To our knowledge, this is the first study to prospectively examine the effects of alcohol with proliferative BBD in relation to subsequent breast cancer risk. The results of this study indicate that recent alcohol consumption does not increase the risk of breast cancer to a greater extent among women with proliferative BBD than in women with nonproliferative BBD.
Previous epidemiologic studies have examined the effect of alcohol consumption among women with a self-report of BBD, with inconsistent results [1
]. In a cohort study conducted in the Netherlands, van den Brandt and colleagues reported a nonsignificant 2.5-fold increased risk of breast cancer among women with BBD who drank ≥ 15 g of alcohol per day compared with nondrinkers (P
for trend = 0.15) [16
]. A pooled analysis of six cohort studies, including the Netherlands cohort just mentioned, reported no significant interaction between alcohol consumption and BBD with respect to breast cancer risk (P
= 0.23) [1
]. In the California Teachers Study, the joint effect of history of BBD and high alcohol consumption was associated with a twofold increased risk of breast cancer (relative risk (RR) = 1.97; 95% CI 1.39 to 2.79), while nondrinkers with biopsy-confirmed BBD had a 35% increased risk of breast cancer (RR = 1.35; 95% CI 1.05 to 1.73) in comparison with nondrinkers with no history of BBD [17
Benign breast conditions are a heterogeneous group of diseases and therefore may not all respond to alcohol exposure in the same manner. Although all of the women in this study had a biopsy removing their benign lesion, we were operating under the assumption that the BBD is a marker of susceptibility and the remaining breast tissue may have a similar susceptibly to the effects of alcohol later in life. Nonproliferative breast diseases comprise a large proportion of reported BBD biopsies. The inconsistencies observed between our study and previous studies may be due to the heterogeneous nature of BBD and variability in distribution of BBD types between different studies.
One proposed mechanism by which alcohol may influence breast cancer risk is by increasing circulating estradiol levels, as has been observed in controlled feeding studies in both premenopausal and postmenopausal women [6
]. A second potential mechanism is that alcohol may function as a cocarcinogen, inhibiting detoxification of carcinogens, or by impairing clearance of carcinogens [3
]. There are data suggesting that alcohol may act early in the carcinogenic process [6
], as well as later, functioning as a tumor promoter [3
The current study failed to support our a priori hypothesis and suggests that recent alcohol consumption does not contribute additional risk to women with proliferative breast disease. One explanation may be that women with proliferative breast diseases are further along on the continuum to breast cancer, and are already at an elevated risk of breast cancer, which is no longer affected by alcohol. In contrast, women with nonproliferative breast disease are not as far along the pathway to breast cancer and specific exposures such as alcohol may exhibit harmful effects and influence the risk of breast cancer.
Byrne and colleagues conducted a similar study in the Nurses' Health Study to examine the effects of postmenopausal hormone use among women with BBD [25
]. Similarly, the results were contrary to their a priori
hypothesis and they observed no elevated risk of breast cancer among women with proliferative breast diseases according to current use of hormones or duration of use, again suggesting that these women may be at such an increased risk of breast cancer that the additional effect of exogenous hormone use is minimal or none.
In Figure , we have schematically described the evidence relating alcohol exposure to breast cancer. It is well accepted that recent alcohol consumption increases the risk of breast cancer [1
] and there is little evidence that alcohol intake early in life affects breast cancer risk [17
]. The three studies examining alcohol and proliferative breast disease suggest that recent alcohol intake does not increase the risk of proliferative breast disease [13
] and may in fact be inversely related to it, with estimates ranging from a 10% [13
] to a 77% [14
] reduction in risk comparing the highest category of alcohol consumption with nondrinkers. These results, along with those from the current study, suggest that the association observed between recent alcohol consumption and breast cancer may not be mediated through proliferative breast disease. However, Byrne and colleagues examined alcohol consumption between the ages of 18 and 22 and proliferative breast disease and reported a 30% increased risk among women consuming ≥ 15 g of alcohol per day compared with nondrinkers (RR = 1.33, 95% CI 1.05 to 1.69) [13
], suggesting that only very early alcohol consumption may affect proliferative BBD.
Schematic description of relation between alcohol consumption, proliferative benign breast disease (BBD), and breast cancer. Solid lines indicate well-established associations and dotted lines indicate less-established relations.
The progression from tumor initiation to breast cancer is not well defined. One hypothesis is that the pathway from normal tissue to breast cancer arises from a series of preinvasive lesions: benign proliferative changes, atypical hyperplasia, and carcinoma in situ. If all breast cancers arise from benign lesions, the results of this study, in conjunction with other data, imply that there may be a narrow window of time when alcohol consumption affects breast cancer risk. An alternative explanation may be that there are multiple pathways to breast cancer and pathways involving proliferative benign breast lesions may not be influenced by alcohol consumption. A third possibility is that the magnitude of risk associated with benign proliferative lesions, and in particular atypical hyperplasia, is so much greater than the effect of alcohol that we were underpowered to detect more subtle increases in risk attributable to alcohol in this study.
Although this is one of the largest studies of its kind, the number of women with proliferative breast disease and high alcohol consumption was small. We were underpowered to examine this relation in greater detail with regard to menopausal status and estrogen receptor status of tumors. A potential concern of the study is that the final study population represents 37% of those originally selected. The study protocol required collection of pathologic specimens in order to have unified review of histologic sections. The major limitation of this is that many hospitals routinely destroy specimens after 5 or 10 years. As a result, many potential cases and controls were excluded, which could result in potential selection bias. There were no significant differences in reasons for loss comparing cases to controls, indicating that any differences that do occur are likely to be due to chance.