lable at ScienceDirect

Annals of Epidemiology 24 (2014) 475e478

Contents lists avai

Annals of Epidemiology

journal homepage: www.annalsofepidemiology.org

Brief communication

Oxidative balance score and risk for incident prostate cancer in a prospectiveU.S. cohort study

Sindhu Lakkur PhD a,d,*, Michael Goodman MDa,b,e, Roberd M. Bostick MD a,b,e, Jessica Citronberg MPHb,William McClellan MDb, William Dana Flanders MDb,e, Suzanne Judd PhD d, Victoria L. Stevens PhD c

aDepartment of Nutrition, Emory University, Atlanta, GAbDepartment of Epidemiology, Emory University, Atlanta, GAc Epidemiology Research Program, American Cancer Society, Atlanta, GAdDepartment of Biostatistics, The University of Alabama at Birmingham, BirminghameWinship Cancer Institute, Emory University, Atlanta, GA

a r t i c l e i n f o

Article history:Received 12 September 2013Accepted 28 February 2014Available online 21 March 2014

Keywords:Oxidative stressProstatic neoplasmsCohort study

* Corresponding author. The University of AlabamaBoulevard, Ryals Public Health Bldg, Suite 514C, Birm(205) 975 9222; fax: þ1 (205) 975 2540.

E-mail address: slakkur@uab.edu (S. Lakkur).

1047-2797/$ e see front matter � 2014 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.annepidem.2014.02.015

a b s t r a c t

Purpose: Oxidative stress is defined as an imbalance between pro-oxidants and antioxidants. Previousresearch found that a single comprehensive oxidative balance score (OBS) that includes individual pro-and anti-oxidant exposures may be associated with various conditions (including prostate cancer) in theabsence of associations with the individual factors. We investigated an OBS-incident prostate cancerassociation among 43,325 men in the Cancer Prevention Study II Nutrition Cohort.Methods: From 1999e2007, 3386 incident cases were identified. Twenty different components, used intwo ways (unweighted or weighted based on literature reviews), were incorporated into the OBS, and theresulting scores were then expressed as three types of variables (continuous, quartiles, or six equal in-tervals). Multivariable-adjusted rate ratios were calculated using Cox proportional hazards models.Results: We hypothesized that the OBS would be inversely associated with prostate cancer risk; however,the rate ratios (95% confidence intervals) comparing the highest with the lowest OBS categories rangedfrom 1.17 (1.04e1.32) to 1.39 (0.90e2.15) for all cases, 1.14 (0.87e1.50) to 1.59 (0.57e4.40) for aggressivedisease (American Joint Committee on Cancer stage III/IV or Gleason score 8e10), and 0.91 (0.62e1.35) to1.02 (1.02e1.04) for nonaggressive disease.Conclusions: Our findings are not consistent with the hypothesis that oxidative balanceerelated expo-sures collectively affect risk for prostate cancer.

� 2014 Elsevier Inc. All rights reserved.

Introduction

Prostate cancer is the second leading cause of cancer deathamong men in the United States [1]. Migration studies [2] stronglysuggest that environmental factors play a role in prostate carcino-genesis. One proposed mechanistic link is oxidative stress [3].Oxidative stress is an imbalance in pro- and anti-oxidants, whichresults in macromolecular damage and disruption of redoxsignaling and control [4]; this can lead to mutagenesis and subse-quent carcinogenesis [5].

Environmental exposures such as smoking and alcohol con-sumption act as pro-oxidants and can increase reactive oxygenspecies (ROS) generation and inflammation [6]. Antioxidants, such

Birmingham, 1665 Universityingham, AL 35294. Tel: þ1

ll rights reserved.

as vitamin E, act as free radical scavengers [7]. Other nutrientsindirectly reduce oxidative stress through anti-inflammatoryproperties (omega-3 fatty acids) [8].

Although antioxidants can reduce cell proliferation and oxida-tive DNA damage in vivo and in vitro, findings from observationalepidemiologic studies that investigated individual antioxidants andprostate cancer risk have been inconsistent [7], and clinical trials ofantioxidants as chemopreventive agents have produced null results[9]. Other studies found that nutrients act synergistically, andcombined factors can be more strongly associated with disease riskthan are individual nutrients [10] leading us to believe that multiplepro- and antioxidant exposures incorporated into a compositemeasure of oxidative balance may be more strongly associated withprostate cancer risk than would be any single factor [11].

In the present study, in follow-up to our pilot case-control studyin which we found an inverse association between oxidative bal-ance score (OBS) and prostate cancer risk [11], we investigated theassociation in a large prospective U.S. cohort study.

S. Lakkur et al. / Annals of Epidemiology 24 (2014) 475e478476

Methods

The men in this analysis were participants in the Cancer Pre-vention Study II Nutrition Cohort, designed to assess associations ofdietary and lifestyle factors with cancer incidence [12]. For thepresent study, follow-up was from 1999 until the date of diagnosisof prostate cancer, death, date of the last returned survey, or June30, 2007. Our analysis begins with 1999 because this was when amore comprehensive 152-item semiquantitative modified WilletFood Frequency Questionnaire (FFQ) was administered. Participantsreported diet in the previous year. Follow-up questionnaires weresent every 2 years to update exposure information. After excludingparticipants with no dietary information (n ¼ 9518), were lost tofollow-up (n ¼ 1570), had a history of prostate cancer (n ¼ 6100) orcancer other than nonmelanoma skin cancer (n ¼ 5468) at before1999, had incomplete or improbable FFQ data or did not reportlifestyle OBS components (n ¼ 3668), and those with unverifiedself-reported prostate cancer (n ¼ 82), a total of 43,325 men wereavailable for analysis. We identified and verified 3386 incidentcases of prostate cancer between 1999 and June 30, 2007. Thisincluded cases verified via medical records (n ¼ 2720) or linkage tocancer registry data (n ¼ 666).

The OBS comprised 20 components, selected based on a prioriknowledge about their relation to oxidative stress. These includeddietary antioxidants alpha- and beta-carotene, beta-cryptoxanthin,zinc, lutein, lycopene, vitamins C and E, omega-3 fatty acids, fla-vonoids and glucosinolates, and selenium (supplements); dietarypro-oxidants omega-6 fatty acids, iron, and saturated fat, as well aslifestyle or medication-related antioxidant (physical activity andnonsteroidal anti-inflammatory drugs use) and pro-oxidant(smoking, alcohol and overweight/obesity) factors. The OBScomponent assignment scheme is shown in Supplementary Table 1.The points assigned to each component were summed to create theoverall OBS. In the analyses of associations with prostate cancer, thescore was used in several ways as (1) a continuous variable, (2)quartiles, and (3) equal interval categories. The cutoffs weredetermined using the distribution of the OBS within the analyticalcohort (specific cutoffs in Tables 2 and 3). Assessing the OBS usingquartiles and equal interval categories allowed for examination ofparticularly low and high scores.

OBS components were included in the score using the followingtwo approaches: (1) equal weights and (2) weighted according tothe reported associations with prostate cancer risk. For each OBScomponent, we identified comprehensive reviews ormeta-analysesthat summarized the magnitude of their associations with prostatecancer. If no published reviews were found, we conducted our ownmeta-analysis of published studies of nutrients and prostate cancer.

Table 1Baseline characteristics of Cancer Prevention Study II men (n ¼ 43,325) by OBS quartile

Characteristics OBS quartile 1

Age at 1999 interview, y, mean (SD) 69.84 (5.69)Caucasian (%) 98.15College education or higher (%) 36.83Family history of prostate cancer in a first-degree

relative (%)* missing 568014.06

Current cholesterol-lowering drug use (%) 23.03Prostate-specific antigen screening (%) 77.20Total energy intake, mean (SD) (kcal/d) 1936.11 (604.96)Total calcium intake, mean (SD) (mg/d)y 752.93 (330.55)Total vitamin D intake, mean (SD) (IU/d)y 278.30 (201.74)Total folate intake, mean (SD) (mcg/d)y 439.36 (203.33)Finasteride use (%) 2.15

Note: All nutrients adjusted for total energy intake.* Self-reported history of recent Prostate-specific antigen screening.y Diet plus supplements.

The weights for pro-oxidants were the pooled adjusted relativerisks derived from published studies, whereas weights for theantioxidant OBS components were calculated using the inverseadjusted relative risks estimates (Supplementary Table 1).

Multivariable-adjusted rate (hazard) ratios (HRs) and their cor-responding 95% confidence intervals (CIs) were calculated usingCox proportional hazards models. All models were tested for pro-portional hazards assumption violations using the likelihood ratiotest. Models were examined for collinearity among independentvariables. Interactions between OBS and each covariate wereassessed using the likelihood ratio test. All analyses were conductedusing SAS statistical software version 9.2 (SAS institute, Cary, NC).

The association between the OBS and prostate cancer risk wasexamined using each OBS version after adjusting for age, total energyintake, total (dietary plus supplemental) calcium intake, totalvitamin D intake, total folate intake, race (white, black, and other),education (less than high school, high school graduate, some college,college graduate, and missing), family history of prostate cancer in afirst-degree relative (no and yes), cholesterol lowering drug use(never, former, current, and missing), finasteride use (never, former,current, and missing), and history of prostate cancer screening (ever,never, and unknown). These potential confounders were selectedbased on evidence in the literature and other a priori considerations.

Associations with the OBS were also examined separately foraggressive and nonaggressive prostate cancer, after adjusting for thesame covariates. Prostate cancerwas classified as aggressive based oneither an advanced stage (American Joint Committee on Cancer stageIII/IV) or high grade (Gleason score of 8e10) as reported previously[13]. In addition, a series of sensitivity analyses examined the impactof individual OBS components by removing each component fromthe score and controlling for it in the model as a covariate.

Results

The baseline characteristics of all participants by OBS quartileare presented in Table 1. There were no significant differences atbaseline in age, race, family history of prostate cancer in a first-degree relative, total energy intake, or finasteride use by OBSquartile. The proportion of participants whowere college educated,had undergone Prostate-specific antigen screening, and wascurrently using cholesterol-lowering drugs increased with anincreasing OBS score. Total calcium, vitamin D, and folate intakealso increased with an increasing OBS score.

Table 2 shows the association between prostate cancer risk andthe OBS with equally weighted and weighted components. Forequally weighted components, after adjusting for covariates, a

OBS quartile 2 OBS quartile 3 OBS quartile 4

70.22 (5.73) 70.45 (5.45) 70.45 (5.77)98.06 97.74 97.5748.11 55.38 64.8213.56 13.77 14.59

27.72 30.11 32.4582.50 85.65 87.881901.33 (578.14) 1892.49 (573.93) 1856.77 (551.58)857.43 (372.79) 946.68 (408.20) 1112.69 (473.24)363.89 (227.98) 427.07 (242.43) 502.37 (253.17)567.95 (239.81) 661.92 (260.78) 794.47 (280.46)2.55 2.73 3.20

Table 2Associations of OBS comprised equally weighted and literature-weighted components with incident prostate cancer in the Cancer Prevention Study II Nutrition Cohort(1999e2007)

OBS Categories Equally weighted OBS components Components weighted by literature reviews

Number ofcases/total

Age-adjusted HR(95% CI)

Multivariable HR*

(95% CI)Number ofcases/total

Age-adjusted HR(95% CI)

Multivariable HR*

(95% CI)

Continuous 3386/43,325 1.008 (1.003e1.012) 1.008 (1.002e1.013) 3386/43,325 1.007 (1.003e1.011) 1.007 (1.002e1.012)Quartilesy

1 810/11,570 1.00 1.00 758/10,831 1.00 1.002 919/11,678 1.12 (1.02e1.23) 1.12 (1.01e1.24) 843/10,830 1.11 (1.00e1.22) 1.11 (0.99e1.23)3 954/9442 1.12 (1.02e1.24) 1.12 (1.01e1.26) 875/10,832 1.14 (1.03e1.25) 1.14 (1.03e1.28)4 903/10,635 1.18 (1.08e1.30) 1.17 (1.04e1.32) 910/10,832 1.17 (1.06e1.29) 1.15 (1.03e1.30)Ptrend

x <.001 .01 .002 .02Equal intervalsz

1 92/1303 1.00 1.00 65/956 1.00 1.002 595/8586 0.97 (0.78e1.21) 0.98 (0.77e1.23) 536/7924 1.00 (0.77e1.29) 1.03 (0.78e1.36)3 1189/15,430 1.08 (0.87e1.33) 1.07 (0.85e1.34) 1188/15,141 1.15 (0.90e1.48) 1.19 (0.91e1.55)4 1073/12,955 1.14 (0.92e1.41) 1.15 (0.91e1.45) 1108/13,598 1.18 (0.92e1.51) 1.21 (0.92e1.60)5 403/4703 1.17 (0.943e1.46) 1.13 (0.88e1.46) 452/5318 1.22 (0.94e1.58) 1.22 (0.92e1.66)6 34/348 1.36 (0.91e2.01) 1.39 (0.90e2.15) 37/388 1.39 (0.93e2.08) 1.41 (0.90e2.21)Ptrend

x <.001 .01 <.001 .01

* Adjusted for age, race, sex, education, family history of prostate cancer in a first degree relative, prostate cancer screening, nonsteroidal anti-inflammatory drugs/aspirinuse, total calcium intake, total vitamin D intake, total energy intake, cholesterol-lowering drug use, and finasteride use.

y Quartile cutoffs: equally weighted (first¼ 8e25, second¼ 26e31, third¼ 32e36, and fourth¼ 36e55); weighted (first¼ 8.1e27.0, second¼ 27.1e33.2, third¼ 33.3e39.4,and fourth ¼ 39.5e60.0).

z Equal interval cutoffs: equally weighted (first ¼ 8e16, second ¼ 17e24, third ¼ 25e32, fourth ¼ 33e40, fifth ¼ 41e48, and sixth ¼ 49e55); weighted (first ¼ 8.1e16.9,second ¼ 16.9e25.7, third ¼ 25.8e34.5, fourth ¼ 34.6e43.3, fifth ¼ 43.3e52.1, and sixth ¼ 52.2e60.0).

x Ptrend assessed using category median.

S. Lakkur et al. / Annals of Epidemiology 24 (2014) 475e478 477

statistically significant, but modest, positive association was foundwhen the OBS was assessed as a continuous variable (HR ¼ 1.00895% CI, 1.002e1.013). Prostate cancer incidence was 17% higher inthe highest versus the lowest OBS quartiles (HR ¼ 1.17; 95% CI,1.04e1.32; Ptrend ¼ .01). A similar analysis using equal intervalcategories yielded a corresponding HR of 1.39 (95% CI, 0.90e2.15;Ptrend ¼ .007; Table 2).

The results of theweighted analyses were essentially identical tothose obtained without weighting: the HRs were 1.007 (95% CI,1.002e1.012) for the OBS treated as a continuous variable, 1.15 (95%CI, 1.03e1.30) when comparing the lowest with the highest OBSquartile (Ptrend¼ .02), and 1.41 (95% CI, 0.90e2.21) when comparingthe lowest with the highest category in the equal interval analyses(Ptrend ¼ 0.01).

Table 3 shows that the associations for aggressive and nonag-gressive tumors were similar to that for all prostate cancer casescombined (Table 2). The weighted results were not substantially

Table 3Associations of OBS comprised equally weighted components with incident aggressive* a(1999e2007)

OBS categories Aggressive

Number of cases/total Multivariable HRy (95

OBS continuous 657/43,325 1.010 (0.991e1.020)OBS quartiles8e25 169/11,570 1.0026e31 170/11,678 1.06 (0.84e1.35)32e36 144/9442 0.93 (0.72e1.20)36e55 174/10,635 1.14 (0.87e1.50)Ptrend

z .47OBS five categories8e16 21/1303 1.0017e26 169/12,116 1.00 (0.62e1.61)27e36 293/19,271 0.98 (0.61e1.58)37e46 164/9801 1.12 (0.67e1.86)47e55 10/834 1.59 (0.57e4.40)Ptrend

z .39

* Prostate cancer was classified as aggressive based on either an advanced stage (Amey Adjusted for age, race, sex, education, family history of prostate cancer in a first degr

use, total calcium intake, total vitamin D intake, total energy intake, physical activity, chz Ptrend assessed using category medians.

different from the unweighted results (data not shown). In thesensitivity analyses assessing the impact of individual OBS com-ponents, (by removing each component from the score andincluding it as a covariate) all Odds Ratio estimates were within 5%of the original model result (data not shown, available on request).

Discussion

The results from this prospective cohort study provide no evi-dence that an OBS is inversely associated with prostate cancer risk.In fact, the data suggest that persons with a higher OBS may be atincreased risk of developing (or being diagnosed with) prostatecancer. Separating aggressive and nonaggressive prostate cancercases did not appreciably change the observed associations. Theseresults are not consistent with the hypothesis that a presumablybeneficial balance of pro- and anti-oxidant exposures protectsagainst prostate carcinogenesis.

nd nonaggressive prostate cancer in the Cancer Prevention Study II Nutrition Cohort

Nonaggressive

% CI) Number of cases/total Multivariable HRy (95% CI)

2729/43,325 1.001 (0.995e1.007)

649/11,570 1.00744/11,678 1.06 (0.94e1.19)602/9442 1.00 (0.88e1.13)734/10,635 1.02 (1.02e1.04)

.85

71/1303 1.00686/12,116 0.79 (0.61e1.03)

1238/19,271 0.84 (0.64e1.09)674/9801 0.82 (0.62e1.09)60/834 0.91 (0.62e1.35)

.77

rican Joint Committee on Cancer stage III/IV) or high grade (Gleason score of 8e10).ee relative, prostate cancer screening, nonsteroidal anti-inflammatory drugs/aspirinolesterol-lowering drug use, and finasteride use.

S. Lakkur et al. / Annals of Epidemiology 24 (2014) 475e478478

Other studies examined the association between OBS and prostatecancer, with conflicting results. In theMarkers of Prostate Cancer casecontrol study, two different methods were used to measure OBScomponents, a FFQ-based method [11] and a combined FFQ andbiomarker-based method [14]. For both methods, a higher OBS wasinversely associated with prostate cancer, although a statisticallysignificant trend was found with the biomarker-based method[11,14]. In the large Canadian Study of Diet, Lifestyle, and Healthcohort study, Agalliu et al. [15] found no association between an OBSand prostate cancer risk with increasing OBS quintiles. Among thesethree studies and the present study, different categorizations of thecomponents for inclusion into the OBS were used, and, althoughunlikely, may be related to the discrepant findings across the studies.

Our study examined a more comprehensive list of OBS compo-nents than did previous studies, which did not include a-carotene,zinc, flavonoids, or glucosinolates, all of which may act as antioxi-dants [16,17]. Our OBS also included physical activity and bodymassindex. While strenuous physical activity increases ROS production(short term), moderate physical activity promotes antioxidant geneexpression by activating Nrf2 [18]. Obesity increases ROS produc-tion and adipocytokine expression, inducing inflammation, a causeand a consequence of oxidative stress [19,20]. A growing body ofevidence suggests that physical activity and obesity are potentialmodifiable risk factors for prostate cancer [21,22]. The Agalliu et al.and Goodman et al. studies used all polyunsaturated fats as a pro-oxidant, whereas we examined the various polyunsaturated fattyacids separately. Omega-3 fatty acids were categorized as antioxi-dants because they promote the transcription of antioxidantenzymes, whereas omega-6 fatty acids were considered pro-oxidants because they serve as precursors to proinflammatoryeicosanoids [23,24]. Furthermore, unlike in the Agalliu study, ouranalyses were adjusted for family history of prostate cancer, and incontrast to all of the above-referenced previous studies, we alsocontrolled for statin use.

The most distinguishing feature of this study was the number ofways the OBS was constructed and assessed. None of the previousstudies compared the results of a weighted and unweighted OBS inrelationwith prostate cancer risk. The unweighted OBS assumes thateach component is equally associated with prostate cancer. Althoughthe literature-based weights account for the different strengths ofassociation, the range of component weights was quite narrow(0.9e1.2), making the weighted and the unweighted results similar.

Theremay be othermethods of weighting OBS components suchas weighting OBS based on the effects of these components onoxidative stress measures. Such alternative weighting could bebased on associations of OBS components with biomarkers ofoxidation such as isoprostanes or disruption of thiol redox circuitsas measured by plasma levels of glutathione redox [25].

Other limitations of our study also should be considered.Carcinogenesis is a multiyear process, and it may be that the OBS atthe time of measurement may not be related to cancer risk.Although our OBS is the most comprehensive of those used in thethree previous prostate cancer studies, it included only extrinsicfactors. Intrinsic factors that affect oxidative stress, such as cellularantioxidant enzymes [17], were not included. Moreover, this cohortincluded a high proportion of health conscious participants, sug-gesting that therewas insufficient variability in lifestyle to detect anassociation between the OBS and prostate cancer. Additionally,participants with missing dietary data were excluded from thisanalysis, which may have led to possible selection bias.

Despite the study limitations and apparent inconsistency withmechanistic evidence, our study affirms that at a population level,pro- and anti-oxidant exposures are unlikely to explain differencesin prostate cancer incidence. Prostate cancer is a difficult disease tostudy, and after much research no modifiable, well-accepted risk

factors have been identified. This study, like many before,underscores the enigmatic nature of this disease.

References

[1] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin2012;62(1):10e29. PubMed PMID: 22237781. Epub 2012/01/13. eng.

[2] Muir CS, Nectoux J, Staszewski J. The epidemiology of prostatic cancer.Geographical distribution and time-trends. Acta Oncol 1991;30(2):133e40.PubMed PMID: 2029395. Epub 1991/01/01. eng.

[3] Khandrika L, Kumar B, Koul S, Maroni P, Koul HK. Oxidative stress in prostatecancer. Cancer Lett 2009;282(2):125e36. PubMed PMID: 19185987. PubmedCentral PMCID: 2789743.

[4] Sies H, Jones D. Oxidative stress. In: Fink G, editor. Encyclopedia of stress. 2nded. Elsevier; 2007. pp. 45e8.

[5] Dreher D, Junod AF. Role of oxygen free radicals in cancer development. Eur JCancer 1996;32A(1):30e8. PubMed PMID: 8695238. Epub 1996/01/01. eng.

[6] Wu D, Cederbaum AI. Alcohol, oxidative stress, and free radical damage.Alcohol Res Health: J Natl Inst Alcohol Abuse Alcohol 2003;27(4):277e84.PubMed PMID: 15540798. Epub 2004/11/16. eng.

[7] Willis MS, Wians FH. The role of nutrition in preventing prostate cancer: areview of the proposed mechanism of action of various dietary substances.Clinica Chim Acta Int J Clin Chem 2003;330(1-2):57e83. PubMed PMID:12636926. Epub 2003/03/15. eng.

[8] Calder PC. n-3 polyunsaturated fatty acids, inflammation, and inflammatorydiseases. Am J Clin Nutr 2006;83(6 Suppl):1505Se19S. PubMed PMID:16841861. Epub 2006/07/18. eng.

[9] Jiang L, Yang KH, Tian JH, Guan QL, Yao N, Cao N, et al. Efficacy of antioxidantvitamins and selenium supplement in prostate cancer prevention: a meta-analysis of randomized controlled trials. Nutr Cancer 2010;62(6):719e27.PubMed PMID: 20661819. Epub 2010/07/28. eng.

[10] Slattery ML, Boucher KM, Caan BJ, Potter JD, Ma KN. Eating patterns and risk ofcolon cancer. Am J Epidemiol 1998;148(1):4e16. PubMed PMID: 9663397.Epub 1998/07/15. eng.

[11] Goodman M, Bostick RM, Dash C, Flanders WD, Mandel JS. Hypothesis:oxidative stress score as a combined measure of pro-oxidant and antioxidantexposures. Ann Epidemiol 2007;17(5):394e9. PubMed PMID: 17462547. Epub2007/04/28. eng.

[12] Calle EE, Rodriguez C, Jacobs EJ, Almon ML, Chao A, McCullough ML, et al. TheAmerican Cancer Society Cancer Prevention Study II Nutrition Cohort: ratio-nale, study design, and baseline characteristics. Cancer 2002;94(9):2490e501.PubMed PMID: 12015775.

[13] Rodriguez C, Patel AV, Mondul AM, Jacobs EJ, Thun MJ, Calle EE. Diabetes andrisk of prostate cancer in a prospective cohort of US men. Am J Epidemiol2005;161(2):147e52.

[14] Goodman M, Bostick RM, Gross M, Thyagarajan B, Dash C, Flanders WD.Combined measure of pro- and anti-oxidant exposures in relation to prostatecancer and colorectal adenoma risk: an update. Ann Epidemiol2010;20(12):955e7. PubMed PMID: 21074110. Pubmed Central PMCID:3008422.

[15] Agalliu I, Kirsh VA, Kreiger N, Soskolne CL, Rohan TE. Oxidative balance scoreand risk of prostate cancer: results from a case-cohort study. Cancer Epidemiol2011;35(4):353e61. PubMed PMID: 21145797. Epub 2010/12/15. eng.

[16] Rao AV, Rao LG. Carotenoids and human health. Pharmacol Res: Official J ItalPharmacol Soc 2007;55(3):207e16. PubMed PMID: 17349800.

[17] Valko M, Rhodes CJ, Moncol J, Izakovic M, Mazur M. Free radicals, metals andantioxidants in oxidative stress-induced cancer. Chem Biol Interact2006;160(1):1e40. PubMed PMID: 16430879. Epub 2006/01/25. eng.

[18] Na HK, Oliynyk S. Effects of physical activity on cancer prevention. Ann N Y AcadSci 2011;1229:176e83. PubMed PMID: 21793853. Epub 2011/07/29. eng.

[19] Furukawa S, Fujita T, Shimabukuro M, Iwaki M, Yamada Y, Nakajima Y, et al.Increased oxidative stress in obesity and its impact on metabolic syndrome.J Clin Invest 2004;114(12):1752e61. PubMed PMID: 15599400. PubmedCentral PMCID: 535065. Epub 2004/12/16. eng.

[20] Codoner-Franch P, Valls-Belles V, Arilla-Codoner A, Alonso-Iglesias E. Oxidantmechanisms in childhood obesity: the link between inflammation andoxidative stress. Transl Res 2011;158(6):369e84. PubMed PMID: 22061044.Epub 2011/11/09. eng.

[21] Young-McCaughan S. Potential for prostate cancer prevention through physicalactivity. World J Urol; 2011. PubMed PMID: 22198724. Epub 2011/12/27. Eng.

[22] Cao Y, Ma J. Body mass index, prostate cancer-specific mortality, andbiochemical recurrence: a systematic review and meta-analysis. Cancer PrevRes (Phila) 2011;4(4):486e501. PubMed PMID: 21233290. Pubmed CentralPMCID: 3071449. Epub 2011/01/15. eng.

[23] Calder PC, Grimble RF. Polyunsaturated fatty acids, inflammation and immu-nity. Eur J Clin Nutr 2002;56(Suppl 3):S14e9. PubMed PMID: 12142955. Epub2002/07/27. eng.

[24] van Beelen VA, Aarts JM, Reus A, Mooibroek H, Sijtsma L, Bosch D, et al. Dif-ferential induction of electrophile-responsive element-regulated genes by n-3and n-6 polyunsaturated fatty acids. FEBS Lett 2006;580(19):4587e90.PubMed PMID: 16889775. Epub 2006/08/08. eng.

[25] Jones DP. Redefining oxidative stress. Antioxid Redox Signal2006;8(9-10):1865e79. PubMed PMID: 16987039. Epub 2006/09/22. eng.

Supplementary Table 1Oxidative balance score assignment scheme

OBS Components* Pooled RRmeta-analysisy Score assignment

Dietary antioxidantsBeta-carotene 0.9[1e16] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileBeta-cryptoxanthin 0.9[1,5e9,11e13] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileAlpha-carotene 0.9[1,3,5e9,11e13] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileZinc 1.1[14,17e26] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileLutein 0.9[1,3,5e9,11e13,27] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileLycopene 0.9[28] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileVitamin C 0.9[5,8,9,12e15,20,23,29e34] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileVitamin E 1.0[5,9,12,13,15,18,20,31e41] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileU-3 fatty acids 0.9[42] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileFlavonoids 1.2[43e45] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileGlucosinolates 0.7[46] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileSelenium, supplements 1.1[18,33,35,37,47] 0 ¼ 0 mcg/d, 1 ¼ 0e20 mcg/d, 2 ¼ 20.1e50 mcg/d, 3 ¼ 50.1þ mcg/d

Dietary pro-oxidantsU-6 fatty acids 0.9[48] 0 ¼ fourth quartile, 1 ¼ third quartile, 2 ¼ second quartile, 3 ¼ first quartileIron 1.0[20,23,49,50] 0 ¼ fourth quartile, 1 ¼ third quartile, 2 ¼ second quartile, 3 ¼ first quartileSaturated fat 1.0[48] 0 ¼ fourth quartile, 1 ¼ third quartile, 2 ¼ second quartile, 3 ¼ first quartile

Nondietary lifestyle antioxidantsPhysical activity 0.9[51] 0 ¼ first quartile, 1 ¼ second quartile, 2 ¼ third quartile, 3 ¼ fourth quartileNSAID (duration) 0.9[52] 0 ¼ never use, 1 ¼ user in only one survey, 2 ¼ user in 92 and 97 or 97 and 99, 3 ¼ user

in 92, 97, and 99Nondietary lifestyle pro-oxidantsSmoking history 1.1[53] 0 ¼ current smoker, 1 ¼ former smoker (>24 y), 2 ¼ former smoker (�24 y), 3 ¼ nonsmokerBody mass 1.1[54] 0 ¼ high BMI and high waist, 1 ¼ high BMI and med. waist, med. BMI, and high waist; 2 ¼ med BMI

and med waist, med BMI and low waist, low BMI and med waist, 3 ¼ low BMI and low waistAlcohol 1.2[55] 0 ¼ 1 > drink/d; 1 ¼ 1 drink/d; 2 ¼ 1e6 drinks/wk; 3 ¼ 0 drinks/d or <1/wk

BMI ¼ body mass index.* Duration of nonsteroidal anti-inflammatory drugswas categorized based on participant’s responses (user or nonuser) to the 1992, 1997, and 1999 surveys. All other dietary

and nondietary lifestyle OBS components were based on self-report in the 1999 questionnaires.y Weights for pro-oxidants were the pooled adjusted relative risks derived from published studies, whereas weights for the anti-oxidants were calculated using inverse

relative risk estimates. For example, an antioxidant exposure inversely associated with prostate cancer risk, with a pooled relative risk of 0.8 would receive a weight of 1/0.8, or1.25 such that 0, 1, 2, and 3 unweighted points would be converted to 0, 1.25, 2.5, and 3.75 points, respectively. Similarly, for a pro-oxidant with a pooled relative risk of 1.1 theunweighted points 0e3 after weighting would assume a range of 0e3.3.

S. Lakkur et al. / Annals of Epidemiology 24 (2014) 475e478 478.e1

Supplementary Table 2OBS components of Cancer Prevention Study II men by quartile

OBS components OBS quartile 1 OBS quartile 2 OBS quartile 3 OBS quartile 4

Total beta-carotene (mcg/d), mean (SD)* 2028.02 (1126.23) 3291.88 (2132.19) 4507.86 (2754.17) 6851.48 (4346.69)Total beta-cryptoxanthin (mcg/d), mean (SD) 108.39 (93.27) 156.24 (114.10) 184.26 (122.72) 227.51 (134.11)Total alpha-carotene (mcg/d), mean (SD) 425.97 (319.93) 673.06 (488.30) 898.87 (646.76) 1256.75 (819.02)Total lutein intake (mg/d), mean (SD) 1144.23 (650.53) 1718.96 (930.01) 2242.43 (1198.91) 3136.25 (1755.94)Total lycopene intake (mg/d), mean (SD) 3959.50 (2945.95) 4923.60 (3426.67) 5544.10 (3547.58) 6669.42 (4133.75)Total zinc intake (mg/d), mean (SD)* 12.84 (11.15) 17.15 (16.21) 21.91 (21.71)Total vitamin C intake (mg/d), mean (SD)* 193.86 (223.39) 340.59 (320.35) 477.10 (396.48) 705.08 (504.14)Total vitamin E intake (mg �TE/d), mean (SD)* 41.30 (67.14) 75.49 (86.03) 105.02 (93.85) 148.54 (99.25)Total U-3 fatty acid intake (gm/d), mean (SD)* 1.21 (0.41) 1.26 (0.41) 1.29 (0.42) 1.36 (0.45)Dietary flavonoid intake (mg/d), mean (SD)y 191.95 (186.03) 252.37 (213.07) 296.63 (236.97) 366.82 (263.05)Dietary glucosinolate intake (mg/d), mean (SD)z 6.59 (5.47) 10.16 (7.90) 13.74 (10.34) 20.32 (14.67)Selenium supplements (mg/d), mean (SD) 3.57 (13.64) 9.24 (25.36) 16.49 (37.34) 38.70 (62.34)Total U-6 fatty acid intake (gm/d), mean (SD)* 12.65 (3.38) 12.19 (3.27) 11.82 (3.22) 11.15 (3.16)Saturated fat intake (gm/d), mean (SD) 22.74 (5.18) 20.53 (4.94) 18.90 (4.69) 16.41 (4.58)Dietary iron intake (mg/d), mean (SD)* 15.64 (7.58) 18.73 (9.45) 20.71 (10.38) 22.78 (11.10)Physical activity (met-h/wk) 12.83 (13.69) 16.78 (15.12) 19.37 (16.25) 23.77 (18.61)BMI (kg/m2), mean (SD) 27.36 (3.88) 26.71 (3.63) 26.22 (3.41) 25.45 (3.15)Waist circumference (inches), mean (SD) 39.65 (4.16) 38.89 (3.95) 38.31 (3.79) 37.33 (3.52)Current smoker, (%) 8.36 3.25 1.87 0.70Former smoker >24 y (%) 31.65 25.56 21.27 15.31Former smoker <24 y (%) 40.03 43.17 43.94 43.441e6 alcoholic drinks/wk, (%) 25.20 27.86 29.23 31.607 alcoholic drinks/wk (%) 13.66 16.12 17.51 17.46>7 alcoholic drinks/wk, (%) 20.88 15.50 12.84 8.90NSAID user in 92, 97, 99 (%) 35.76 45.74 50.66 58.28NSAID user in 92 and 97 or 97 and 99 (%) 15.75 17.12 16.90 15.59NSAID user in only one survey (%) 24.98 20.61 18.95 15.36

Note: all nutrients adjusted for total energy intake.BMI ¼ body mass index; MET ¼ metabolic equivalents; NSAID ¼ nonsteroidal anti-inflammatory drugs.

* Total intake ¼ diet plus supplements. The use of vitamin supplements was assessed by questions on frequency of multivitamin use and preferred brand as well as in-dividual supplements and dosages. Nutrient intakes from supplements were estimated by using the Diet Analysis System version 3.8a [56].

y Flavonoid intake was determined using nutrient databases published by the U.S. Department of Agriculture.z Glucosinolate intake was derived from estimates of glucosinolate content in common food sources of the compound [57].

S. Lakkur et al. / Annals of Epidemiology 24 (2014) 475e478478.e2

References

[1] Peters U, Leitzmann MF, Chatterjee N, Wang Y, Albanes D, Gelmann EP, et al.Serum lycopene, other carotenoids, and prostate cancer risk: a nested case-control study in the prostate, lung, colorectal, and ovarian cancer screeningtrial. Cancer Epidemiol Biomarkers Prev 2007;16(5):962e8. PubMed PMID:17507623. Epub 2007/05/18. eng.

[2] Beilby J, Ambrosini GL, Rossi E, de Klerk NH, Musk AW. Serum levels of folate,lycopene, beta-carotene, retinol andvitamin E andprostate cancer risk. Eur J ClinNutr 2010;64(10):1235e8. PubMed PMID: 20683458. Epub 2010/08/05. eng.

[3] McCann SE, Ambrosone CB, Moysich KB, Brasure J, Marshall JR,Freudenheim JL, et al. Intakes of selected nutrients, foods, and phytochemicalsand prostate cancer risk in western New York. Nutr Cancer 2005;53(1):33e41.PubMed PMID: 16351504. Epub 2005/12/15. eng.

[4] Ambrosini GL, de Klerk NH, Fritschi L, Mackerras D, Musk B. Fruit, vegetable,vitamin A intakes, and prostate cancer risk. Prostate Cancer Prostatic Dis2008;11(1):61e6. PubMed PMID: 17519926. Epub 2007/05/24. eng.

[5] Schuurman AG, Goldbohm RA, Brants HA, van den Brandt PA. A prospectivecohort study on intake of retinol, vitamins C and E, and carotenoids andprostate cancer risk (Netherlands). Cancer Causes Control 2002;13(6):573e82.PubMed PMID: 12195647. Epub 2002/08/28. eng.

[6] Jian L, Du CJ, Lee AH, Binns CW. Do dietary lycopene and other carotenoidsprotect against prostate cancer? Int J Cancer 2005;113(6):1010e4. PubMedPMID: 15514967. Epub 2004/10/30. eng.

[7] Giovannucci E, Ascherio A, Rimm EB, Stampfer MJ, Colditz GA, Willett WC.Intake of carotenoids and retinol in relation to risk of prostate cancer.J Natl Cancer Inst 1995;87(23):1767e76. PubMed PMID: 7473833. Epub1995/12/06. eng.

[8] Cohen JH, Kristal AR, Stanford JL. Fruit and vegetable intakes and prostatecancer risk. J Natl Cancer Inst 2000;92(1):61e8. PubMed PMID: 10620635.Epub 2000/01/06. eng.

[9] Stram DO, Hankin JH, Wilkens LR, Park S, Henderson BE, Nomura AM, et al.Prostate cancer incidence and intake of fruits, vegetables and related micro-nutrients: the multiethnic cohort study* (United States). Cancer Causes Con-trol 2006;17(9):1193e207. PubMed PMID: 17006725. Epub 2006/09/29. eng.

[10] Mettlin C, Selenskas S, Natarajan N, Huben R. Beta-carotene and animal fatsand their relationship to prostate cancer risk. A case-control study. Cancer1989;64(3):605e12. PubMed PMID: 2743255. Epub 1989/08/01. eng.

[11] Bosetti C, Talamini R, Montella M, Negri E, Conti E, Franceschi S, et al.Retinol, carotenoids and the risk of prostate cancer: a case-control studyfrom Italy. Int J Cancer 2004;112(4):689e92. PubMed PMID: 15382052. Epub2004/09/24. eng.

[12] Kirsh VA, Hayes RB, Mayne ST, Chatterjee N, Subar AF, Dixon LB, et al. Sup-plemental and dietary vitamin E, beta-carotene, and vitamin C intakes andprostate cancer risk. J Natl Cancer Inst 2006;98(4):245e54. PubMed PMID:16478743. Epub 2006/02/16. eng.

[13] Hodge AM, English DR, McCredie MR, Severi G, Boyle P, Hopper JL, et al. Foods,nutrients and prostate cancer. Cancer Causes Control 2004;15(1):11e20.PubMed PMID: 14970730. Epub 2004/02/19. eng.

[14] Andersson SO, Wolk A, Bergstrom R, Giovannucci E, Lindgren C, Baron J, et al.Energy, nutrient intake and prostate cancer risk: a population-based case-control study in Sweden. Int J Cancer 1996;68(6):716e22. PubMed PMID:8980172. Epub 1996/12/11. eng.

[15] Shibata A, Paganini-Hill A, Ross RK, Henderson BE. Intake of vegetables,fruits, beta-carotene, vitamin C and vitamin supplements and cancer inci-dence among the elderly: a prospective study. Br J Cancer1992;66(4):673e9. PubMed PMID: 1419605. Pubmed Central PMCID:1977409. Epub 1992/10/01. eng.

[16] Virtamo J, Pietinen P, Huttunen JK, Korhonen P, Malila N, Virtanen MJ, et al.Incidence of cancer and mortality following alpha-tocopherol and beta-carotene supplementation: a postintervention follow-up. JAMA2003;290(4):476e85. PubMed PMID: 12876090. Epub 2003/07/24. eng.

[17] Gonzalez A, Peters U, Lampe JW, White E. Zinc intake from supplements anddiet and prostate cancer. Nutr Cancer 2009;61(2):206e15. PubMed PMID:19235036. Pubmed Central PMCID: 2664741. Epub 2009/02/24. eng.

[18] Zhang Y, Coogan P, Palmer JR, Strom BL, Rosenberg L. Vitamin and mineral useand risk of prostate cancer: the case-control surveillance study. Cancer CausesControl 2009;20(5):691e8. PubMed PMID: 19093214. Pubmed CentralPMCID: 2755205. Epub 2008/12/19. eng.

[19] Leitzmann MF, Stampfer MJ, Wu K, Colditz GA, Willett WC, Giovannucci EL.Zinc supplement use and risk of prostate cancer. J Natl Cancer Inst2003;95(13):1004e7. PubMed PMID: 12837837. Epub 2003/07/03. eng.

[20] Kristal AR, Stanford JL, Cohen JH, Wicklund K, Patterson RE. Vitamin andmineral supplement use is associated with reduced risk of prostate cancer.Cancer Epidemiol Biomarkers Prev 1999;8(10):887e92. PubMed PMID:10548317. Epub 1999/11/05. eng.

[21] Key TJ, Silcocks PB, Davey GK, Appleby PN, Bishop DT. A case-control study ofdiet and prostate cancer. Br J Cancer 1997;76(5):678e87. PubMed PMID:9303371. Pubmed Central PMCID: 2228001. Epub 1997/01/01. eng.

[22] Kolonel LN, Yoshizawa CN, Hankin JH. Diet and prostatic cancer: a case-control study in Hawaii. Am J Epidemiol 1988;127(5):999e1012. PubMedPMID: 3358418. Epub 1988/05/01. eng.

[23] Vlajinac HD, Marinkovic JM, Ilic MD, Kocev NI. Diet and prostate cancer: acase-control study. Eur J Cancer 1997;33(1):101e7. PubMed PMID: 9071908.Epub 1997/01/01. eng.

[24] Gallus S, Foschi R, Negri E, Talamini R, Franceschi S, Montella M, et al. Dietaryzinc and prostate cancer risk: a case-control study from Italy. Eur Urol2007;52(4):1052e6. PubMed PMID: 17292532. Epub 2007/02/13. eng.

[25] Lee MM, Wang RT, Hsing AW, Gu FL, Wang T, Spitz M. Case-control study ofdiet and prostate cancer in China. Cancer Causes Control 1998;9(6):545e52.PubMed PMID: 10189039. Epub 1999/04/03. eng.

[26] West DW, Slattery ML, Robison LM, French TK, Mahoney AW. Adult dietaryintake and prostate cancer risk in Utah: a case-control study with specialemphasis on aggressive tumors. Cancer Causes Control 1991;2(2):85e94.PubMed PMID: 1873441. Epub 1991/03/01. eng.

[27] Key TJ, Appleby PN, Allen NE, Travis RC, Roddam AW, Jenab M, et al. Plasmacarotenoids, retinol, and tocopherols and the risk of prostate cancer in theEuropean Prospective Investigation into Cancer and Nutrition study. Am J ClinNutr 2007;86(3):672e81. PubMed PMID: 17823432. Epub 2007/09/08. eng.

[28] Etminan M, Takkouche B, Caamano-Isorna F. The role of tomato products andlycopene in the prevention of prostate cancer: a meta-analysis of observa-tional studies. Cancer Epidemiol Biomarkers Prev 2004;13(3):340e5. PubMedPMID: 15006906. Epub 2004/03/10. eng.

[29] Parker AS, Cerhan JR, Putnam SD, Cantor KP, Lynch CF. A cohort study offarming and risk of prostate cancer in Iowa. Epidemiology 1999;10(4):452e5.PubMed PMID: 10401883. Epub 1999/07/13. eng.

[30] Raimondi S, Mabrouk JB, Shatenstein B, Maisonneuve P, Ghadirian P. Diet andprostate cancer risk with specific focus on dairy products and dietary calcium:a case-control study. Prostate 2010;70(10):1054e65. PubMed PMID:20232354. Epub 2010/03/17. eng.

[31] Ramon JM, Bou R, Romea S, Alkiza ME, Jacas M, Ribes J, et al. Dietary fat intakeand prostate cancer risk: a case-control study in Spain. Cancer Causes Control2000;11(8):679e85. PubMed PMID: 11065004. Epub 2000/11/07. eng.

[32] Bidoli E, Talamini R, Zucchetto A, Bosetti C, Negri E, Lenardon O, et al. Dietaryvitamins E and C and prostate cancer risk. Acta Oncol 2009;48(6):890e4.PubMed PMID: 19452333. Epub 2009/05/20. eng.

[33] Kristal AR, Arnold KB, Neuhouser ML, Goodman P, Platz EA, Albanes D, et al.Diet, supplement use, and prostate cancer risk: results from theprostate cancer prevention trial. Am J Epidemiol 2010;172(5):566e77.PubMed PMID: 20693267. Pubmed Central PMCID: 2950820. Epub 2010/08/10. eng.

[34] Gaziano JM, Glynn RJ, Christen WG, Kurth T, Belanger C, MacFadyen J, et al.Vitamins E and C in the prevention of prostate and total cancer in men: thePhysicians’ Health Study II randomized controlled trial. JAMA2009;301(1):52e62. PubMed PMID: 19066368. Pubmed Central PMCID:2774210. Epub 2008/12/11. eng.

[35] Peters U, Littman AJ, Kristal AR, Patterson RE, Potter JD, White E. Vitamin E andselenium supplementation and risk of prostate cancer in the Vitamins andlifestyle (VITAL) study cohort. Cancer Causes Control 2008;19(1):75e87.PubMed PMID: 17943452. Epub 2007/10/19. eng.

[36] Tzonou A, Signorello LB, Lagiou P, Wuu J, Trichopoulos D, Trichopoulou A. Dietand cancer of the prostate: a case-control study in Greece. Int J Cancer1999;80(5):704e8. PubMed PMID: 10048971. Epub 1999/02/27. eng.

[37] Hartman TJ, Albanes D, Pietinen P, Hartman AM, Rautalahti M, Tangrea JA,et al. The association between baseline vitamin E, selenium, and prostatecancer in the alpha-tocopherol, beta-carotene cancer prevention study. CancerEpidemiol Biomarkers Prev 1998;7(4):335e40. PubMed PMID: 9568790. Epub1998/05/06. eng.

[38] Lawson KA, Wright ME, Subar A, Mouw T, Hollenbeck A, Schatzkin A, et al.Multivitamin use and risk of prostate cancer in the National Institutes ofHealth-AARP Diet and Health Study. J Natl Cancer Inst 2007;99(10):754e64.PubMed PMID: 17505071. Epub 2007/05/17. eng.

[39] Wright ME, Weinstein SJ, Lawson KA, Albanes D, Subar AF, Dixon LB, et al.Supplemental and dietary vitamin E intakes and risk of prostate cancer in alarge prospective study. Cancer Epidemiol Biomarkers Prev2007;16(6):1128e35. PubMed PMID: 17548674. Epub 2007/06/06. eng.

[40] Rodriguez C, Jacobs EJ, Mondul AM, Calle EE, McCullough ML, Thun MJ.Vitamin E supplements and risk of prostate cancer in U.S. men. Cancer Epi-demiol Biomarkers Prev 2004;13(3):378e82. PubMed PMID: 15006912. Epub2004/03/10. eng.

[41] Chan JM, Stampfer MJ, Ma J, Rimm EB, Willett WC, Giovannucci EL. Supple-mental vitamin E intake and prostate cancer risk in a large cohort of men inthe United States. Cancer Epidemiol Biomarkers Prev 1999;8(10):893e9.PubMed PMID: 10548318. Epub 1999/11/05. eng.

[42] Carayol M, Grosclaude P, Delpierre C. Prospective studies of dietary alpha-linolenic acid intake and prostate cancer risk: a meta-analysis. Cancer Cau-ses Control 2010;21(3):347e55. PubMed PMID: 19921446. Epub 2009/11/19.eng.

[43] Mursu J, Nurmi T, Tuomainen TP, Salonen JT, Pukkala E, Voutilainen S. Intakeof flavonoids and risk of cancer in Finnish men: The Kuopio Ischaemic HeartDisease Risk Factor Study. Int J Cancer 2008;123(3):660e3. PubMed PMID:18338754. Epub 2008/03/15. eng.

[44] Bosetti C, Bravi F, Talamini R, Parpinel M, Gnagnarella P, Negri E, et al. Fla-vonoids and prostate cancer risk: a study in Italy. Nutr Cancer2006;56(2):123e7. PubMed PMID: 17474856. Epub 2007/05/04. eng.

S. Lakkur et al. / Annals of Epidemiology 24 (2014) 475e478 478.e3

[45] Knekt P, Jarvinen R, Seppanen R, Hellovaara M, Teppo L, Pukkala E, et al.Dietary flavonoids and the risk of lung cancer and other malignant neo-plasms. Am J Epidemiol 1997;146(3):223e30. PubMed PMID: 9247006. Epub1997/08/01. eng.

[46] Steinbrecher A, Nimptsch K, Husing A, Rohrmann S, Linseisen J. Dietary glu-cosinolate intake and risk of prostate cancer in the EPIC-Heidelberg cohortstudy. Int J Cancer 2009;125(9):2179e86. PubMed PMID: 19585501. Epub2009/07/09. eng.

[47] Etminan M, FitzGerald JM, Gleave M, Chambers K. Intake of selenium in theprevention of prostate cancer: a systematic review and meta-analysis. Can-cer Causes Control 2005;16(9):1125e31. PubMed PMID: 16184479. Epub2005/09/27. eng.

[48] Dennis LK, Snetselaar LG, Smith BJ, Stewart RE, Robbins ME. Problems with theassessment of dietary fat in prostate cancer studies. Am J Epidemiol2004;160(5):436e44. PubMed PMID: 15321840. Epub 2004/08/24. eng.

[49] Choi JY, Neuhouser ML, Barnett MJ, Hong CC, Kristal AR, Thornquist MD,et al. Iron intake, oxidative stress-related genes (MnSOD and MPO) andprostate cancer risk in CARET cohort. Carcinogenesis 2008;29(5):964e70.PubMed PMID: 18296681. Pubmed Central PMCID: 2902382. Epub 2008/02/26. eng.

[50] Sinha R, Park Y, Graubard BI, Leitzmann MF, Hollenbeck A, Schatzkin A, et al.Meat and meat-related compounds and risk of prostate cancer in a largeprospective cohort study in the United States. Am J Epidemiol2009;170(9):1165e77. PubMed PMID: 19808637. Pubmed Central PMCID:2781742. Epub 2009/10/08. eng.

[51] Liu Y, Hu F, Li D, Wang F, Zhu L, Chen W, et al. Does physical activity reduce therisk of prostate cancer? A systematic review and meta-analysis. Eur Urol2011;60(5):1029e44. PubMed PMID: 21802197. Epub 2011/08/02. eng.

[52] Mahmud SM, Franco EL, Aprikian AG. Use of nonsteroidal anti-inflammatorydrugs and prostate cancer risk: a meta-analysis. Int J Cancer2010;127(7):1680e91. PubMed PMID: 20091856. Epub 2010/01/22. eng.

[53] Huncharek M, Haddock KS, Reid R, Kupelnick B. Smoking as a risk factor forprostate cancer: a meta-analysis of 24 prospective cohort studies. Am J PublicHealth 2010;100(4):693e701. PubMed PMID: 19608952. Pubmed CentralPMCID: 2836346. Epub 2009/07/18. eng.

[54] Guh DP, Zhang W, Bansback N, Amarsi Z, Birmingham CL, Anis AH. The inci-dence of co-morbidities related to obesity and overweight: a systematic reviewand meta-analysis. BMC Public Health 2009;9:88. PubMed PMID: 19320986.Pubmed Central PMCID: 2667420. Epub 2009/03/27. eng.

[55] Middleton Fillmore K, Chikritzhs T, Stockwell T, Bostrom A, Pascal R. Alcohol useand prostate cancer: a meta-analysis. Mol Nutr Food Res 2009;53(2):240e55.PubMed PMID: 19156715. Epub 2009/01/22. eng.

[56] Block GCL, Smucker R, Harlan L. Health habits and history questionnaire: diethistory and other risk factors: personal computer system documentation.Bethesda, MD: National Cancer Institute Division of Cancer Prevention andControl, National Institutes of Health; 1989.

[57] McNaughton SA, Marks GC. Development of a food composition database for theestimation of dietary intakes of glucosinolates, the biologically active constit-uents of cruciferous vegetables. Br J Nutr 2003;90(3):687e97. PubMed PMID:13129476. Epub 2003/09/18. eng.

S. Lakkur et al. / Annals of Epidemiology 24 (2014) 475e478478.e4