Alcohol and breast cancer risk among Asian-American women in Los Angeles County

The study population and methods used in this population-based case-control study have been described previously [5]. In brief, breast cancer patients were identified by the LAC Cancer Surveillance Program (CSP), the population-based cancer registry covering Los Angeles, a member of the statewide California Cancer Registry, and the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program. Case patients were identified as Chinese, Japanese, or Filipino between the ages of 25 and 74 years, inclusive, at the time of diagnosis of an incident breast cancer. Cases were diagnosed between 1995 and 2001 and between 2003 and 2006. In total, we identified 3,797 eligible case patients (1,496 Chinese, 865 Japanese, 1,436 Filipino) and interviewed 2,303 cases (929 Chinese, 547 Japanese, 827 Filipino). Among those who did not participate, 869 declined to be interviewed (375 Chinese, 222 Japanese, 272 Filipino), 77 had died (17 Chinese, 24 Japanese, 36 Filipino), and 548 could not be located (175 Chinese, 72 Japanese, 301 Filipinos). The 2,035 control subjects (923 Chinese, 518 Japanese, 594 Filipino) were selected from the neighborhoods where the case patients resided at the time of diagnosis. A well-established algorithm was used to identify neighborhood controls that we have used in numerous case-control studies [6, 7]. We initially defined a specified sequence of houses to be visited in the neighborhoods where index cases lived at the time of diagnosis. We then sought to interview the first eligible resident in the sequence. If the first eligible control subject refused to participate, the second eligible one in the sequence was asked, and so on. Letters were left when no one was home, and follow-up was by mail and telephone. Controls were sought to frequency-match to the cases on specific Asian ethnicities and 5-year age groups. On average, a suitable control was identified after visiting a median of 60 households. Of the controls interviewed, 64% were the first-identified eligible control, 18% were the second-identified eligible control, and 18% were the third or later eligible control.

Cases and controls were interviewed in person by trained multilingual interviewers by using a standardized, structured questionnaire. In brief, the questionnaire elicited information on race/ethnicity, migration history, education, family history of cancer, menstrual and reproductive histories, use of exogenous hormones, smoking history, lifetime physical-activity patterns, birth weight, and height and weight history. We used two separate measures to assess alcohol intake. In the beginning of the interview, by using a quantitative food-frequency questionnaire (FFQ) that was modeled after the validated diet instrument used in the Multiethnic Cohort Study in Hawaii and Los Angeles [8], we asked women about usual intake of specific foods/food groups and beverages during the year before cancer diagnosis (for cases) or during the past year (for controls). Separate questions were asked regarding intake of beer, wine, and hard liquor. The FFQ recorded frequency of intake (never or < 1/month, 1 time per month, 2 to 3 times per month, 1 time per week, 2 to 3 times per week, 4 to 6 times per week, once per day, 2 or more times per day) for each food/beverage and the usual amounts of intake for at least monthly consumption. A second measure of alcohol intake was based on lifetime-history assessment; these questions were asked toward the end of the interview. Specifically, we asked if they ever drank alcoholic beverages on a regular basis, defined as at least once a week for 6 months or longer or at least 26 drinks per year before the reference date. For regular drinkers, we asked the age they first started drinking alcoholic beverages regularly, the age they last drank regularly, the total number of years they drank regularly, and the average number of drinks they consumed per day or per week when they drank regularly. These questions were asked separately for intake of beer, wine, sake, and hard liquor. Subjects were told that one drink is equivalent to one bottle or 12-oz can of beer, one glass (or 4 oz) of wine, 1-oz cup sake, or one cocktail or mixed drink of hard liquor (1.5-oz shot). Information on tumor characteristics including estrogen receptor (ER) and progesterone receptor (PR) status was routinely collected by the LAC-CSP as part of SEER study protocol.

Variables on body size, use of menopausal hormones, diabetes, selected dietary factors, and dietary pattern have been reported in the first group of study participants (1,384 cases, 1,225 controls) [9‐11]. The relation between birth weight and breast cancer risk has been reported in all participants [5]. The study was approved by the Institutional Review Board of the Keck School of Medicine at the University of Southern California. Informed consent was obtained from each case and control before her interview.

Blood specimens were collected from interviewed breast cancer cases and control women at the end of the interview. We conducted a cross-sectional study among a subset of postmenopausal control women in which sex hormones and growth factors were measured. Methods used to measure these biomarkers have been published in our previous studies [10‐12]. In brief, we used commercially available enzyme-linked immunosorbent assay (ELISA) kits to measure insulin-like growth factor (IGF)-1, IGF-binding protein (BP)-3, sex-hormone-binding globulin (SHBG), adiponectin, and dehydroepiandrosterone sulfate (DHEAS). Serum levels of estradiol (E2), estrone (E1), testosterone (T), and androstenedione (adione) were measured with previously described radioimmunoassay methods [13].

In total, 2,229 cases and 2,002 controls with complete information on lifetime alcohol intake and the covariates included for adjustment were included in this study (74 cases and 33 controls with missing data on alcohol intake or covariates were excluded). For drinkers, the number of grams of ethanol consumed per day was estimated by multiplying the number of drinks of each specific type of alcohol consumed per week by the appropriate number of grams of ethanol per drink, and then summing across the type of alcohol and dividing by seven. Ethanol values attributed to a 12-oz serving of beer, 4-oz serving of wine, 1-oz serving of sake, and 1.5-oz serving of hard liquor were 13, 11, 4, and 14 grams, respectively.

We calculated odds ratios (ORs; relative risk estimates) and their corresponding 95% confidence intervals (CIs) and P values by conditional logistic regression methods, with matched sets defined jointly by reference age (< 39, 40 to 44, 45 to 49, 50 to 54, 55 to 59, 60 to 64, 65 to 69, and 70+ years), and specific Asian ethnicity (Chinese, Japanese, Filipino). All regression models included the following covariates: birthplace and years of residence in the United States among non-US born (> 20 years, 11 to 20 years, ≤ 10 years), education (less than high school, high school, some college, college graduate), interviewer, age at menarche (< 12, 12 to 13, or ≥ 14 years), parity (0, one, two, three, or four or more births), current body mass index (BMI (kg/m²); quartiles of controls), years of regular (that is, at least 1 hour per week) recreational physical activity (< 5, 5 to 9, 10 to 19, or ≥ 20 years), total calories (continuous), intake of soy (quartiles), green tea (0, < 120 ml/day, or ≥ 120 ml/day), and black tea (0, < 120 ml/day, or ≥ 120 ml/day), menopausal status (premenopausal, natural menopause, bilateral oophorectomy, simple hysterectomy, or "hormone therapy menopause" defined as starting hormone therapy before periods had stopped), age at menopause (≤ 44, 45 to 49, 50 to 54, or 55+ years), and family history of breast cancer. We conducted case-control comparisons for all subjects combined, and separately by Asian ethnicity (Chinese, Japanese, Filipino), nativity (US born, non-US born), and by the six Asian ethnic/nativity groups, and menopausal status. To examine the potential effect modification of smoking, BMI, soy intake, folate intake, and use of menopausal hormones on association with alcohol and breast cancer, we separately constructed interaction terms between these factors and alcohol intake and used likelihood-ratio tests to assess the significance of these interaction terms. We used analysis of variance (ANOVA) and analysis of covariance (ANCOVA) to assess the relations between alcohol intake and serum hormone levels and other biomarkers in a subgroup of control women after adjusting for age, Asian ethnicity, migration history, education, age at menarche, parity, and current BMI. Hormone and other biomarker measurements were transformed logarithmically to achieve approximate normal distributions for statistical analysis. We evaluated whether the geometric mean of hormone and other biomarker levels differed by current drinking status. P values less than 0.05 are considered statistically significant, and all P values quoted are two-sided. All analyses were performed by using the EPILOG statistical software system (version 1.01s; Pasadena, CA, USA) and the SAS statistical software (version 8.0; SAS Institute, Cary, NC, USA).