A Case-Control Study of Risk Factors for Fibrocystic Breast Conditions

of 16

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
16 pages
0 downs
A Case-Control Study of Risk Factors for Fibrocystic Breast Conditions
  945   Am J Epidemiol   2004;160:945–960 American Journal of EpidemiologyCopyright © 2004 by the Johns Hopkins Bloomberg School of Public HealthAll rights reservedVol. 160, No. 10 Printed in U.S.A. DOI: 10.1093/aje/kwh318 A Case-Control Study of Risk Factors for Fibrocystic Breast Conditions Shanghai Nutrition and Breast Disease Study, China, 1995–2000 Chunyuan Wu 1 , Roberta M. Ray 1 , Ming Gang Lin 1 , Dao Li Gao 2 , Neilann K. Horner 1 , Zakia C. Nelson 1 , Johanna W. Lampe 1,3 , Yong Wei Hu 4 , Jackilen Shannon 1,5 , Helge Stalsberg 6 , Wenjin Li 1 , Dawn Fitzgibbons 1 , Peggy Porter 1 , Ruth E. Patterson 1,3 , Jessie A. Satia 7 , and David B. Thomas 1,3 1  Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA. 2  Department of Epidemiology, Zhong Shan Hospital Cancer Center, Shanghai, China. 3  Department of Epidemiology, University of Washington, Seattle, WA. 4  Shi Dong Hospital, Shanghai, China. 5  Veterans Administration Medical Center, Portland, OR. 6  Institute of Medical Biology, University of Tromsø, Tromsø, Norway. 7  Department of Nutrition, School of Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC. Received for publication December 16, 2003; accepted for publication June 10, 2004. This study was conducted to identify reproductive and dietary factors associated with benign proliferativemammary epithelial cell changes. Subjects were women enrolled in a randomized trial of breast self-examinationin Shanghai, China. Women who developed fibrocystic breast conditions classified as nonproliferative (175women), proliferative (181 women), or proliferative with atypia (33 women) between 1995 and 2000 and 1,070unaffected trial participants were administered general risk factor and food frequency questionnaires. Conditionallogistic regression was used to estimate adjusted odds ratios and 95% confidence intervals. High parity andconsumption of fresh fruits and vegetables were more strongly associated with a reduced risk of proliferative andatypical lesions than with nonproliferative conditions. For the fourth quartile of consumption versus the first, oddsratios for lesions diagnosed as nonproliferative, proliferative, and proliferative with atypia were 0.4 (95%confidence interval (CI): 0.2, 0.7), 0.2 (95% CI: 0.1, 0.4), and 0.1 (95% CI: 0.03, 0.5), respectively, for fruit intakeand 0.6 (95% CI: 0.3, 1.1), 0.4 (95% CI: 0.2, 0.7), and 0.1 (95% CI: 0.1, 0.9), respectively, for vegetable intake.Reduced but nonsignificant risks in relation to soy products were observed for proliferative and atypical lesions.No single nutrient or botanical family was appreciably more strongly associated with proliferative conditions thanwith nonproliferative conditions, after results were controlled for total fruit and vegetable consumption. A diet richin fruits and vegetables may reduce cellular proliferation in the mammary epithelium; this is one mechanism bywhich such a diet could reduce risk of breast cancer.breast diseases; diet; estrogens; fibrocystic disease of breast; fruit; risk factors; soy foods; vegetables Fibrocystic breast conditions, formerly referred to as“fibrocystic breast disease” (1, 2), with proliferative epithe-lial cell elements have been associated with increased risk of subsequent breast cancer, especially if accompanied by atyp-ical cellular changes (3–6). Since factors that stimulate cellproliferation may enhance the likelihood of malignantchange, the identification of risk factors for proliferativemammary epithelial cell changes could provide insights intothe early stages of mammary carcinogenesis.Previous studies have shown increased risk of fibrocysticbreast conditions in the aggregate to be associated with highsocial class (3, 4, 7, 8), late age at menopause (3, 4, 6, 8, 9),estrogen replacement therapy (3), nulliparity (3, 4, 6, 8–10),low body mass index (3, 4, 7, 8, 11), and family history of breast cancer (4, 6–9, 12), while high parity (3, 7–9), oralcontraceptive use (3, 4, 6, 7, 10, 13), and physical activity(14, 15) have been related to decreased risk. Most reportshave shown no substantial effect of ever, former, or current Correspondence to Dr. David B. Thomas, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, MP-474, Seattle, WA 98109 (e-mail: dbthomas@fhcrc.org).   b  y  g u e s  t   on J  ul   y 2 2  ,2  0 1 1  a j   e. ox f   or  d  j   o ur n al   s . or  gD  ownl   o a d  e d f  r  om   946   Wu et al.  Am J Epidemiol   2004;160:945–960 smoking on development of fibrocystic conditions, evenacross different grades of atypia (3, 6, 8). Results have beeninconsistent for young age at first full-term pregnancy (3, 7,9, 16) and lactation (7). Age at menarche has not been asso-ciated with risk (3, 4, 10).Among the dietary factors studied, the most consistentfinding is an apparently protective effect of fruits and vege-tables (17–20). Diets high in fiber (21–23) and soy products(21) have also been associated with reduced risk, but arandomized trial showed increased cell division in lobularmammary epithelium after soy protein supplementation(24). Consumption of green vegetables has been inverselyassociated specifically with risk of proliferative benignbreast disease (20). Results from studies of total fat areinconsistent (3, 19, 25, 26). In one study, high daily intake of  energy was reported to be positively associated with benignepithelial cell proliferative disease (22).We conducted this study to evaluate the role that reproduc-tive and dietary factors, particularly antioxidants and certainphytoestrogens, may play in the development of mammaryepithelial cell proliferation in Chinese women with fibro-cystic breast conditions. MATERIALS AND METHODSStudy setting and subjects Subjects for this case-control study were selected fromwomen who were enrolled between 1989 and 1991 in arandomized trial of breast self-examination among textileworkers in Shanghai, China (27, 28). When a woman in thetrial detected a breast lump, she was evaluated by a medicalworker in her factory who, if indicated, referred her to one of three hospitals operated by the Shanghai Textile IndustryBureau or to other hospitals that had contractual agreementswith individual factories. Study personnel reviewed thepathology reports and other medical records of all womenfound to have a histologically confirmed benign breastlesion, recorded tumor size and histologic classification on astandardized form, and obtained slides of all tumors for ship-ment to Seattle, Washington, for review. Women in the trialwho received a breast biopsy in any of the three ShanghaiTextile Industry Bureau hospitals and were subsequentlydiagnosed with a fibrocystic breast condition betweenSeptember 1995 and July 2000 were eligible for the presentstudy.A total of 622 women with fibrocystic breast conditionswere identified for this study, and in-person interviews werecompleted for 551 (89 percent) of them. Of these women,389 (71 percent) had satisfactory slides for review (at leastfive scanning power fields) and were included in the anal-yses for this report.Women whose breast conditions were diagnosed betweenSeptember 1995 and August 1997 were simultaneouslyenrolled in an ongoing immunocytochemical study of mammary cell proliferation (unpublished data). Controls forthe present study were selected from unaffected women inthe breast self-examination trial cohort. For each benign andmalignant case also enrolled in the previous study of cellproliferation, 20 potential controls of the same age wererandomly selected. Women were contacted, starting with thefirst two names on the list, until two women of the same ageand menstrual status as their matched case were recruited.All 367 controls recruited in this manner (64 percent of theeligible women contacted) were included in the analyses forthis report. Controls for the cases who were not enrolled inthe study of cell proliferation were frequency matched tocases eligible for this study, as well as to cases also eligiblefor two concurrent studies of breast cancer and fibro-adenoma, by 5-year age group and hospital affiliation of their factories at baseline. In-person interviews werecompleted for 704 (82 percent) of 862 controls selected inthis manner. Data collection A baseline questionnaire was administered orally to allwomen in the breast self-examination trial by factorymedical clinic workers between October 1989 and October1991. The women who participated in the present study werereinterviewed from 1995 through 2001 by members of ateam of specially trained former medical clinic workers,utilizing a detailed general risk factor questionnaire and afood frequency questionnaire. Cases were interviewed inhospitals or clinics, usually at the time of their initial visit fortheir breast problem, before the histologic diagnosis wasmade. Controls were interviewed in their homes or factories.Information was elicited on demographic characteristics,reproductive and gynecologic history, smoking and alcoholhabits, medical history, family history of breast cancer, andoccupational and recreational physical activities. Becauseprior breast surgery reported by many women erroneouslyincluded the surgery conducted for the present disease, infor-mation from the baseline questionnaire on prior breastsurgery was utilized. The food frequency questionnaire wasbased on a previously validated instrument (29, 30) andascertained data on the frequency of intake of 99 food itemsduring adult life. Respondents provided information on theirusual frequency of intake (per day, week, month, or year) of each item. The seasonality of fruit and vegetable consump-tion was accounted for by asking subjects to report thenumber of months of the year in which they consumed eachitem.Informed consent was obtained from each woman prior tointerview. The study was approved by the institutionalreview boards of the Fred Hutchinson Cancer ResearchCenter and the Station for Prevention and Treatment of Cancer of the Shanghai Textile Industry Bureau, in accor-dance with assurance filed with the Office for HumanResearch Protections of the US Department of Health andHuman Services. Validation of diagnosis and histologic classification A single study pathologist (M. L.) read slides from allcases without knowledge of their srcinal diagnosis. Herecorded the number of scanning power fields examined andclassified the women’s conditions into three different typesof fibrocystic breast conditions according to the schemedeveloped by Stalsberg (31). Cases with atypia were those   b  y  g u e s  t   on J  ul   y 2 2  ,2  0 1 1  a j   e. ox f   or  d  j   o ur n al   s . or  gD  ownl   o a d  e d f  r  om   Risk Factors for Fibrocystic Breast Conditions 947   Am J Epidemiol   2004;160:945–960 with atypical ductal hyperplasia, atypical lobular hyper-plasia, or moderate apocrine atypia (33 cases). Cases withproliferative changes were those with moderate or floridductal hyperplasia or moderate or predominant sclerosingadenosis and no atypia (181 cases). Subjects with mild or noductal hyperplasia and subjects with mild or no sclerosingadenosis were classified as having nonproliferative fibro-cystic breast conditions (175 cases).Randomly selected slides from the three major groups of fibrocystic breast conditions were read by a reference pathol-ogist (H. S.). There was a satisfactory level of agreementbetween the readings made by the study pathologist and thereference pathologist on the presence or absence of prolifera-tion and atypia (weighted κ   = 0.4), and the benign conditionswere categorized into 1) nonproliferative, 2) proliferativewithout atypia, and 3) proliferative with atypia for analysis. Inall instances, the diagnoses made by the study pathologistwere used. Data analysis All individual food consumption frequencies wereconverted to annual frequencies. Intake of fruits and vegeta-bles was computed by weighting the frequency of intake bythe number of months per year during which the food itemwas eaten. The 99 foods were grouped into 19 mutuallyexclusive food groups (Appendix 1). The individual annualfrequencies of intake for the foods in each group weresummed to create values for consumption of the food group.Amounts of sesame and soybean oils consumed were esti-mated by dividing the amounts used by the woman’s familyper day by the number of family members. Nine fried fooditems were combined into a separate high-fat group. Fruitsand vegetables were further classified into mutually exclu-sive botanical families (Appendix 2). Total daily energyintake was calculated from all macronutrients, oils, andalcohol using 1991 Chinese food composition tables (32, 33)and estimated average serving sizes from the 1992 ChinaHealth and Nutrition Survey (34). Total daily dietary intakesof vitamin E (mg of α -tocopherol equivalents), carotenoids( µ g), vitamin C (mg), and crude fiber (g) were also calcu-lated from the same Chinese food composition tables (32,33). One woman with a nonproliferative fibrocystic breastcondition and one control were excluded because their calcu-lated daily energy intakes were considered implausible atmore than 4,000 kcal.To assess potential trends in risk with level of consump-tion, we divided food groups, botanical families, and nutri-ents into quartiles according to the distribution of consumption among the controls. We created fewer catego-ries for food groups with too little variation in intake tocreate meaningful quartiles. Because controls were inter-viewed after the corresponding cases, they tended to havebeen interviewed at a later date. Therefore, we stratifiedstudy subjects by year of interview (1995–1996, 1997,1998–1999, 2000–2001) in all analyses. We computed oddsratios (as estimates of relative risk) and associated 95 percentconfidence intervals using conditional logistic regression(35). For small sample sizes (<5), we used exact logisticregression to calculate odds ratios and confidence intervals.All odds ratios were adjusted for age using 5-year age cate-gories. We assessed residual confounding by age bycomparing selected results in these analyses with results of analyses in which age was modeled as a continuous variable;no residual confounding by age was detected. The oddsratios for food groups were further adjusted for total energyintake (36). Odds ratios for botanical groups and specificmicronutrients were adjusted for age and total fruit and vege-table intake. We performed tests for trend by entering thecategorical variables in regression models as ordinal vari-ables. We evaluated potential confounding by adding vari-ables independently to the appropriate model. Variableswere considered as confounders if they changed the esti-mated β  coefficient of the main independent variable by 10percent or more. RESULTS As expected, because the controls for this study were alsoselected for studies of breast cancer, they tended to be olderthan the cases without atypia in this study (table 1). Womenwith atypia were older than women with nonproliferativeand proliferative conditions. Almost 80 percent of thewomen who developed fibrocystic breast conditions wereyounger than age 50 years at diagnosis.Although results were based on small numbers of post-menopausal women and the findings were not statisticallysignificant, the risk of all three types of conditions was lowerfor women who had gone through natural menopause than TABLE 1. Age distribution of cases with fibrocystic breast conditions and controls, Shanghai Nutrition and Breast Disease Study, 1995–2000 Age group (years)Nonproliferative casesProliferative casesAtypia casesControlsNo.%No.%No.%No.% 35–39 2514.3168.826.1131.240–447140.68446.41339.447043.945–494626.35128.2618.221920.550–591910.8105.526.112411.4 ≥ 60 148.02011.11030.324423.0Total 175100.0181100.033100.01,070100.0   b  y  g u e s  t   on J  ul   y 2 2  ,2  0 1 1  a j   e. ox f   or  d  j   o ur n al   s . or  gD  ownl   o a d  e d f  r  om   948   Wu et al.  Am J Epidemiol   2004;160:945–960 for women of the same age who had not (table 2). Thisfinding was not observed for artificial menopause. Risks of all three conditions declined with increasing number of live-births, although only strongly and significantly so for prolif-erative lesions with and without atypia. Risk of proliferativeconditions declined with age at menarche. Because only TABLE 2. Risk of fibrocystic breast conditions in relation to general risk factors, stratified by year of interview, Shanghai Nutrition and Breast Disease Study, 1995–2000 * ExposureNo. of controlsNonproliferative casesProliferative casesAtypia casesNo. of casesOR†,‡95% CI†No. of casesOR‡95% CINo. of casesOR‡95% CI Total1,07017518133MenopausalNo6871451.0§1421.0§221.0§Yes383300.60.3, 1.4390.90.4, 2.1110.40.05, 3.0Type of menopauseNatural331230.40.2, 1.1320.90.4, 2.1100.40.1, 2.7Surgical5050.80.2, 2.671.00.3,, 11.5No. of livebirths¶,#17131361.0§1461.0§211.0§2124140.40.2, 0.9110.20.1, 0.750.30.1, 1.4 ≥ 3188160.60.2, 2.1140.20.1, 0.860.10.02, 0.6 p   trend ** of lactation (months)#,††0180311.0§361.0§71.0§ ≤ 6209451.30.7, 2.5541.80.9, 3.750.70.2, 2.87–12361590.80.4, 1.5500.70.4, 1.4110.70.2, 2.213–24115160.80.3, 2.7131.50.4, 5.340.70.1, 4.4 ≥ 25152110.60.2, 2.3131.70.4, 7.651.0§0.1, 11.0 p   trend‡‡ (years) at first livebirth#,†† ≤ 24273291.0§251.0§121.0§25–296001121.00.5, 1.91131.10.5, 2.2160.30.1, 1.0 ≥ 30152250.80.3, 1.8330.90.4,, 1.2 p   trend ** 0.510.780.09Age (years) at menarche ≤ 13171331.0§291.0§61.0§14204370.90.5, 1.8500.90.5, 1.870.80.2, 2.715215381.10.6, 2.1360.70.3, 1.350.60.2, 2.216221340.60.3, 1.2290.50.2,, 1.5 ≥ 17257330.90.4, 1.7370.60.3, 1.2100.80.2, 2.5 p   trend ** 0.360.050.53Duration (years) of oral contraceptive use09781521.0§1601.0§321.0§,§§ ≤ 135101.00.4, 2.591.00.4, 2.600.70, 4.5>156131.60.7, 3.6121.60.7, 3.810.60.01, 4.8 p   trend‡‡0.230.37>0.99Use of an intrauterine deviceNever382511.0§451.0§141.0§Ever6881241.00.6, 1.61361.61.0, 2.7191.10.4, 3.0 Table continues   b  y  g u e s  t   on J  ul   y 2 2  ,2  0 1 1  a j   e. ox f   or  d  j   o ur n al   s . or  gD  ownl   o a d  e d f  r  om   Risk Factors for Fibrocystic Breast Conditions 949   Am J Epidemiol   2004;160:945–960 TABLE 2. Continued *  Women with missing values for any variable were excluded from analysis for that variable.† OR, odds ratio; CI, confidence interval.‡ All odds ratios were adjusted for age.§ Reference category.¶ Also adjusted for age at first livebirth.# Restricted to parous women. **  Test for trend; the reference category was included.†† Also adjusted for number of livebirths.‡‡ Test for trend among exposed women; the reference category was excluded.§§ Exact logistic regression.¶¶ At administration of the baseline questionnaire. ExposureNo. of controlsNonproliferative casesProliferative casesAtypia casesNo. of casesOR†,‡95% CI†No. of casesOR‡95% CINo. of casesOR‡95% CI Duration (years) of intrauterine device use0382511.0§451.0§141.0§ ≤ 9113401.70.9, 3.4281.50.7,, 4.010–14239401.00.5, 1.9612.31.2, 4.560.90.2, 3.7 ≥ 15295340.90.5, 1.8381.70.9, 3.471.60.4, 6.1 p   trend ** 0.530.050.50Duration (years) of living with a spouse who smoked0375611.0§681.0§131.0§,§§ ≤ 15295591.00.6, 1.6640.80.4, 1.4101.00.3, 3.316–20223351.30.7, 2.3350.80.4, 1.551.50.4, 5.8 ≥ 21171170.90.4, 1.8140.50.2,, 2.4 p   trend‡‡0.870.530.44First-degree relative with breast cancer No1,0531691.0§1761.0§321.0§,§§Yes 1763.80.9, 16.852.80.6, 13.613.20.04, 63.2EducationElementary school or less211131.0§161.0§61.0§,§§Middle or high school8271491.60.7, 3.81591.80.8, 4.1241.80.4, 8.3College or more31132.30.7, 7.460.70.2, 2.733.50.4, 27.4 p   trend ** 0.150.970.18Prior breast lump¶¶Never 1,0101531.0§1481.0§281.0§,§§Ever 32162.91.2, 7.2276.02.7, 13.534.50.6, 22.3Frequency of breast self-examination (times/year)0725641.0§661.0§201.0§,§§ ≤ 6137321.71.0, 3.1291.80.9, 3.351.10.3, 3.57–12198753.92.5, 6.3783.32.0, 5.581.40.5, 3.7 ≥ 137413.12.9, 58.5857.115.6, 209.006.20, 59.8 p   trend‡‡<0.001<0.0010.75Daily energy intake (kcal)<1,647.3267501.0§601.0§101.0§1,647.3–1,868.4268420.80.5, 1.5430.60.4,, 2.01,868.5–2,128.2268420.70.4, 1.2460.70.4,, 1.4>2,128.2267410.70.4, 1.3320.60.3,, 1.9 p   trend **   b  y  g u e s  t   on J  ul   y 2 2  ,2  0 1 1  a j   e. ox f   or  d  j   o ur n al   s . or  gD  ownl   o a d  e d f  r  om 
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks