Benign breast disease, recent alcohol consumption, and risk of breast cancer: a nested case–control study

The Nurses' Health Study (NHS I) cohort was initiated in 1976, when 121,700 US registered nurses ages 30 to 55 returned an initial questionnaire. The Nurses' Health Study II (NHS II) is also an ongoing cohort study of over 116,000 US female nurses who were 25 to 42 years of age in 1989 when the study was initiated. Every 2 years, information on reproductive variables, body mass index, exogenous hormone use, and disease outcomes is collected in both cohorts. Semiquantitative food frequency questionnaires were included as part of the biennial questionnaire in 1980, 1984, 1986, and 1990 in NHS I, and in 1991 in NHS II. Alcohol consumption was also assessed on the baseline 1989 questionnaire in NHS II. The methods developed to follow up participants and confirm incident cancers and death in the Nurses' Health Study have been described previously in detail elsewhere [22] and have been applied to NHS II.

Beginning with the initial NHS I questionnaire in 1976, participants have been asked on every biennial questionnaire to report any diagnosis of fibrocystic disease or other BBD. Early questionnaires (1976, 1978, and 1980) asked whether the respondent had ever been diagnosed as having 'fibrocystic disease' or 'other BBD' and whether she had been hospitalized with this diagnosis. Beginning in 1982, the NHS I questionnaires sought specific details of a history of biopsy-confirmed BBD. The initial 1989 NHS II questionnaire and all subsequent biennial questionnaires also asked participants to report any diagnosis of BBD and to indicate if it was confirmed by biopsy or aspiration.

We conducted a case–control study nested within the subcohort of women with a biopsy-confirmed BBD. Incident breast cancer cases in both cohorts were identified through the nurses' own reports and were confirmed by review of medical records. Cases in the study are cases of breast cancer diagnosed by 1 June 1994 (NHS I) or 1 June 1995 (NHS II) with a previous confirmed BBD biopsy and available pathology specimens (histologic sections and/or tissue blocks). Controls are women who did not have a diagnosis of breast cancer when the case was diagnosed and also had a previous biopsy-confirmed BBD and available pathology specimens. Controls were matched to cases on year of birth and year of biopsy. Attempts were made to identify four matched controls for each case, although this was not always possible. This study was approved by the Committee on Human Subjects at Brigham and Women's Hospital.

We identified incident confirmed breast cancer cases diagnosed after the return of the initial questionnaire through the 1994/1995 follow-up cycle and controls that also reported a previous biopsy-confirmed BBD. A total of 1,080 cases were originally identified for this study, and 4,353 controls matched on age and year of benign breast biopsy were selected. After selection, 2.1% were found to be no longer eligible and were excluded. Seventy-five percent of eligible participants confirmed their BBD biopsy and granted permission to review their pathology slides. In response to our slide requests, we received pathology specimens for 55% of the study subjects. The primary reason given by pathology departments for not sending slides or tissue blocks was that the specimens had been destroyed or were no longer available (33%). There were no significant differences in the success of obtaining slides for breast cancer cases and controls. A detailed description of the steps of specimen collection and exclusions of selected cases and controls during the collection process are included in Table 1. An additional 156 women were excluded for the following reasons: reported bilateral BBD, but tissue specimen did not identify breast side (n = 2), report of breast cancer either before study start or after the 1994/1995 follow-up cycle (n = 10), and date of benign breast diagnosis was within six months of the breast cancer diagnosis date (cases) or index date (controls) (n = 144).

Slides were coded by a research assistant and submitted to one of two collaborating pathologists (SJS, JLC) in a blinded fashion. The pathologists independently reviewed the BBD biopsy slides. Any slide identified as having either questionable atypia or atypia was jointly reviewed by the two pathologists. For each set of slides reviewed, a detailed work sheet was completed quantifying the extent of proliferative and atypical changes, morphologic features, and other details. BBDs were classified according to the Page classification system [23] into one of three categories: nonproliferative, proliferative without atypia, and atypical hyperplasia. Upon pathology review, it was determined that 29 women with an original BBD diagnosis had evidence of carcinoma in situ (n = 25) or invasive carcinoma (n = 4). These women were excluded from further analysis. There were 353 breast cancer cases and 1,495 controls with pathology slides available from their BBD biopsy.

Information on alcohol consumption was obtained from semiquantitative food frequency questionnaires. In NHS I, questions regarding alcohol consumption were asked in 1980, 1984, 1986, and 1990. Women were asked about their average consumption of beer, wine, and liquor separately in the prior year. One drink was considered equal to one can or bottle of beer, a 4-ounce glass of wine, or one drink or shot of liquor. Participants were asked to select from the following categories: almost never, 1 to 3 per month, 1 per week, 2 to 4 per week, 5 to 6 per week, 1 per day, 2 to 3 per day, 4 to 6 per day, 6 or more per day. Similarly, women in NHS II answered questions on consumption of alcohol in the 1989 and 1991 questionnaires. In 1991, the NHS II alcohol questions were expanded to include red wine, white wine, light beer, regular beer, and liquor. Total alcohol consumption per questionnaire cycle was calculated by adding the alcohol contributions from beer, wine, and liquor. Grams of ethanol per day were calculated based on the following equivalents of 12.8 g for regular beer, 11.3 g for light beer, 11.0 g for wine, and 14.0 g for liquor. Alcohol consumption in NHS I has been shown to be valid and highly reproducible in repeated assessments [24].

Women were assigned the alcohol exposure from the cycle preceding the diagnosis of breast cancer. Cases and controls in NHS I with diagnosis (or index) dates preceding the initial food frequency questionnaire administered in 1980 were excluded from the analysis (n = 36 cases, 129 controls). If alcohol consumption was missing from the questionnaire before the index date, the exposure from the preceding cycle was assigned. The Spearman correlation between reported alcohol consumption in questionnaire cycles was 0.80 (P<0.0001) or greater for all consecutive cycles. Women with missing alcohol exposure data from two cycles preceding the index date were excluded from the analyses (n = 35 cases, 143 controls). Thus, there were 282 patients with breast cancer and 1,223 controls eligible for this study with both BBD pathology and detailed alcohol exposure preceding their index date.

Odds ratios (ORs) and 95% confidence intervals (CIs) were determined using logistic regression models controlling for matching factors (age, year of BBD biopsy) and follow-up time using the SAS software package (version 8.0; SAS institute, Cary, NC). Follow-up time was defined as the time from BBD diagnosis to breast cancer diagnosis date (cases) or index date (controls). Information on potential confounding variables was obtained from responses on biennial questionnaires. For each 2-year cycle of case–control selection, covariate information was determined from the responses on questionnaires immediately preceding the cycle in which the breast cancer case was diagnosed. In addition to matching factors and follow-up interval, multivariate analyses were adjusted for the following confounders and breast cancer risk factors: first-degree family history of breast cancer (yes/no); quartiles of body mass index (≤ 21.6, 21.7 to 23.6, 23.7 to 26.6, ≥ 26.7 kg/m²); age at menarche (<12, 12, 13, ≥ 14 years); age at first birth/parity (nulliparous; one to four children, and age at first birth if ≤ 24 years of age; one to four children, and age at first birth if >24 years of age; five or more children, and age at first birth if ≤ 24 years of age; five or more children age at first birth >24); duration of postmenopausal hormone use (never, <5 years, ≥ 5 years); and menopausal status/type of menopause (premenopausal, natural, bilateral oopherectomy, other or unknown type of menopause).

Analyses examining the combined effect of benign breast histology and alcohol consumption were based on a cross-classified variable that used nonproliferative breast disease and consumption of 0 g of alcohol per day as the reference group. To examine whether the association between alcohol consumption and breast cancer was modified by benign breast histology, we conducted a likelihood ratio test to assess statistical significance of an interaction term using an ordered scale for benign breast histology and alcohol consumption.