Carotenoids and the Prostate

7/25/2019 Carotenoids and the Prostate

1/10

1023

American Journal of Epidemiology

Copyright 2002 by the Johns Hopkins Bloomberg School of Public Health

All rights reserved

Vol. 155, No. 11

Printed in U.S.A.

Serum Lycopene, Other Carotenoids, and Prostate Cancer Vogt et al.

Serum Lycopene, Other Serum Carotenoids, and Risk of Prostate Cancer inUS Blacks and Whites

T. M.Vogt,1,2 S. T. Mayne,1 B. I. Graubard,2 C. A. Swanson,2 A. L. Sowell,3 J. B. Schoenberg,4 G. M. Swanson,5

R. S. Greenberg,6 R. N. Hoover,2 R. B. Hayes,2 and R. G. Ziegler2

Epidemiologic studies investigating the relation between individual carotenoids and risk of prostate cancerhave produced inconsistent results. To further explore these associations and to search for reasons prostatecancer incidence is over 50% higher in US Blacks than Whites, the authors analyzed the serum levels ofindividual carotenoids in 209 cases and 228 controls in a US multicenter, population-based case-control study(19861989) that included comparable numbers of Black men and White men aged 4079 years. Lycopene wasinversely associated with prostate cancer risk (comparing highest with lowest quartiles, odds ratio (OR) = 0.65,95% confidence interval (CI): 0.36, 1.15; test for trend, p= 0.09), particularly for aggressive disease (comparing

extreme quartiles, OR = 0.37, 95% CI: 0.15, 0.94; test for trend, p= 0.04). Other carotenoids were positivelyassociated with risk. For all carotenoids, patterns were similar for Blacks and Whites. However, in both thecontrols and the Third National Health and Nutrition Examination Survey, serum lycopene concentrations weresignificantly lower in Blacks than in Whites, raising the possibility that differences in lycopene exposure maycontribute to the racial disparity in incidence. In conclusion, the results, though not statistically significant,suggest that serum lycopene is inversely related to prostate cancer risk in US Blacks and Whites. Am JEpidemiol 2002;155:102332.

Blacks; carotenoids; case-control studies; nutrition; prostatic neoplasms

Received for publication May 29, 2001, and accepted for publica-tion September 13, 2001.

Abbreviations: CI, confidence interval; NHANES III, Third Health

and Nutrition Examination Survey; OR, odds ratio.1 Department of Epidemiology and Public Health, Yale UniversitySchool of Medicine, New Haven, CT.

2 Division of Cancer Epidemiology and Genetics, National CancerInstitute, Bethesda, MD.

3 National Center for Environmental Health, Centers for DiseaseControl and Prevention, Chamblee, GA.

4 Cancer Epidemiology Services, New Jersey Department ofHealth and Senior Services, Trenton, NJ.

5 College of Human Medicine, Michigan State University, EastLansing, MI.

6 Medical University of South Carolina, Charleston, SC.Reprint requests to Dr. Tara M. Vogt, Division of Cancer

Epidemiology and Genetics, National Cancer Institute, 6120Executive Boulevard, Suite 320, Bethesda, MD 20892 (e-mail:[email protected]).

Prostate cancer is the most commonly diagnosed cancerand the second leading cause of cancer mortality amongmen living in the United States (1). African-American menbear a disproportionately heavy burden from this disease,with incidence rates over 50 percent higher than those forUS Whites (1). The etiology of prostate cancer and an expla-nation for the racial disparity in incidence remain elusive.However, migrant studies and temporal shifts in incidencewithin countries suggest that modifiable factors such as dietcould be involved (24).

Carotenoids are natural pigments synthesized in plantsand bacteria that aid in photosynthesis and photoprotection

(5). Humans are exposed to these compounds mainlythrough intake of fruits and vegetables. Although their func-tional importance is not entirely understood, it is known thatsome carotenoids are converted to vitamin A in vivo. Thecancer-preventive potential of carotenoids has been hypoth-esized to lie in their ability to quench singlet oxygen andperform antioxidant functions, although other mechanismshave also been suggested (6). Although there is evidencethat carotenoid-rich foods decrease the risk of many cancers,the literature supporting a protective role for prostate cancerhas been inconsistent (7). However, several studies havesuggested that increased exposure to lycopene, a carotenoidfound primarily in tomatoes, may reduce the risk of prostatecancer (8).

To explore reasons for the racial disparity in prostate can-cer incidence in the United States, the National Cancer

Institute conducted a multicenter, population-based case-control study of prostate cancer that included comparablenumbers of Black men and White men between the ages of40 and 79 years. By purposely oversampling Blacks, a pop-ulation often underrepresented in epidemiologic studies, thisdesign provided a unique opportunity to examine risk pat-terns among Blacks separately and in comparison with thoseof Whites. Blood was collected and potential risk factorswere assessed in detail by structured at-home interviews.The present analysis investigates the risk of prostate cancerassociated with serum concentrations of lycopene and otherindividual carotenoids, including -carotene, -carotene, -cryptoxanthin, and lutein/zeaxanthin.

y

g

y

pj

j

g
http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/


2/10

1024 Vogt et al.

Am J Epidemiol Vol. 155, No. 11, 2002

MATERIALS AND METHODS

Study design

Subjects comprised a subset from a multicenter, popula-tion-based case-control study of four cancers, includingmultiple myeloma and cancers of the prostate, esophagus,and pancreas, that occur excessively in Blacks. Eligiblecases were between 40 and 79 years of age with histologi-cally confirmed, incident prostate cancer diagnosed betweenAugust 1986 and April 1989. They were identified fromrecords of hospitals covered by the population-based cancerregistries for Atlanta, Georgia; Detroit, Michigan; and NewJersey (10 counties). For each study center, cases wereselected by a race- and age-stratified sampling scheme thatoversampled Blacks and younger men in order to ensureapproximately equal numbers of men from each race andacross a broad age range. Population-based controls wereidentified with either random digit dialing (9) (under age 65years) or Health Care Financing Administration records (age65 years or older) and were frequency matched to the antic-

ipated case distribution by study center, race, and 5-year agegroup.

After informed consent from participants was obtained, in-person structured interviews were conducted, usually in thesubjects homes. Questions were asked regarding demograph-ics, family history of cancer, medical and sexual history, alco-hol and tobacco use, dietary intake, and occupational history.Diet was ascertained via a 60-item food frequency question-naire in which subjects were asked to recall their usual fre-quency of consumption of specific foods consumed by USBlacks and Whites (10) over their adult lives, excluding thepast 5 years.

Study participation

Of the 981 prostate cancer cases and 1,315 controls whowere successfully interviewed, a subset was selected todonate blood for further analyses. Cases were consideredineligible if they had undergone orchiectomy or had or werepresently undergoing hormone treatment, chemotherapy, orradiation therapy. A total of 483 (234 Black, 249 White)cases were eligible and invited to participate. Of these,blood was successfully obtained from 127 Black cases (54percent) and 147 White cases (59 percent). Eligible controlswere selected to be frequency matched to cases on studycenter, race, and age. A total of 467 (213 Black, 254 White)

controls were invited to participate. Blood was successfullycollected from 137 Black controls (64 percent) and 158White controls (62 percent). For both the interview andblood collection, overall participation rates for eligible sub-jects were as follows: Black cases, 42 percent; Black con-trols, 45 percent; White cases, 44 percent; White controls,42 percent.

Because of budgetary constraints, individual carotenoidswere measured in blood from a subset of cases and controlsthat was balanced by age and race for each study center.These 437 study participants included 209 cases (99Blacks, 110 Whites) and 228 controls (108 Blacks, 120Whites).

Biochemical analysis

A phlebotomist visited each participating subject in hishome and drew approximately 50 ml of blood. The medianinterval between prostate cancer diagnosis and blood drawwas 3.7 months (ranging from 27 days to 2.3 years). Fornutrient analyses, blood was collected in serum separator

tubes, previously covered in aluminum foil to limit exposureto natural light. After at least 30 minutes for clot formation,samples were refrigerated. Within 6 hours of blood collec-tion, serum was separated by centrifugation, and 0.5-mlaliquots were stored at 70C.

In 1991, individual carotenoids (all-trans-carotene, all-trans -carotene, all-trans -cryptoxanthin, lutein/zeaxan-thin, all-trans lycopene) were measured in the sera by theCenters for Disease Control and Prevention using isocraticreverse-phase high-performance liquid chromatographywith multiwavelength ultraviolet/visible light detection at450 nm (11). After the addition of nonapreno -carotene, theinternal standard, all samples were extracted into hexane

and then redissolved in ethanol, and an equal volume of ace-tonitrile was added. A filtrate was then injected onto a 15-cmC18 reverse-phase column and eluted with 50 percentethanol:50 percent acetonitrile containing 100 l/liter ofdiethylamine. The chromatograms were quantified by com-paring the peak height of each carotenoid with that of thesame carotenoid in a standard solution and adjusting for thepeak height of the internal standard.

To evaluate quality control, we inserted blinded sampleswith high and low levels of the individual carotenoids priorto shipment to the Centers for Disease Control andPrevention for analysis. For each of the five carotenoids, thevariation in these blinded samples met all of the quality con-

trol requirements of Westgard et al. (12). Coefficients ofvariation were between 4 and 8 percent, except the low -carotene samples that had a coefficient of variation of 12percent.

Statistical analysis

Cases and controls were compared in terms of variousdemographic and other factors using chi-square tests.Among controls, Pearsons correlation coefficients and par-tial correlation coefficients (adjusted for serum cholesterol)were calculated for serum carotenoids, fruit and vegetableintake, and intake of lycopene-rich foods. The median

serum carotenoid concentrations in cases and controls and inBlack and White controls were compared using Wilcoxonsrank sum tests.

For both stratified and unstratified analyses, serumcarotenoid concentrations were categorized into quartilesbased on the total population of controls. Unconditionallogistic regression (13) was used to generate adjusted oddsratios and 95 percent confidence intervals for prostate can-cer risk, with the lowest quartile serving as the referent. Totest for linear trend, a variable with values equal to themedian among controls for each quartile was treated as con-tinuous and tested in the model. When cases were stratifiedaccording to disease aggressiveness, each case group was

y

g

y

pj

j

g


3/10

Serum Lycopene, Other Carotenoids, and Prostate Cancer 1025


compared with the controls using polychotomous logisticregression (14).

To evaluate confounding, we screened suspected prostatecancer risk factors, as well as other potential confoundersidentified in the literature. These factors included familyhistory of prostate cancer, personal history of benign prosta-tic hyperplasia, personal history of vasectomy, daily alcohol

consumption, current smoking status, number of cigarettessmoked per day, Quetelets index, daily energy intake,intake of foods high in animal fat, education, income, everversus never married, month of blood draw, and serum cho-lesterol. For each carotenoid, potentially confounding vari-ables were added one at a time to models adjusted only forstudy design factors. The variable was considered a con-founder if, upon addition to the model, all carotenoid oddsratios shifted in a consistent direction and the proportionalchange for at least one level of exposure exceeded 10 per-cent. Confounding was further evaluated by using forwardand backward modeling in which potentially confoundingvariables were added/subtracted sequentially. The month of

blood draw was the only confounder identified. Therefore,all models were adjusted for the study design factors (agestratified into 10-year categories, race, study center) andmonth of blood draw stratified into 2-month categories.Interaction was assessed by both examining odds ratiosacross serum carotenoid levels and statistical significancetesting of multiplicative interaction terms.

All statistical tests were two tailed with 0.05. TheSAS system (15) was used for all analyses except the poly-chotomous logistic regression modeling for whichSUDAAN software (16) was used to perform maximumlikelihood inference.

Third National Health and Nutrition Examination Surveyanalyses

To explore differences between US Blacks and Whites,we examined serum lycopene concentrations for 990 Blackmen and 2,826 White men aged 4079 years who were par-ticipating in the Third Health and Nutrition ExaminationSurvey (NHANES III) (17). Individual carotenoids in serumhad been analyzed at the Centers for Disease Control andPrevention by high-performance liquid chromatographymethods comparable with those used in the present study.Age-specific medians were calculated for men of both races.In addition, a statistical test for racial differences in the

mean serum lycopene concentration was performed usinganalysis of covariance. Age was included as a covariate, andthis surveys complex sampling and weighting scheme wastaken into account (18).

RESULTS

Of interviewed subjects asked to donate blood, participa-tion was nearly equal by race (59 percent for Blacks and 61percent for Whites). Subjects were more likely to participateif they were younger, lived in Atlanta versus Detroit or NewJersey, had a positive family history of prostate cancer,earned a higher income, or were more educated, while pat-

terns were not observed for the intake of fruit and vegetablesor lycopene-rich foods. Participation by cases versus con-trols did not vary across categories of these factors in con-sistent ways (table 1).

Among the subjects who gave blood, cases were similarto controls by race and, within each race, by study center,month of blood draw, and educational achievement (table

2). However, cases were slightly older than controls, espe-cially among Blacks (p 0.05).

All serum carotenoids were positively and statisticallysignificantly correlated with one another (table 3). Fruit andvegetable intake was positively, but weakly, correlated witheach serum carotenoid (all correlation coefficients 0.10)except lycopene (correlation coefficient (r) 0.02). There

TABLE 1. Participation rates by selected characteristics for

interviewed cases and controls asked to donate blood from a

US multicenter case-control study, 19861989

Participation rates

Cases (%) Controls (%)

Overall participation

RaceBlacks

Whites

Age (years)


4/10

1026 Vogt et al.


was a weak positive correlation between the intake oflycopene-rich foods and serum lycopene (r 0.12).Adjustment for serum cholesterol altered the correlationsvery little.

The median serum lycopene concentration was loweramong cases than controls for both races. In contrast, theconcentrations for all other carotenoids were higher amongcases compared with controls for both races (table 4). Onlythe case-control difference among Blacks for lutein/zeaxan-thin was statistically significant (p 0.04).

Odds ratios by serum carotenoid quartile, as well as testfor trend p values, are presented in table 5 for all cases com-bined and for cases stratified by disease aggressiveness.When all cases were combined, every carotenoid was posi-tively, but nonsignificantly, associated with prostate cancer

risk except lycopene, which was inversely associated (testfor trend, p 0.09). Among the carotenoids with positiveassociations, the strongest effects were observed for -carotene and lutein/zeaxanthin. The inverse association withlycopene was particularly apparent among those withaggressive disease (comparing the highest with lowest quar-tiles, odds ratio (OR) 0.37, 95 percent confidence interval(CI): 0.15, 0.94; test for trend, p 0.04). Otherwise, strik-ing differences between nonaggressive and aggressive dis-ease were not observed. Associations were not substantiallyaltered when fully adjusted for all potential confounders; forexample, for all cases, the odds ratios by increasing quartilefor -carotene 1.00, 1.59, 1.49, and 2.09; the odds ratios

by increasing quartile for lycopene 1.00, 0.90, 0.73, and0.59. However, adjusting for total carotenoids (excludinglycopene) strengthened the inverse association withlycopene for all cases combined (odds ratios by increasingquartile 1.00, 0.86, 0.62, and 0.48; test for trend, p 0.01).

Race-specific models revealed that the associationbetween serum lycopene and the risk of prostate cancer wassimilar for Blacks and Whites with apparent, although notstatistically significant, inverse trends seen for each race(table 6). The positive associations with -carotene andlutein/zeaxanthin, though unstable, appeared stronger inBlacks. For other carotenoids, individual odds ratios fluctu-ated widely, and it was difficult to evaluate differences byrace. None of the tests for interaction between race and indi-

vidual serum carotenoids was statistically significant (datanot shown).

In a previously published analysis using the full studypopulation of 981 cases and 1,315 controls, the intake offoods high in animal fat was identified as the dietary factormost predictive of prostate cancer risk (20). Thus, wedecided a posteriori to examine the joint effects on prostatecancer risk of serum lycopene and animal fat intake in thesubgroup analyzed for serum carotenoids (table 7).Compared with those with low serum lycopene concentra-tions and high intake of foods high in animal fat (cutpointswere based on the medians among controls), the odds ratioswere statistically significantly decreased for men with high

TABLE 2. Description by selected characteristics of Black cases and controls and White cases and

controls from a US multicenter case-control study, 19861989

Blacks (%)

Cases(n= 99)

Controls(n= 108)

Cases(n= 110)

Age (years)

4049

505960697079

Study centerAtlanta, Georgia

Detroit, MichiganNew Jersey

Month of blood drawJanuaryFebruaryMarchAprilMayJuneJulyAugust

SeptemberOctoberNovemberDecember

Highest education attainedGrades 08Grades 911Grade 12

Some college/professional school

First degree relative with prostate cancerPositiveNegative

1

293633*

37

2438

20102723

127

462117

15

694

10

263231

3831

31

18242416

711

342921

16

298

10

293229

2935

36

1321232410

10

10123246

694

15

302926

3338

29

13321614

169

131321

53

397

Controls(n= 120)

Whites (%)

* p< 0.05 for Black cases compared with Black controls.

y

g

y

pj

j

g


5/10



serum lycopene concentrations and low animal fat intake(OR 0.40, 95 percent CI: 0.22, 0.74). The interactionbetween these two variables was not statistically significant(p 0.54), indicating independence of the association ofeach factor with prostate cancer risk.

Effect modification was also evaluated for age (


6/10

1028 Vogt et al.


(21) and the other evaluating serum lycopene levels in acohort of Washington County, Maryland, residents (22),noted that higher lycopene concentrations were associatedwith decreased risk of prostate cancer. Relative risks com-paring the highest with lowest quantiles were 0.60 (95 per-

cent CI: 0.37, 0.97) and 0.50 (95 percent CI: 0.20, 1.29),respectively. In 1995, Giovannucci et al. (23) reported thatthe risk of prostate cancer in a cohort of male health profes-sionals was significantly reduced by 35 percent among menwith the highest, relative to lowest, intake of tomatoes andtomato-based products. This inverse association was partic-ularly strong among those with more aggressive disease, forwhom risk was reduced by over 50 percent. More recently,data from a cohort developed from the Physicians HealthStudy demonstrated an inverse relation with baseline levelsof serum lycopene, which was also especially strong andstatistically significant for cases with aggressive disease (44percent decrease in risk comparing the highest with lowest

quintiles) (19). Conversely, a prospective study conductedamong Japanese-American men living in Hawaii did notprovide evidence for an inverse association with serumlycopene (24) nor did a Dutch cohort study assessing tomatointake (25). It should be noted, however, that exposure to

lycopene in these two study populations was substantiallylower than in other studies reporting an inverse association.Several case-control studies also examined this relation withone reporting a statistically significant inverse associationwith plasma lycopene (26). Among case-control studiesusing dietary estimates of lycopene exposure, some sug-gested an inverse association with risk (2730), while othersfound no evidence for an association in either direction (20,3135).

Our study included comparable numbers of Black menand White men, thus permitting race-specific analyses andleading to the new finding that higher serum lycopene con-centrations are similarly associated with reduced risk of

TABLE 5. Prostate cancer odds ratios for all cases and for nonaggressive and aggressive cases from a US multicenter

case-control study, 19861989

Quartile of serum carotenoid

1 (low) 4 (high)2

Range Odds ratio RangeRange Range Odds ratioOdds ratio

(reference)Odds ratio

Test fortrend,

pvalue

a-Carotene (mg/dl)All casesNonaggressive

casesAggressive

cases

b-Carotene (mg/dl)All casesNonaggressive

casesAggressive

cases

b-Cryptoxanthin(mg/dl)

All casesNonaggressive

casesAggressive

cases

Lutein/zeaxanthin(mg/dl)

All casesNonaggressive

casesAggressive

cases

Lycopene (mg/dl)All casesNonaggressive

casesAggressive

cases

0.01.4

0.38.2

0.34.9

3.614.1

0.510.7

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.53.2

8.314.8

5.07.1

14.219.6

10.817.1

1.40

1.08

1.65

1.56

1.16

1.57

1.39

1.09

2.03

1.11

1.04

0.91

0.97

1.05

0.93

3.34.7

14.923.1

7.211.1

19.726.6

17.224.7

1.29

1.30

1.01

1.54

1.48

1.34

1.29

1.27

1.57

1.03

1.35

0.57

0.74

0.72

0.79

4.829.4

23.2117.5

11.244.3

26.780.8

24.857.4

1.24

0.95

1.91

1.64

1.54

1.61

0.98

0.93

1.22

1.51

1.32

1.73

0.65

0.79

0.37*

0.60

0.91

0.17

0.22

0.20

0.41

0.65

0.82

0.84

0.17

0.36

0.10

0.09

0.36

0.04

3

* p< 0.05. Quartile cutpoints were based on the distribution for each exposure among controls. All models were adjusted for age, race, study center, and month of blood draw. Among cases, distributions by stage and grade, respectively, were as follows: 146 localized, 23 regional, and 21 distant; 75 well differentiated, 70

moderately differentiated, 38 poorly differentiated, and one undifferentiated. Stage and grade were combined to form categories of disease aggressiveness. Afterthe exclusion of 33 subjects because of missing information on grade and/or stage, nonaggressive disease included 111 cases with well- or moderately differen-tiated grade and localized stage. Aggressive disease included 65 cases with poorly differentiated to undifferentiated grade and/or regional to distant stage.Thissystem of categorization seeks to distinguish between disease that is more versus less likely to progress and become fatal (Gann et al., Cancer Res1999;59:122530).

y

g

y

pj

j

g


7/10



FIGURE 1. Median serum lycopene concentrations by age and race among males from the Third National Health and Nutrition ExaminationSurvey, 19881994.

TABLE 6. Prostate cancer odds ratios for Blacks and Whites from a US multicenter case-control study, 19861989

Quartile of serum carotenoid

1 (low) 4 (high)2

Range Odds ratio RangeRange Range Odds ratioOdds ratio Odds ratio

Test fortrend,

pvalue

a-Carotene (mg/dl)

BlacksWhites

b-Carotene (mg/dl)BlacksWhites

b-Cryptoxanthin(mg/dl)

BlacksWhites

Lutein/zeaxanthin(mg/dl)

BlacksWhites

Lycopene (mg/dl)BlacksWhites

0.01.4

0.38.2

0.34.9

3.614.1

0.510.7

1.001.00

1.001.00

1.001.00

1.001.00

1.001.00

1.53.2

8.314.8

5.07.1

14.219.6

10.817.1

1.491.54

2.79*1.09

1.741.00

1.741.00

0.880.96

3.34.7

14.923.1

7.211.1

19.726.6

17.224.7

1.191.40

1.421.87

2.020.99

1.510.83

0.670.69

4.829.4

23.2117.5

11.244.3

26.780.8

24.857.4

1.431.09

2.43*1.20

1.130.91

2.031.53

0.730.59

0.530.95

0.220.57

0.890.64

0.180.45

0.360.17

3

* p< 0.05. Quartile cutpoints were based on the distribution for each exposure among controls for Blacks and Whites combined. All models were adjusted for age, study center, and month of blood draw.

TABLE 7. Joint effects of serum lycopene and intake of foods high in animal fat from a US multicenter case-control study,

19861989

Serum lycopene concentration

Odds ratio95%

confidenceinterval

No. of cases/No. of controls

22 times/week


8/10

1030 Vogt et al.


prostate cancer for men of both races. In the analysis ofdietary data from our study, there was no evidence of aninverse association between lycopene and the risk ofprostate cancer in Black men or White men (20). However,commonly consumed sources of lycopene, such as pizza andlasagna, were not included in the questionnaire, and the cor-relation between intake of lycopene-rich foods and serum

lycopene was only 0.12. The only other study to address thisassociation in Blacks and Whites separately was a recentmultiethnic case-control study that reported an inverse asso-ciation between the dietary intake of cooked tomatoes andthe risk of prostate cancer in Blacks but not Whites (32).

Although our data did not suggest that risk patterns var-ied by race, the median serum lycopene concentrations wereapproximately 18 percent lower among Black controls com-pared with White controls. Similarly, median values forBlack men from NHANES III were 12 percent lower thanthose for White men within the same age range as our con-trols (19.9 vs. 22.6 g/dl, respectively). Lycopene intakewas also statistically significantly lower among US Blacks

compared with Whites in an analysis of the 1987 NationalHealth Interview Survey data (36). Given lycopenes poten-tially protective role, Blacks appear to be at a disadvantage,raising the possibility that differences in lycopene concen-trations could contribute to a portion of the racial disparityin prostate cancer incidence. These observations warrantfurther investigation.

In contrast to our lycopene results, increasing serum con-centrations of -carotene, -carotene, -cryptoxanthin, andlutein/zeaxanthin were either unassociated or positivelyassociated with increased risk of prostate cancer. Althoughnot statistically significant, these positive associations wereparticularly strong for -carotene and lutein/zeaxanthin.

Similar nonsignificant positive associations have been sug-gested in other observational studies using serologic data(19, 22, 24).

There is increasing evidence that foods high in fat, espe-cially fat from animal sources, may increase the risk ofprostate cancer (20, 37). The results of the present analysisadditionally suggest that the effects of animal fat and lycopeneare independent and that the greatest benefit may occur whenboth animal fat intake is reduced and lycopene intake isincreased. It is reassuring that these results are consistent withUS Department of Agriculture recommendations suggestingplentiful and varied consumption of fruits and vegetables andreduced intake of high fat meat and dairy products (38).

In addition to the inclusion of comparable numbers ofBlack participants and White participants, our study hadseveral strengths. We estimated carotenoid exposure bymeasuring blood carotenoid concentrations. In a recentreport by Freeman et al. (39), individual carotenoids mea-sured in the blood, but not in the diet, were strongly and sta-tistically significantly correlated with prostate tissue levels(for lycopene, r 0.56 and 0.06 for blood and diet, respec-tively). Our laboratory reproducibility was excellent withcoefficients of variation generally 8 percent or less forblinded quality control materials. The potential for degrada-tion during sample collection and storage seemed minimalbecause the mean concentrations of serum lycopene, the

most labile of the individual carotenoids, were similar in ourcontrols and those in NHANES III (21.18 mg/dl and 22.85mg/dl, respectively), using race-specific age standardiza-tion. In an effort to determine whether the cases included inthis analysis were representative of incident prostate canceracross the United States, we compared the distribution ofdisease characteristics with national cancer statistics col-

lected by the Surveillance, Epidemiology, and End ResultsProgram. For each race, the case distribution by stage wassimilar to that of the national data. However, a slightlyhigher proportion of both Black cases and White cases haddisease that was well differentiated compared withSurveillance, Epidemiology, and End Results data (data notshown) (40, 41).

Our results suggest that low concentrations of serumlycopene may be more strongly associated with aggressiveprostate cancer. This finding implies that disease progres-sion may be especially susceptible to the protective effectsof lycopene. It is also possible that aggressive prostate can-cer itself reduced serum lycopene levels. To investigate this

possibility, we examined the mean serum lycopene concen-trations among cases by disease aggressiveness. Lycopenelevels decreased with increasingly aggressive grade andstage (19.6, 16.8, and 15.1 g/dl for well-differentiated,moderately differentiated, and poorly/undifferentiated dis-ease, respectively, test for trend, p 0.04; 17.9, 17.0, and13.7 g/dl for localized, regional, and distant disease,respectively, test for trend, p 0.03). However, because thispattern was not observed for any other carotenoid, andbecause stronger lycopene associations have been observedfor aggressive disease in prospective studies (19, 23), a biasdue to disease seems unlikely to explain our findings.

We also considered whether differential participation by

cases and controls in the blood collection phase of the studycould have distorted our results. Analyses indicated that par-ticipation was comparable for cases and controls in terms ofrace, age, study center, family history of prostate cancer,income, education, intake of fruit and vegetables, and intakeof lycopene-rich foods. Thus, we found no evidence thatparticipation bias could account for the findings.Additionally, we considered the possibility that cases couldhave altered their diet as a result of disease, potentiallyimpacting serum carotenoid levels. In fact, comparablenumbers of cases (22 percent) and controls (19 percent)reported major dietary changes in the 6 months prior toblood draw. Excluding these subjects served to modestly

strengthen the positive association with -carotene, forexample (ORs by increasing quartile 1.00, 1.62, 1.66, and1.76), but did not substantially alter the results for lycopene(ORs by increasing quartile 1.00, 0.80, 0.76, and 0.64).

Lycopene is unique among the common carotenoids inthat overall fruit and vegetable intake is not a reliable pre-dictor of blood lycopene concentration (42). Indeed, in ourstudy the correlation coefficient between serum lycopene andtotal fruit and vegetable intake was 0.02. Lycopene entersthe US diet in the form of tomato-based products (43).Absorption and, thus, serum concentrations are enhanced byheat processing and concurrent fat intake (44). Items such asspaghetti, lasagna, and pizza, as opposed to fresh tomatoes,

y

g

y

pj

j

g


9/10



may be the dominant sources of lycopene in the UnitedStates (45). Therefore, we believe that the inverse associationbetween serum lycopene and prostate cancer risk cannot eas-ily be attributed to the increased fruit and vegetable intaketypical of healthy diets. Furthermore, in a recent validationstudy, scores from an index of healthy eating were signifi-cantly correlated with all plasma carotenoid concentrations

with the exception of lycopene (46).A protective role for lycopene in prostate cancer etiology

is biologically plausible. Relative to other carotenoids,lycopene has exceptionally high singlet oxygen-quenchingability (47). In this way, lipids, nucleic acids, and proteinsmay be protected from oxidative damage that could other-wise lead to cancer. It is also possible that lycopene, inde-pendent of its singlet oxygen-quenching abilities, enhancescellular differentiation, inhibits cell proliferation, maintainsintercellular communication, interacts with growth factors,or influences carcinogenesis through other mechanisms (8,48, 49).

In summary, our findings provide support for the protec-

tive role of lycopene in prostate carcinogenesis. Because weincluded large numbers of US Blacks, these results can nowbe extended to include this racial subgroup. Although we didnot find evidence that the association varied for Black menand White men, it is provocative that racial differences inserum lycopene concentrations were observed, both amongour controls and in NHANES III. Thus, differences inlycopene intake or metabolism may contribute to the racialdisparity in prostate cancer incidence. We also found that theinverse relation with lycopene was particularly pronouncedfor aggressive disease. Hypotheses that other individualcarotenoids are associated with a reduced risk of prostatecancer were not supported. Our lycopene findings are

encouraging because the etiology of prostate cancer, the sec-ond leading cause of cancer mortality in US men, remainselusive, with few modifiable strategies.

ACKNOWLEDGMENTS

This study was funded, in part, through National CancerInstitute contracts to the Michigan Cancer Foundation(NO1-CP-5109 and NO1-CN-05225), the New Jersey StateDepartment of Health (NO1-CP-51089 and NO1-CN-31022), the Georgia Center for Cancer Statistics (NO1-CP-

51092 and NO1-CN-05227), and Westat, Inc. (NO1-CP-51087).

The authors thank Dr. Joseph L. Gastwirth for his helpfulsuggestions.

REFERENCES

1. Ries LAG, Eisner MP, Kosary CL, et al, eds. SEER cancerstatistics review, 19731997. Bethesda, MD: National CancerInstitute, 2000.

2. Zaridze DG, Boyle P, Smans M. International trends in prosta-tic cancer. Int J Cancer 1984;33:22330.

3. Shimizu H, Ross RK, Bernstein L, et al. Cancers of theprostate and breast among Japanese and white immigrants inLos Angeles County. Br J Cancer 1991;63:9636.

4. Lilienfeld AM, Levin ML, Kessler II. Cancer in the UnitedStates. (APHA monograph). Cambridge, MA: HarvardUniversity Press, 1972.

5. Mayne ST. Beta-carotene, carotenoids, and disease preventionin humans. FASEB J 1996;10:690701.

6. van Poppel G, Goldbohm RA. Epidemiologic evidence forbeta-carotene and cancer prevention. Am J Clin Nutr 1995;62(suppl):1393S402S.

7. Ziegler RG. Vegetables, fruits, and carotenoids and the risk ofcancer. Am J Clin Nutr 1991;53(suppl):251S9S.

8. Giovannucci E. Tomatoes, tomato-based products, lycopene,and cancer: review of the epidemiologic literature. J NatlCancer Inst 1999;91:31731.

9. Waksberg J. Sampling methods for random digit dialing. J AmStat Assoc 1978;73:406.

10. Swanson CA, Gridley G, Greenberg R, et al. A comparison ofdiets of Blacks and Whites in three areas of the United States.Nutr Cancer 1993;20:15365.

11. Sowell AL, Huff DL, Gunter EW, et al. Identification of cis-carotenoids in human sera analyzed by reversed-phase high-performance liquid chromatography with diode array detection.J Chromatogr 1988;431:42430.

12. Westgard JO, Barry PL, Hunt MR, et al. Amulti-rule Shewhartchart for quality control in clinical chemistry. Clin Chem 1981;27:493501.

13. Breslow NE, Day NE, eds. Statistical methods in cancerresearch. Vol 1. The analysis of case-control studies. Lyon,France: International Association for Research on Cancer,1980. (IARC scientific publication no. 32).

14. Hosmer DW Jr, Lemeshow S. Applied logistic regression. NewYork, NY: Wiley-Interscience, 1989.

15. SAS Institute, Inc. SAS/STAT users guide, version 6. 4th ed.Vols 1 and 2. Cary, NC: SAS Institute, Inc, 1989.

16. Shah BV, Barnwell BG, Bieler GS. SUDAAN users manual,release 7.5. Research Triangle Park, NC: Research TriangleInstitute, 1997.

17. National Center for Health Statistics. Third National Healthand Nutrition Examination Survey, 19881994. NHANES III

laboratory data file (CD-ROM). Hyattsville, MD: Centers forDisease Control and Prevention, 1996. (Public use data filedocumentation no. 76200).

18. Korn EL, Graubard BI. Analysis of health surveys. New York,NY: Wiley-Interscience, 1999.

19. Gann PH, Ma J, Giovannucci E, et al. Lower prostate cancerrisk in men with elevated plasma lycopene levels: results of aprospective analysis. Cancer Res 1999;59:122530.

20. Hayes RB, Ziegler RG, Gridley G, et al. Dietary factors and risksfor prostate cancer among Blacks and Whites in the UnitedStates. Cancer Epidemiol Biomarkers Prev 1999;8:2534.

21. Mills PK, Beeson WL, Phillips RL, et al. Cohort study of diet,lifestyle, and prostate cancer in Adventist men. Cancer 1989;64:598604.

22. Hsing AW, Comstock GW, Abbey H, et al. Serologic precur-sors of cancer. Retinol, carotenoids, and tocopherol and risk of

prostate cancer. J Natl Cancer Inst 1990;82:9416.23. Giovannucci E, Ascherio A, Rimm EB, et al. Intake of

carotenoids and retinol in relation to risk of prostate cancer. JNatl Cancer Inst 1995;87:176776.

24. Nomura AMY, Stemmermann GN, Lee J, et al. Serummicronutrients and prostate cancer in Japanese Americans inHawaii. Cancer Epidemiol Biomarkers Prev 1997;6:48791.

25. Schuurman AG, Goldbohm RA, Dorant E, et al. Vegetable andfruit consumption and prostate cancer risk: a cohort study inthe Netherlands. Cancer Epidemiol Biomarkers Prev 1998;7:67380.

26. Lu QY, Hung JC, Heber D, et al. Inverse associations betweenplasma lycopene and other carotenoids and prostate cancer.Cancer Epidemiol Biomarkers Prev 2001;10:74956.

27. Schuman LM, Mandal JS, Radke A, et al. Some selected fea-tures of the epidemiology of prostate cancer: Minneapolis-St.

y

g

y

pj

j

g


10/10

1032 Vogt et al.


Paul, Minnesota case-control study, 19761979. In: MagnusK, ed. Trends in cancer incidence: causes and practical impli-cations. Washington, DC: Hemisphere Publishing Corp, 1982:34554.

28. Tzonou A, Signorello LB, Lagiou P, et al. Diet and cancer ofthe prostate: a case-control study in Greece. Int J Cancer 1999;80:7048.

29. Norrish AE, Jackson RT, Sharpe SJ, et al. Prostate cancer and

dietary carotenoids. Am J Epidemiol 2000;151:11923.30. Jain MG, Hislop GT, Howe GR, et al. Plant foods, antioxi-dants, and prostate cancer risk: findings from case-controlstudies in Canada. Nutr Cancer 1999;34:17384.

31. Key TJA, Silcocks PB, Davey GK, et al. A case-control studyof diet and prostate cancer. Br J Cancer 1997;76:67887.

32. Kolonel LN, Hankin JH, Whittemore AS, et al. Vegetables,fruits, legumes and prostate cancer: a multi-ethnic case-controlstudy. Cancer Epidemiol Biomarkers Prev 2000;9:795804.

33. Cohen JH, Kristal AR, Stanford JL. Fruit and vegetable intakesand prostate cancer risk. J Natl Cancer Inst 2000;92:618.

34. Villeneuve PJ, Johnson KC, Kreiger N, et al. Risk factors forprostate cancer: results from the Canadian National EnhancedCancer Surveillance System. Cancer Causes Control 1999;10:35567.

35. Le Marchand L, Hankin JH, Kolonel LN, et al. Vegetable andfruit consumption in relation to prostate cancer risk in Hawaii:a reevaluation of the effect of dietary -carotene. Am JEpidemiol 1991;133:21519.

36. Nebeling LC, Forman MR, Graubard BI, et al. The impact oflifestyle characteristics on carotenoid intake in the UnitedStates: the 1987 National Health Interview Survey. Am JPublic Health 1997;87:26871.

37. Kolonel LN. Nutrition and prostate cancer. Cancer CausesControl 1996;7:8394.

38. US Department of Agriculture, US Department of Health andHuman Services. Nutrition and your health: dietary guidelinesfor Americans. 5th ed. Hyattsville, MD: US Department of

Agriculture, 2000. (Home and garden bulletin no. 232).39. Freeman VL, Meydani M, Yong S, et al. Prostatic levels of

tocopherols, carotenoids, and retinol in relation to plasmalevels and self-reported usual dietary intake. Am J Epidemiol2000;151:10918.

40. Ries LAG, Kosary CL, Hankey BF, et al, eds. SEER cancerstatistics review, 19731996. Bethesda, MD: National CancerInstitute, 1999.

41. Stanford JL, Stephenson RA, Coyle LM, et al. Prostate cancertrends 19731995, SEER Program. Bethesda, MD: NationalCancer Institute, 1999. (NIH publication no. 99-4543).

42. Campbell DR, Gross MD, Martini MC, et al. Plasmacarotenoids as biomarkers of vegetable and fruit intake. CancerEpidemiol Biomarkers Prev 1994;3:493500.

43. Chug-Ahuja JK, Holden JM, Forman MR, et al. The develop-ment and application of a carotenoid database for fruits, veg-etables, and selected microcomponent foods. J Am Diet Assoc1993;93:31823.

44. Stahl W, Sies H. Uptake of lycopene and its geometrical iso-mers is greater from heat-processed than from unprocessedtomato juice in humans. J Nutr 1992;122:21616.

45. Mayne ST, Cartmel B, Silva F, et al. Plasma lycopene concen-trations in humans are determined by lycopene intake, plasmacholesterol concentrations and selected demographic factors. JNutr 1999;129:84954.

46. Hann CS, Rock CL, King I, et al. Validation of the HealthyEating Index with use of plasma biomarkers in a clinical sam-ple of women. Am J Clin Nutr 2001;74:47986.

47. Di Mascio P, Kaiser S, Sies H. Lycopene as the most efficientbiological carotenoid singlet oxygen quencher. Arch BiochemBiophys 1989;274:5328.

48. Gerster H. The potential role of lycopene for human health. JAm Coll Nutr 1997;16:10926.

49. Mucci LA, Tamimi R, Lagiou P, et al. Are dietary influenceson the risk of prostate cancer mediated through the insulin-likegrowth factor system? BJU Int 2001;87:81420.

y

g

y

pj

j

g

Carotenoids and the Prostate

Documents

Transcript of Carotenoids and the Prostate