D. Y. Sex-hormone-binding globulin and breast cancer risk. Antican-cer Res., 7: 1039-1048, 1987.

69. Moore, J. W., Clark, G. M. G., Takatani, O., Wakabayashi, Y., Hayward, J.L., and Bulbrook, R. D. Distribution of 17/3-estradiol in the sera of normalBritish and Japanese women. J. Nati. Cancer Inst., 71: 749-754, 1983.

70. Fentiman, I. S., Caleffi, M., Wang, D. Y., Hampson, S. J., Hoare, S. A.,Clark, G. M. G., Moore. J. W., Bruning, P., and Bonfrer, J. M. G. Thebinding of blood-born estrogens in normal vegetarian and omnivorous womenand the risk of breast cancer. Nutr. Cancer, //: 101-106, 1988.

71. Plymate, S. R., Moore, D. E., Cheng, C. Y., Bardin, C. W., Southworth, M.B., and Levinski, M. J. Sex hormone-binding globulin changes during themenstrual cycle. J. Clin. Endocrinol. Metab., 61: 993-996, 1985.

72. Apter, D., Bolton, N. J.. Hammond, G. L., and Vihko, R. Serum sexhormone-binding globulin during puberty in girls and in different types ofadolescent menstrual cycles. Acta Endocrinol., 107:413-419, 1984.

73. Hugues, J. N., Coste, T., Perret, G., Jayle, M. F., Sebaoun, J., and Modigliani, E. Hypothalamo-pituitary ovarian function in thirty-one women withchronic alcoholism. Clin. Endocrino!., 12: 543-551, 1980.

74. Valimaki, M., Pelkonen, R., Salaspuro, M., Harkonen, M., Hirvonen, E.,and Ylikahri, R. Sex hormones in amenorrheic women with alcoholic liverdisease. J. Clin. Endocrinol. Metab., 59: 133-138, 1984.

75. Armstrong, B. K., Brown, J. B., Clarke, H. T., Crooke, D. K., Hahnel, R.,Masarel, J. R., and Ratajczak, T. Diet and reproductive hormones: a studyof vegetarian and nonvegetarian postmenopausal women. J. Nati. CancerInst., 67:761-767, 1981.

583

Research. on January 19, 2016. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

1992;52:578-583. Cancer Res   Victoria W. Persky, Robert T. Chatterton, Linda V. Van Horn, et al.   Girls: Potential Implications for Breast Cancer RiskHormone Levels in Vegetarian and Nonvegetarian Teenage

  Updated version

  http://cancerres.aacrjournals.org/content/52/3/578

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

  [email protected] at

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

  Permissions

  [email protected] at

To request permission to re-use all or part of this article, contact the AACR Publications

Research. on January 19, 2016. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from