Fax: +45-39-20 80-10
Interactions between intakes of alcohol and postmenopausal hormones on risk of breast cancer
Article first published online: 26 OCT 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 122, Issue 5, pages 1109–1113, 1 March 2008
How to Cite
Nielsen, N. R. and Grønbæk, M. (2008), Interactions between intakes of alcohol and postmenopausal hormones on risk of breast cancer. Int. J. Cancer, 122: 1109–1113. doi: 10.1002/ijc.23195
- Issue published online: 24 DEC 2007
- Article first published online: 26 OCT 2007
- Manuscript Accepted: 24 AUG 2007
- Manuscript Received: 11 JUL 2007
- Danish Ministry of the Interior and Health
- The Health Insurance Foundation
- breast cancer;
- alcohol intake;
- hormone use;
- menopausal status
Alcohol and postmenopausal hormone use are well-established modifiable risk factors for breast cancer. Alcohol may decrease the metabolic clearance of estradiol, whereby the risk of breast cancer associated with hormone use may depend on blood alcohol levels. The objective is to determine whether alcohol interacts with hormone use on risk of breast cancer. The 5,035 postmenopausal women who participated in the Copenhagen City Heart Study were asked about their alcohol intake and hormone use at baseline in 1981–1983 and were followed until 2002 in the Danish cancer registry, with <0.1% loss to follow-up. Proportional hazard models were used to analyze data. During follow-up, 267 women developed breast cancer. Alcohol consumption was associated with a small increased risk of breast cancer (hazard ratio = 1.11 per drink/day, 95% CI: 0.99–1.25). Women who used hormones also had a higher risk of breast cancer (HR = 2.00, 95% CI: 1.52–2.61) compared to nonhormone users. We found an interaction between these 2 factors; those who had an intake of more than 2 drinks per day and took hormones had a risk of 4.74 (95% CI: 2.61–8.59) for breast cancer compared to abstainers who did not use hormones. Alcohol was not associated with breast cancer among women who did not use hormones (HR = 0.98 per drink/day, 95% CI: 0.82–1.78). In conclusion, the interaction between alcohol and hormone use should, if confirmed in other studies, have an impact both on the prescription of hormones and on sensible drinking limits for postmenopausal women. © 2007 Wiley-Liss, Inc.
Breast cancer is the most common cancer in women and the most frequent cause of death among women below the age of 70 in most developed countries.1 The association between alcohol intake and breast cancer has been studied in several cohort and case-control studies and the evidence has been summarized in reviews and meta-analyses.2, 3, 4, 5, 6 The results suggest a small, but consistent increase in the risk of breast cancer with increasing alcohol intake. Although the increased risk of breast cancer associated with alcohol intake is relatively low, the association and shape of the risk function are of great importance because of the seemingly lack of a threshold, the large number of women drinking a small amount of alcohol and the high incidence of the disease. Together with alcohol, postmenopausal hormone is one of the important modifiable risk factors for breast cancer. In a number of studies, hormone use has been shown to imply a 20–100% increased risk of breast cancer, depending on type of hormone and length of therapy.7, 8, 9
Both alcohol and hormones seem to influence breast cancer risk through their ability to affect circulating levels of steroid hormones, especially estrogens. Hence, among postmenopausal women who take estradiol orally, a more than 3-fold higher serum estradiol has been observed in those who drink alcohol as compared to those who abstain.10 Only few studies have specifically addressed a possible interaction between alcohol and hormone use on the risk of breast cancer in large populations.11, 12, 13, 14 These studies have consistently found the combined effect of alcohol and hormones to be stronger than the effect of either one of them alone.11, 12, 13, 14 However, in a pooled analysis, the association between alcohol and breast cancer was not modified by hormone use6 and no interaction between alcohol and hormone use has been reported in several case-control studies.15, 16, 17 A possible explanation to the discrepancies is that women in several of the studies had a narrow range intake of alcohol, which may have impaired thorough analyses of a possible interaction with hormone use. The objective of the present study is to assess and quantify a possible interaction between alcohol and postmenopausal hormone use on risk of breast cancer in a prospective cohort study with long-term follow-up.
Material and methods
The Copenhagen City Heart Study was initiated in 1976. An age-stratified sample of 19,698 men and women were randomly drawn from the Central Population Registry and invited to participate in the study. In 1981–1983, the study population was supplemented with 500 men and women and additional study examinations were performed and the response rate at this examination was 70%. The 12,698 participants in the second examination counted 7,018 women and all participants gave informed consent. The study participants were asked about their weekly alcohol consumption and hormone use at the second examination and this examination is therefore used as baseline for the present study. The 5,297 postmenopausal women were included in the study. Women with breast cancer before baseline (n = 127) or with lacking information on alcohol consumption (n = 26) or other covariates (n = 109) were excluded, leaving 5,035 women for the analyses. A detailed description of the Copenhagen City Heart Study has previously been published elsewhere.18
Measure of alcohol consumption
As part of the self-administered questionnaire, the participants were asked about their average monthly, weekly, or daily consumption of beer (bottles), wine (glasses), or spirits (units). The total weekly alcohol consumption can be calculated by adding the reported average number of drinks from each type of alcohol. It is assumed that 1 bottle of beer contains 12 g of ethanol, which can also be considered the average for 1 glass of wine and 1 unit of spirit. One standard glass of wine is one sixth of a bottle of wine and a standard serving of spirits is 4 centilitres. For the analyses, alcohol consumption was classified into 5 categories according to total weekly intake of alcohol (drinks/week): less than 1, 1 to 7, 8 to 14, 15 to 21 and more than 21.
Measure of postmenopausal hormone use
The women answered questions about their menopausal status (pre/post) and whether they used any hormonal substitution (yes/no). Women who were postmenopausal and reported to use hormones were defined as users of postmenopausal hormones.
The following variables were considered as potential confounders for the analyses: body mass index (BMI) (<18.5, 18.5–24.9, 25–29.9, or ≥30 kg/m2), number of children (0, 1–2, 3 or more), physical activity in leisure time (none or very little activity; 2–4 hr of light activity per week; more than 4 hr of light activity or 2–4 hr of high level activity; and competitional level or more than 4 hr of hard level activity per week), perceived stress (continuous), tobacco smoking (never-smoker, ex-smoker, smokers of 1–14 g per day, 15–24 g per day and more than 24 g per day) and education (less than 8 years, 8–11 years or 12 or more years).
Participants were followed from date of the second examination till date of first diagnosis of primary breast cancer (n = 267), death (n = 2430), emigration (n = 6) or end of follow-up on December 31, 2002 (n = 2332). Using the civil registry number, which is unique to every Danish citizen, primary breast cancer events were identified through linkage to the Danish National Cancer Registry, which contains data on all cancer diagnoses in Denmark. The following ICD7-codes were used to identify primary invasive breast cancer cases: 170.0–170.5, 470.0–470.5 and 870.0–870.2. The vital status of the study population was followed in the Central Death Registry. Information on diagnoses of breast cancer was updated until 2002, making it possible to follow the women for 19 to 21 years for a first-time primary diagnosis of breast cancer.
Data were analyzed by means of Cox regression models with age as the time scale using SAS/STAT software version 8.2. Alcohol consumption and hormone use both met the assumption of proportional hazards. Initially, we estimated the age-adjusted hazard ratio of primary breast cancer associated with alcohol consumption and hormone use separately. Subsequently, a multivariate Cox regression model was fitted to adjust for potential confounding from baseline covariates. Trend analyses were used to address dose-response relations between alcohol and breast cancer. To address a possible interaction between alcohol consumption and hormone use in terms of breast cancer risk, we applied 2 different strategies. First, we assessed the association between alcohol consumption and breast cancer risk separately for women who used hormones and women who did not use hormones. Second, to examine biological interaction (defined as the coparticipation of 2 causes in the same causal mechanism), we redefined alcohol and hormone use into a single composite exposure variable and entered the combination into the proportional hazard model as a factored set of terms.19 To address the magnitude of the biologic interaction, we used a synergy index to assessed whether the relative excess risk for each of the component causes exceeded the sum of the relative excess risk for each of the components. A synergy index above 1 means positive interaction or synergy.
The women were aged from 39 to 91 years (mean 62 years) at baseline. Forty-six percentage of them were light to moderate drinkers while 48% were nondrinkers and 5% had an intake above the sensible drinking limits of 14 drinks/week (Table I). Twenty percentage of the women used hormones. The mean BMI was 25 kg/m2, with a total of 14% being obese. Heavy drinkers were slightly younger, had more frequent intake of hormones and a higher percentage were current smokers and had higher levels of stress than non- and moderate drinkers (Table I).
|Study population||Weekly alcohol intake (drink)/wk|
|Persons, n (% of study sample)||5035||2395 (48)||1764 (35)||580 (11)||175 (3)||121 (2)|
|Mean age (SD)||62 (8)||63 (8)||61 (8)||61 (8)||60 (7)||59 (7)|
|Postmenopausal hormone use, n (%)||1001 (20)||409 (17)||377 (21)||134 (23)||43 (25)||38 (31)|
|Nulliparous (%)||1199 (24)||556 (23)||440 (25)||122 (21)||50 (29)||31 (26)|
|Mean body mass index, kg/m2 (SD)||25 (5)||26 (5)||25 (4)||24 (4)||25 (4)||24 (3)|
|Obese (BMI < 30 kg/m2), n (%)||700 (14)||416 (17)||212 (12)||42 (7)||20 (11)||10 (8)|
|Current smokers, n (%)||2638 (52)||1179 (49)||919 (52)||343 (59)||111 (63)||86 (71)|
|Low education, n (%)||2620 (52)||1520 (63)||795 (45)||218 (38)||61 (35)||26 (21)|
|Physically inactive, n (%)||985 (20)||564 (24)||273 (15)||89 (15)||37 (21)||22 (18)|
|High level of stress, n (%)||560 (11)||297 (12)||161 (9)||53 (9)||24 (14)||25 (21)|
Alcohol and breast cancer
A total of 267 women developed breast cancer during follow-up. Intake of alcohol was associated with a small increased risk of breast cancer in a linear manner (p-value for trend: 0.06; Table II). A moderate intake (8–14 drinks per week) of alcohol was associated with a relative risk of breast cancer of 1.28 (95% CI: 0.87–1.89), whereas a high intake (more than 21 drinks per week) was associated with a relative risk of 1.54 (95% CI: 0.77–3.10) as compared to abstainers. When including alcohol as a continuous variable, 1 drink increase in alcohol intake per day implied an increase in risk of 1.11 (95% CI: 0.99–1.25).
|Cancers (n)||Incidence per 100,000 years||Age-adjusted HR (95% CI)||Multi-adjusted1 HR (95% CI)|
|Weekly alcohol consumption|
|<1 drink/wk (n = 2,395)||107||289||1 (reference)||1 (reference)|
|1–7 drinks/wk (n = 1,764)||101||346||1.23 (0.94–1.62)||1.19 (0.90–1.57)|
|8–14 drinks/wk (n = 580)||36||378||1.35 (0.93–1.97)||1.28 (0.87–1.89)|
|15–21 drinks/wk (n = 175)||14||483||1.72 (0.99–1.97)||1.61 (0.92–2.84)|
|>21 drinks/wk (n = 121)||9||479||1.79 (0.91–3.54)||1.54 (0.77–3.10)|
|Test for trend||0.01||0.06|
|Drink/day (continuous)||267||1.14 (1.02–1.27)||1.11 (0.99–1.25)|
|No (n = 4,034)||182||287||1 (reference)||1 (reference)|
|Yes (n = 1,001)||85||498||1.96 (1.50–2.55)||2.00 (1.52–2.61)|
Hormone use and breast cancer
Women who were users of hormones at baseline had a risk of breast cancer of 2.00 (95% CI: 1.52–2.61) compared to women who did not take hormones (Table II).
Alcohol and hormone use and risk of breast cancer
In stratified analyses, intake of alcohol meant no increased risk of breast cancer (HR = 0.98 per drink/day, 95% CI: 0.82–1.78) among nonusers of hormones (Table III). Hence, nonusers of hormones who had an intake of alcohol higher than the sensible drinking limits of 14 drinks per week had no increased risk of breast cancer (HR = 1.02, 95% CI: 0.51–2.04). This was in contrast to hormone users, where 1 drink increase in alcohol consumption was associated with a hazard ratio of 1.27 (95% CI: 1.09–1.49) and where drinking 15–21 drinks/week was associated with a markedly higher risk of developing breast cancer (HR = 3.23, 95% CI: 1.48–7.07). This implied more than multiplicative interaction (p = 0.11), despite the relatively low power to assess a statistical interaction.
|Cancers, n||Incidence per 100,000 years||Age-adjusted HR (95% CI)||Multi-adjusted1 HR (95% CI)|
|Women who used hormones at baseline|
|Weekly alcohol consumption|
|<1 drink/wk (n = 409)||28||410||1 (reference)||1 (reference)|
|1–7 drinks/wk (n = 377)||28||422||1.03 (0.61–1.73)||0.95 (0.55–1.64)|
|8–14 drinks/wk (n = 134)||15||665||1.62 (0.87–3.03)||1.67 (0.87–3.22)|
|15–21 drinks/wk (n = 43)||9||1297||3.25 (1.53–6.90)||3.23 (1.48–7.07)|
|>21 drinks/wk (n = 38)||5||796||2.06 (0.79–5.33)||2.17 (0.79–5.93)|
|p-value for trend||0.004||0.004|
|Drinks/day (continuous)||85||1.25 (1.09–1.43)||1.27 (1.09–1.49)|
|Women who did not use hormones at baseline|
|Weekly alcohol consumption|
|<1 drink/wk (n = 1,986)||79||262||1 (reference)||1 (reference)|
|1–7 drinks/wk (n = 1,387)||73||324||1.29 (0.94-1.78)||1.29 (0.93-1.79)|
|8–14 drinks/wk (n = 446)||21||289||1.16 (0.71-1.87)||1.10 (0.67-1.81)|
|15–21 drinks/wk (n = 132)||5||227||0.91 (0.37-2.24)||0.88 (0.35-2.18)|
|>21 drinks/wk (n = 83)||4||326||1.39 (0.51-3.79)||1.28 (0.46-3.57)|
|p-value for trend||0.58||0.79|
|Drinks/day (continuous)||182||1.01 (0.85-1.20)||0.98 (0.82-1.78)|
Women who drank 8–14 drinks per week and used hormones had a 3-fold higher risk of developing breast cancer (HR = 2.99, 95% CI: 1.69–5.29) than women who did not drink alcohol or used hormones (Fig. 1). Of the 83 women who used hormones and had an intake of more than 14 drinks per week, 18% developed breast cancer during the twenty years of follow-up, which corresponds to a hazard ratio of 4.74 (95% CI: 2.61–8.59) as compared to nondrinkers who took no hormones. The synergy index for the joint association of hormone use and an alcohol intake of more than 14 drinks per week was 4.19 (95% CI: 1.09–16.31), which implies more than additive interaction. This means that the joint association was 4 times as large as what would have been expected to be by the sum of the individual risks of hormone use and high alcohol consumption.
In this prospective study, we found large differences in the influence of alcohol on breast cancer risk among users and nonusers of hormones, and there was a strong interaction between these 2 modifiable risk factors. Heavy drinkers who used hormones had an almost 5 times higher risk of breast cancer as compared to nondrinkers who did not use hormones, and a markedly higher breast cancer risk was also observed for hormone-users with even a moderate alcohol consumption of between 1 and 2 drinks daily. No increased risk of developing breast cancer was observed for women who consumed alcohol, but did not use hormones.
Four other prospective studies have suggested that use of hormones and ingestion of alcohol may synergistically enhance the risk of breast cancer.11, 12, 13, 14 In the Iowa Women's Health Study cohort, consumption of more than 15 g of alcohol per day was associated with a relative risk of 1.83 (95% CI: 1.18–2.85) among ever users of noncontraceptive estrogens, whereas there was no association between alcohol consumption and breast cancer risk in never users of hormones.13 A recent study from Sweden also found a statistically significant multiplicative interaction between alcohol intake and use of postmenopausal hormones on risk of estrogen receptor (ER) and progesterone receptor (PR) positive tumors.11 They found that consumption of more than 10 g of alcohol per day and ever having used hormones was associated with about a doubling of the risk of ER+PR+ tumors as compared to abstainers who had never used hormones.11 In the Nurses' Health Study, current long-time users of postmenopausal hormones who consumed more than 20 g of alcohol daily had a relative risk of 1.99 (95% CI: 1.42–2.79), compared to nondrinking nonhormone users, but the assessment of multiplicative interaction between alcohol and hormone use was not statistically significant.12 Results from the Women's Health Study have also suggested that the use of both estrogen and alcohol may increase the risk of breast cancer more than either one of them alone.14 Although none of the previous studies have quantified the interaction between alcohol and hormone use in terms of combined risks as we do in the present study, the suggested interactions in previous studies support our findings and makes it a relatively consistent finding with important policy implications.
In postmenopausal women who use hormones, alcohol consumption may have a pronounced effect on the blood estradiol concentration because of altered clearance of estradiol. Therefore, women who drink alcohol and use hormones may have a much higher estradiol concentration than anticipated from use of hormones alone.10 This may be one explanation for the very high risk of breast cancer in this group of women.
The alcohol-breast cancer relation has been a matter of debate for years. It has been discussed whether there is a linear increase in risk of breast cancer, as suggested by recent meta-analyses2, 3, 5, 6 or a threshold effect, as suggested by several cohort studies.20, 21, 22 In particular, the results arising from different populations seem to diverge. If there truly is a strong interaction between hormone and alcohol use, and hormone use is differently distributed in the varying populations studied, our results may give an explanation for the diverging results. Our findings support the epidemiological evidence suggesting a linear relation between alcohol consumption and the risk of breast cancer in women who use hormones, but not in women who do not use hormones. In 2002, a meta-analysis by the Collaborative Group on Hormonal Factors in Breast Cancer found that the relative risk of breast cancer increases by 7.1% for each additional 10 g per day intake of alcohol,2 which is similar to our findings in the analysis where we do not take hormone use into account.
We had no information of type or duration of hormones used, which may have led to some exposure misclassification. Furthermore, in this study the women were asked about drinking habits and hormone use at one point in time and the results were used as an indicators of the general level of exposure to these factors. This implies a certain risk of misclassification, because the general level of alcohol consumption or hormone use could change during follow-up. However, these forms of misclassifications are most likely to result in lower or more conservative estimates than the true ones.
We did not have information on important biological risk factors for breast cancer, such as family history of breast cancer, age at menarche and age at first birth. To confound the analyses, however, these factors would also have to be related to alcohol consumption and hormone use. Experience of breast cancer in a near relative may have resulted in a higher awareness of risk factors for breast cancer, and if anything, we would expect women with a family history of breast cancer to consume less alcohol and be more reluctant to the use of hormones. This would attenuate the true relation between alcohol, hormones and breast cancer, and confounding from this factor is therefore unlikely to explain our results. Furthermore, we find it unlikely that neither age at menarche nor age at first birth would be strongly associated with alcohol intake and hormone use later in life. Another limitation of this study was the small number of breast cancer cases in some of the alcohol categories, especially after stratifying on hormone use, which resulted in a low statistical power to test for interactions.
The prospective design of the Copenhagen City Heart Study ensured temporality between alcohol consumption, hormone use and first-time incidence of breast cancer, and linkage of civil registry numbers to nationwide population-based registers enabled identification of virtually all cases of breast and allowed for nearly complete long-term follow-up.
In conclusion, we find an interaction between alcohol consumption and hormone use on risk of breast cancer among postmenopausal women. We report a linear dose-response relation between alcohol intake and risk of breast cancer among users of postmenopausal hormones, but not among nonusers of hormones. If these results are confirmed in similar studies, the findings should have an impact both on prescription of postmenopausal hormones and on sensible drinking limits for postmenopausal women.
The original data for the present study were collected and provided by the Copenhagen City Heart Study http://www.copenhagencityheartstudy.dk/.
- 9Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and hormone replacement therapy: collaborative reanalysis of data from 51 epidemiological studies of 52,705 women with breast cancer and 108,411 women without breast cancer. Lancet 1997; 350: 1047–59.
- 18The Copenhagen City Heart Study. A book of tables with data from the first examination (1976–78) and a five year follow-up (1981–83). Scand J Soc Med 1989; 170: 1–160., , , , , for
- 19Measuring interactions. In: RothmanK, ed. Epidemiology: an introduction. New York: Oxford University Press, 2002. 168–80..