Alcoholic beverage consumption and female breast cancer risk: A systematic review and meta‐analysis of prospective cohort studies

Abstract

Alcohol consumption is an established cause of female breast cancer. This systematic review examines in detail the association between alcohol and female breast cancer overall and among the described subgroups, using all of the evidence to date. A systematic review of PubMed and Embase was performed according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines. The search included articles published up to November 15, 2023. Meta‐analyses and regressions were performed for alcohol consumption of less than 1 standard drink (10 g of ethanol) per day and for a range of alcohol consumption categories in relation to breast cancer. Analyses by menopausal status, hormone receptor status, human epidermal growth factor receptor 2 status, and molecular subtype were performed. The search yielded 5645 publications, of which 23 publications of individual and pooled studies examined the association between overall alcohol consumption and breast cancer incidence. The meta‐regression showed a positive association; relative risks (RR) of breast cancer were 1.05 (95% CI: 1.04, 1.06), 1.10 (95% CI: 1.08, 1.12), 1.18 (95% CI: 1.15, 1.21), and 1.22 (95% CI: 1.19, 1.25) for 0.5, 1, 2, and 3 standard drinks per day compared with nondrinking, respectively. A meta‐analysis of nine studies indicated that for consumption of less than one standard drink per day, the RR estimate of breast cancer was 1.04 (95% CI: 1.01, 1.07) compared with nondrinking. Consumption of an additional 1 standard drink per day was associated with a higher risk of premenopausal (RR: 1.03 (95% CI: 1.01, 1.06)) and postmenopausal (RR: 1.10 (95% CI: 1.08, 1.12)) breast cancer. Alcohol consumption increases female breast cancer risk, even for women who consume one drink per day. Furthermore, alcohol consumption is associated with both pre‐ and postmenopausal breast cancer risk. These findings support evidence‐based cancer prevention guidelines to reduce alcohol‐related risks.

Based on all available evidence from prospective studies of breast cancer incidence published up to November 15, 2023, this review confirms that alcohol is a risk factor for female breast cancer, including postmenopausal and premenopausal cases. Additionally, the review found that consuming less than 1 standard drink per day (10 g of ethanol) significantly increases the risk of breast cancer. Public health information campaigns should be implemented to raise awareness of the impact of alcohol consumption on breast cancer risk.

INTRODUCTION

While higher levels of alcohol consumption are linked to greater harm (Rehm et al., 2017), it remains unclear whether there is a level of alcohol intake that poses no risk of any health consequences (Anderson et al., 2023). The World Health Organization stated that “no level of alcohol consumption is safe when it comes to human health” (Anderson et al., 2023). This statement is based on the fact that alcohol was established as a Group 1 human carcinogen by the International Agency for Research on Cancer (IARC) in 1988 (International Agency for Research on Cancer, 1988) and last updated in 2012 (International Agency for Research on Cancer, 2012), and accruing evidence that a large number of alcohol‐attributable cancers are caused by levels of drinking that are not classified as heavy drinking (International Agency for Research on Cancer, 1988, 2010; Rumgay et al., 2021).

The causal association between consumption of alcoholic beverages and female breast cancer risk was established by IARC in 2007 (International Agency for Research on Cancer, 2010). Additionally, the World Cancer Research Fund (WCRF) Continuous Update Project (CUP) found convincing evidence that any amount of alcohol consumption increases the risk of breast cancer (World Cancer Research Fund/American Institute for Cancer Research, 2018a). A 2016 pooled analysis of individual‐level data from 20 prospective cohort studies of breast cancer incidence found a dose–response association up to 55 g/day, after which the relative risk of breast cancer plateaued (Jung et al., 2016). In that analysis, even low amounts of consumption were associated with a higher risk (Relative Risk (RR): 1.03 (95% CI: 1.00, 1.06) for >0 to <5 g of ethanol per day compared with nondrinking) (Jung et al., 2016). Further, the RR for 10 g of ethanol per day increase was greater for post‐menopausal (RR: 1.09; 95% CI: 1.07, 1.11) than for premenopausal (RR: 1.03; 95% CI: 0.99, 1.08) breast cancer risk (Jung et al., 2016).

To further clarify the association between alcoholic beverage consumption and breast cancer risk, based on all of the evidence available from prospective studies of breast cancer incidence published up to November 15, 2023, the present systematic review was designed to assess (i) the association between consumption of less than 1 standard drink (defined according to the World Health Organization's call for an international definition of a standard drink as 10 g of ethanol) per day and breast cancer incidence compared with not consuming alcohol, (ii) the dose–response relationship between alcohol consumption and the risk of breast cancer, (iii) whether the risk of breast cancer is significantly associated with heavy episodic drinking (HED), age at initiation (i.e., age at first drink), or drinking between menarche and first birth, (iv) whether alcohol consumption is associated with the risk of both premenopausal and postmenopausal breast cancer, and (v) whether alcohol consumption is associated with breast cancer by hormone receptor status, human epidermal growth factor receptor 2 status, and molecular subtype.

METHODS

The present systematic review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines (Moher et al., 2009). This work is a sub‐study of a systematic review pre‐registered with PROSPERO (CRD42023431730) and is limited to prospective cohort studies that examined the association between alcohol consumption and breast cancer incidence.

For the purposes of this review, alcohol consumption was defined as average alcohol intake (measured in grams of ethanol per day), prevalence, frequency, or intensity of heavy episodic drinking (HED), age at initiation (i.e., age at first drink), and drinking between menarche and first birth (yes or no, and average alcohol intake in grams of ethanol per day). Alcohol use was not separated by types of alcoholic beverages (i.e., beer, wine, spirits, and others). While some articles have hypothesized that antioxidants found in red wine have potential chemo‐preventive effects (Kraft et al., 2009; National Cancer Institute, 2002), a margin of exposure for cancer between ethanol and resveratrol was over 100,000 (i.e., for every cancer prevented by resveratrol, ethanol causes over 100,000 new cancer cases) (Lachenmeier et al., 2014). Therefore, the cancer preventative effects of resveratrol in red wine are negligible compared to the cancer causative effects of ethanol.

For the purposes of this review, breast cancer was defined as the incidence of breast cancer (i.e., when abnormal breast cells grow out of control and form tumors (National Cancer Institute, 2024)). For the definition of breast cancer, no limitations were established for the ICD‐10 coding, invasive status of the breast cancer, or how the breast cancer was diagnosed.

Search strategy

A systematic electronic search was performed using PubMed and Embase. The search was limited to articles published up to November 15, 2023. There were no restrictions by age group, geographic location, or language. The search strategy and the keywords used are presented in Tables S1 and S2. References contained in relevant articles were also checked for any additional relevant articles that were missed by the systematic search.

Inclusion and exclusion criteria

Studies were included if the following criteria were met: (1) the study was a prospective cohort study or reanalyzed data from individual prospective cohort studies; (2) the exposure was alcohol consumption; (3) the outcome was breast cancer incidence; and (4) the study reported alcohol and breast cancer risk ratio(s), relative risk(s), incident rate ratio(s), or hazards ratio(s), with their corresponding error estimates or confidence intervals.

Studies were excluded if the results were beverage‐specific only (i.e., wine, beer, or spirits only). Preprints and other unpublished articles that had not undergone peer review were excluded from the systematic review due to the potential for changes in results during the review process, which could lead to discrepancies between the preprint and the final publication (Brietzke et al., 2023). Additionally, the quality of these preprints may be questionable (Brietzke et al., 2023). These were not included in the systematic review.

In cases where there was more than one publication or analysis reporting RR data from the same cohort, for each such analysis we prioritized first pooled analyses over individual analyses, second the most recent analyses of pooled or individual cohort data, and thirdly the analyses that controlled for the greatest number of relevant confounders. In cases where pooled analyses existed, we excluded all other analyses from the individual cohorts included in the pooled analyses. The selection of the most relevant analyses was performed separately for each analysis (i.e., a separate selection process was performed to determine the publications included in the meta‐analysis for the RR of breast cancer for consuming less than 10 g of alcohol per day and for the RR for alcohol consumption and premenopausal breast cancer, etc.). Menopausal status was defined as pre‐menopausal or post‐menopausal and did not include peri‐menopausal status.

Selection of studies, data extraction, and quality assessment

Selection of studies, data extraction, and quality assessment were performed in duplicate by two of nine trained researchers (Ivneet S, MS, LV, AF, GS, MJ, AI, HS, and KS). Articles were assessed for inclusion based first on title and abstract. Subsequently, full‐text articles were assessed for eligibility; any discrepancies were resolved through discussion with the senior author. From each article that met the inclusion criteria, the authors' names, country, age of participants, recruitment method, inclusion and exclusion criteria, sample size, follow‐up time, exposure measurement, outcome assessment, effect modifier measurement, risk relationships (i.e., risk ratios, RRs, incident rate ratios, or hazards ratios), and conflicts of interest were extracted. All extracted data were cross‐checked by the senior author.

Data synthesis

Studies that examined the relationship between drinking less than 1 drink per day and breast cancer risk were pooled using DerSimonian‐Laird random‐effect meta‐analysis. For this analysis, all estimates where the upper bound of consumption was less than 10 g of alcohol per day were included. In cases where multiple estimates were published for alcohol consumption of <10 g/day (e.g., a RR estimate for >0 to 4.9 and >4.9 to 9.9 g/day), all relevant RR estimates were included.

A sensitivity analysis was conducted to assess the relationship between consuming less than one drink per day and breast cancer risk. This involved two approaches: (i) a meta‐regression was employed to account for multiple estimates (from different alcohol consumption strata) originating from the same study, and (ii) a meta‐regression was also used to address differences in the amount of alcohol consumed by utilizing the midpoint of each consumption stratum and account for multiple estimates (from different alcohol consumption strata) originating from the same study.

For all other analyses, studies were pooled using DerSimonian‐Laird random‐effect meta‐regression. Meta‐regressions accounted for multiple estimates from the same study by applying random effects at the study level. Variations among the results of the studies were quantified using the I ² statistic (Higgins & Thompson, 2002). Examination of potential publication bias was performed using Egger's regression‐based test (Egger et al., 1997). Meta‐regressions included a regression term for grams per day. Heterogeneity in the results of breast cancer estimates resulting from factors such as confounding, study design, and populations surveyed was accounted for using a random‐effects model (Dettori et al., 2022).

For categorical outcomes, meta‐analyses were conducted on the natural log scale. Relative risks for categories of alcohol consumption with lower and upper boundaries were transformed into RRs for the category midpoints. Point estimates for categories with no upper boundary were estimated by adding the lower boundary of the highest category and half of the range of the second highest category. For dose–response analyses, to model the RR curves nonlinearly, we utilized restricted cubic splines (Royston & Lambert, 2011). To examine which restricted cubic spline was the most accurate, we utilized a three, four, and five knot spline (with knot placements based on Royston & Lambert, 2011). The number of knots that produced the most accurate function to model the relationship between alcohol consumption and risk of breast cancer incidence was determined by comparing the Akaike information criterion and the Bayesian information criterion across models. Based on this comparison, a three‐knot spline was utilized in the final model. To assess where the RR changed at 10, 20, 30, and 40 g/day, we performed a backward selection meta‐regression using a term for alcohol consumption in grams per day as well as regression terms that accounted for each additional g/day above 10, 20, 30, and 40 g/day.

The analysis also assessed the association between consuming less than 1 standard drink per day and breast cancer risk compared with not consuming alcohol (World Health Organization, 2024). In cases where the reference group was not current abstainers or lifetime abstainers, the RRs were adjusted by dividing all categorical RRs by the RR estimates for current abstainers or lifetime abstainers (i.e., so that the RR for lifetime abstainers was equal to 1.00). To estimate the uncertainty of these adjusted RRs, we utilized 1000 sets of simulated RR estimates with the simulated RRs sampled from their respective error distributions. These simulated RRs were then used to estimate a set of 1000 RRs that were adjusted so that the RR for current abstainers or lifetime abstainers was equal to 1.00. The 2.5th and 97.5th percentiles of these 1000 RRs were then used to estimate the 95% confidence intervals (CIs).

All analyses were performed using STATA statistical software, version 14, and p < 0.05 (two‐sided) was used as the definition of statistical significance.

Risk of bias in included studies

The Newcastle‐Ottawa Scale was used to assess the risk of bias in included studies (Peterson et al., 2011). The meta‐analyses did not adjust for the risk of bias in a study.

RESULTS

The search strategy yielded 5645 publications after duplicates were excluded (see Figure 1 for the PRISMA flow chart). Based on title and abstract screening, 4952 of these studies were excluded, and 692 underwent full‐text assessments. After completion of the full‐text assessments, a total of 42 unique publications were included in the current review (see Table S3 for the publications included in each analysis). Table S4 outlines the operationalization and tools used to measure alcohol consumption, the definitions and methods used to diagnose breast cancer, as well as the recruitment and follow‐up period for all included studies.

FIGURE 1.

PRISMA flow diagram for the systematic search.

A total of 23 publications, which encompassed three analyses of pooled cohorts and 20 analyses of individual cohorts, provided RR estimates for categories of alcohol consumption (see Table 1 and Table S5). Of these studies, 19 studies provided risk estimates for breast cancer overall, one study provided risk estimates stratified by BRCA carrier status only, and three studies provided risk estimates stratified by menopausal status only. Nineteen studies measured alcohol consumption defined by current alcohol consumption among drinkers and current abstention, three measured alcohol consumption defined by current alcohol consumption among drinkers, lifetime abstention, and former drinkers, and one examined lifetime alcohol consumption. Furthermore, 11 studies examined the risk of all breast cancers, 10 examined the risk of invasive breast cancers only, and two studies did not indicate whether the analysis was limited to invasive breast cancers (Figure 2).

TABLE 1.

Characteristics of cohort studies measuring the risk relationship between alcohol consumption and breast cancer risk.

Author	Cohort [location]	Years of study	Sample size	Exposure (g/day)	Cases	Relative risk (95% CIs)	Confounders adjusted for
Simon et al. (1991)	Tecumseh community cohort [USA]	Baseline: 1959 to 1960 End of follow‐up: 1987 to 1988	2299	Lifetime abstainer	657	1.00	Age, BMI, subscapular and triceps skinfold measurements, education levels, cigarette use, family history of breast cancer, age at menarche, mother's age at first live birth, and parity
				Former drinker	199	0.93 (0.40, 2.18)
				0 to <14	924	1.08 (0.64, 1.82)
				14 to <28	97	1.23 (0.49, 3.10)
				≥28	37	1.12 (0.25, 5.01)
Byrne et al. (1996)	NHANES I/NHEFS cohort [USA]	Baseline: 1982 to 1984 End of follow‐up: 1986 to 1987	6156	Abstainer	29	1.00	Age
				>0 to 4	12	0.70 (0.40, 1.40)
				>4 to 14	4	0.61 (0.20, 1.70)
				>14	7	1.40 (0.60, 3.20)
Zhang et al. (1999)	Framingham Heart Study [USA]	Baseline: Original cohort: 1948, Follow‐up cohort: 1971, End of follow‐up: 1993	5514	Abstainer	69	1.00	Education, height, BMI, physical activity index, age at first pregnancy, parity, age at menarche, age at menopause, average number of cigarettes smoked, HRT
				0.1 to <5.0	110	0.80 (0.60, 1.10)
				5.0 to <15.0	55	0.70, (0.50, 1.10)
				≥15	53	0.70, (0.50, 1.10)
Mattisson et al. (2004)	Malmö and Cancer (MDC) [Sweden]	Baseline: 1991 to 1996 End of follow‐up: 2001	11,670	Abstainer	22	1.00 (0.64, 1.55)	Diet, interviewer, method version, season of diet interview, age at baseline, TE, change of dietary habits, height, waist, HRT, age at birth of first child, age at menarche, leisure time physical activity, smoking habits, educational level
				>0 to <15	257	1.10
				15 to 30	39	0.97 (0.68, 1.36)
				>30	11	1.85 (1.00, 3.43)
Allen et al. (2009)	UK Million Women Study [UK]	Baseline: 1996 to 2001 End of follow‐up: 2002 to 2006	1,280,296	Abstainer	6409	1.00 (0.97, 1.03)	Age, region, SES, BMI, smoking, physical activity, OC, HRT
				>0 to <3.4	7841	1.00 (0.98, 1.02)
				3.4 to <8.0	6642	1.08 (1.05, 1.10)
				8.0 to <17.1	5672	1.13 (1.10, 1.16)
				≥17.1	1816	1.29 (1.23, 1.35)
Land et al. (2014)	NSABP P‐1 trial [Canada, USA]	Baseline: 1992 to 1997 End of follow‐up: NS	13,388	Abstainer	–	1.00	Breast cancer risk (gail score analyzed as a continuous variable), diabetes
				>0 to 14	–	0.85 (0.59, 1.22)
				>14	–	0.92 (0.74, 1.19)
Hippisley‐Cox and Coupland (2015)	QResearch database; Egton Medical Information Systems (EMIS) [UK]	Baseline: 1998 End of follow‐up: 2013	4,943,765	Abstainer	–	1.00	Age, town, family history, ethnicity, benign breast disease, OC, HRT, manic or schizophrenia, lung cancer, blood cancer, ovarian cancer, BMI, smoking, diabetes
				>0 to <8	–	1.05 (1.03, 1.08)
				8 to <24	–	1.11 (1.07, 1.15)
				24 to <56	–	1.21 (1.16, 1.26)
				56 to 72	–	1.31 (1.07, 1.61)
				>72	–	1.25 (0.92, 1.71)
Klatsky et al. (2015)	Kaiser Permanente (KP) cohort [USA]	Baseline: 1978 to 1985 End of follow‐up: 2012	124,193	Lifetime abstainer	–	1.00	Age, sex, race or ethnicity, BMI, education, marital status, smoking
				Former drinker	–	1.30 (1.10, 1.60)
				<14	–	1.10 (1.00, 1.20)
				14 to <42	–	1.20 (1.10, 1.40)
				≥42	–	1.30 (1.10, 1.50)
Romieu et al. (2015)	European Prospective Investigation into Cancer and Nutrition (EPIC) [Greece, Spain, Italy, France, Germany, the Netherlands, United Kingdom, Denmark, Sweden, and Norway]	Baseline: 1992 to 1998, End of follow‐up: 2004 to 2010	334,850	Lifetime abstainer	753	1.00 (0.85, 1.18)	Menopausal status, OC, HRT, height, weight, menopause weight, smoking status, educational level, physical activity, age at first menses, age at first full‐term pregnancy, age at menopause, TE without alcohol intake, time since quitting alcohol
				>0 to 5 a	3951	1.10
				>5 to 15 a	3001	1.20 (1.14, 1.26)
				>15 to 30 a	1106	1.30 (1.21, 1.40)
				>30 a	214	1.19 (1.02, 1.36)
Jung et al. (2016)	β‐Carotene and Retinol Efficacy Trial [USA]; Breast Cancer Detection Demonstration Project Follow‐up Study [USA]; California Teachers Study [USA]; Canadian National Breast Screening Study [Canada]; Cancer Prevention Study II Nutrition Cohort [USA]; CLUE 2: Campaign Against Cancer and Heart Disease [USA]; Iowa Women's Health Study [USA]; Japan Public Health Center‐Based Study Cohort I [Japan]; Melbourne Collaborative Cohort Study [Australia]; Multiethnic Cohort Study [USA]; Netherlands Cohort Study [Netherlands]; New York University Women's Health Study [USA]; NIH‐AARP Diet and Health Study [USA]; Nurses' Health Study (a) [USA]; Nurses' Health Study (b) [USA]; Nurses' Health Study II [USA]; Prospective Study on Hormones, Diet and Breast Cancer [Italy]; Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial [USA]; Swedish Mammography Cohort [Sweden]; Women's Health Study [USA]; Women's Lifestyle and Health Study [Sweden]	NS	1,089,273	Abstainers	13,255	1.00	Ethnicity, education, BMI, height, physical activity, smoking status, age at menarche, joint effects of menopausal status and HRT, OC, joint effects of parity and age at first birth, family history of breast cancer, personal history of benign breast disease, TE
				>0 to <5	12,202	1.03 (1.00, 1.06)
				5 to <15	6235	1.10 (1.06, 1.14)
				15 to <30	2686	1.19 (1.14, 1.25)
				≥30	1805	1.32 (1.23, 1.41)
White et al. (2017)	Sister Study [Puerto Rico, USA]	Baseline: 2003 to 2009 End of follow‐up: 2014	50,884	Lifetime abstainer	65	1.00	Age, race/ethnicity, education, age at menarche, age at pack‐years of smoking, HRT, age at menopause and menopausal status, BMI
				Former drinker	277	1.04 (0.79, 1.73)
				<14	1219	1.06 (0.82, 1.36)
				14 to <28	170	1.10 (0.82, 1.48)
				≥28	110	1.22 (0.89, 1.68)
Ellingjord‐Dale et al. (2017)	Norwegian Breast Cancer Screening Program (NBCSP) [Norway]	Baseline: 2006 to 2014 End of follow‐up: 2014	29,162 (4402 cases; 24,760 controls)	Abstainer	725	1.00	Smoking, physical activity, height, BMI, education, age at menarche, number of pregnancies, and menopausal status, physical activity
				>0 to <24	935	1.07 (0.96, 1.19)
				24 to <36	850	1.07 (0.96, 1.20)
				36 to <60	1073	1.11 (1.00, 1.24)
				60 to <72	448	1.16 (1.02, 1.33)
				72	371	1.26 (1.09, 1.45)
Arthur et al. (2018)	Women's Health Initiative [USA]	Baseline: 1993 to 1998 End of follow‐up: 2016	131,833	Abstainer	1516	1.00 (0.94, 1.06)	Age at entry, education, non‐alcohol TE intake, ethnicity, age at menarche, parity, breastfeed, history of mammograms, HRT, OC, age at menopause, family history, history of BBD, diet, alcohol intake, physical activity, BMI, smoking
				>0 to 4.9	4021	1.01
				>4.9 to 9.9	943	1.00 (0.93, 1.08)
				>9.9 to 19.9	992	1.07 (1.00, 1.15)
				>19.9	698	1.18 (1.08, 1.28)
Betts et al. (2018)	Health and Lifestyle Survey (HALS1) [United Kingdom]	Baseline: 1984 to 1985 End of follow‐up: 2009	6721	Abstainer	90	1.00	Ethnicity, income, self‐rated health, smoking status, BMI, exercise
				>0 to <17.1	49	0.91 (0.63, 1.30)
				17.1 to 32	8	1.80 (0.89, 3.63)
				>32	2	2.22 (0.53, 9.20)
Heberg et al. (2019)	Danish Nurse Cohort Study [Denmark]	Baseline: 1993 End of follow‐up: 2016	16,106	Abstainer	245	1.00 (0.93, 1.07)	Smoking, BMI, physical activity, self‐rated health, cohabitation, parity, HRT, OC, no. of births, age at menarche
				>0 to <17.1	475	0.93
				17.1 to 24	350	1.07 (1.00, 1.14)
				>24	337	1.09 (1.02, 1.17)
Arriaga et al. (2019)	Blue Mountains Eye Study (BMES) [Australia]	Baseline: 1992 to 1993 End of follow‐up: 2012	PostM: 1891	Abstainer	–	1.00	Age
				>0 to <10	–	1.74 (0.85, 3.58)
				>10	–	0.93 (0.45, 1.91)
	Australian Longitudinal Study on Women's Health (ALSWH) [Australia]	Baseline: 1996 End of follow‐up: 2012	PreM: 21,052	Abstainer	–	1.00	Age
				>0 to <10	–	1.28 (0.90, 1.83)
				10 to 20	–	1.51 (1.01, 2.26)
				>20	–	2.21 (1.32, 3.71)
			PostM: 13,419	Abstainer	–	1.00	Age
				>0 to <10	–	0.94 (0.69, 1.28)
				>10	–	1.20 (0.93, 1.56)
	Australian Diabetes, Obesity and Lifestyle Study (AusDiab) [Australia]	Baseline: 1999 to 2000 End of follow‐up: 2012	PreM: 2518	Abstainer	–	1.00	Age
				>0 to <10	–	1.10 (0.45, 2.71)
				10 to 20	–	2.44 (1.02, 5.82)
				>20	–	0.68 (0.09, 5.20)
			PostM: 2592	Abstainer	–	1.00	Age
				>0 to <10	–	0.82 (0.43, 1.55)
				>10	–	1.15 (0.68, 1.95)
	North West Adelaide Health Study (NWAHS) [Australia]	Baseline: 1999 to 2003 End of follow‐up: 2012	PostM: 778	Abstainer	–	1.00	Age
				>0 to <10	–	0.96 (0.32, 2.86)
				>10	–	1.24 (0.42, 3.69)
	Sax Institute's 45 and Up Study [Australia]	Baseline: 2006 to 2009 End of follow‐up: 2012	PreM: 15,016	Abstainer	–	1.00	Age
				>0 to <10	–	1.12 (0.81, 1.54)
				10 to 20	–	1.32 (0.93, 1.88)
				>20	–	1.45 (0.90, 2.35)
			PostM: 103,074	Abstainer	–	1.00	Age
				>0 to <10	–	1.08 (0.96, 1.21)
				>10	–	1.19 (1.06, 1.34)
Viner et al. (2019)	Alberta's Tomorrow Project (ATP) [Canada]	Baseline: 2001 to 2008 End of follow‐up: 2017	PreM: 8447	Abstainer	24	1.00	Age, marital status, highest level of education, smoking status, PYs of cigarettes, BMI, menopausal status, history of breast cancer screening
				0 to 13.45	140	1.13 (0.73, 1.75)
				>13.45	23	1.01 (0.56, 1.81)
			PostM: 8133	Abstainer	62	1.00
				0 to 13.45	185	0.88 (0.66, 1.18)
				>13.45	31	0.90 (0.57, 1.40)
Zeinomar et al. (2019)	Prospective Family Study Cohort [Australia, Canada, USA]	NS	17,435	Abstainer	492	1.00	Study site, race/ethnicity, BMI, education, OC, smoking
				<13.45	275	0.99 (0.85, 1.16)
				≥13.45	193	1.10 (0.92, 1.32)
Arthur et al. (2020)	UK Biobank [UK]	Baseline: 2006 to 2010 End of follow‐up: 2015 to 2016	PreM: 35,457	Abstainer	–	1.00	Age at recruitment, socioeconomic status, age at menarche, parity and age at first pregnancy, family history of breast cancer, history of mammograms, OC, age at menopause, HRT, first five principal components of ancestry and genotyping batch, genetic score, diet score, alcohol consumption, physical activity, BMI, smoking
				>0 to 14	–	1.06 (0.77, 1.39)
				>14	–	1.11 (0.82, 1.51)
			PostM: 110,869	Abstainer	–	1.00
				>0 to 14	–	1.08 (0.94, 1.23)
				>14	–	1.15 (0.99, 1.33)
Rainey et al. (2020)	“KARolinska MAmmography project for risk reduction of breast cancer” (KARMA) [Sweden]	Baseline: 2011 to 2013 End of follow‐up: 2017	57,654	Abstainer	159	1.00	Age, BMI, ethnicity, OC, HRT, family history of breast cancer in mother or sister, age at menarche, parity and age at first birth, education level, menopausal status
				>0 to <10	579	1.17 (0.96, 1.41)
				>10	220	1.26 (1.01, 1.59)
Li et al. (2020)	International BRCA1/2 Carrier Cohort Study (IBCCS) [Austria, Belgium, Denmark, France, Germany, Hungary, Iceland, Italy, the Netherlands, Spain, Sweden, UK], Kathleen Cuningham Foundation Consortium for Research Into Familial Breast Cancer (kConFab) Follow‐Up Study [Australia, New Zealand], Breast Cancer Family Registry (BCFR) [Australia, Canada, USA]	NS	BRCA1 carriers:	Abstainer	–	1.00	Tobacco consumption, bilateral Oophorectomy, age at first full‐term pregnancy, number of full‐term pregnancies, BMI, OC, number of affected relatives with breast cancer, age at menarche
				>0 to <12	–	1.12 (0.81,1.55)
				12 to 24	–	1.07 (0.75,1.54)
				>24	–	1.06 (0.55,2.05)
			BRCA2 carriers:	Abstainer	–	1.00
				>0 to <12	–	0.99 (0.65,1.52)
				12 to 24	–	1.28 (0.81,2.04)
				>24	–	0.91 (0.37,2.19)
Iwase et al. (2021)	JPHC‐I; JPHC‐II; JACC; MIYAGI‐I; MIYAGI‐I; MIYAGI‐II; AICHI; OHSAKI; LSS [Japan]	Baseline: 1984 to 1995 End of follow‐up 2000 to 2013	131,686	Abstainer	1638	1.00	Age, area, menopausal status at baseline, smoking status, BMI, age at menarche, and parity
				>0 to <11.5	220	1.08 (0.92, 1.27)
				11.5 to <23	51	1.09 (0.82, 1.45)
				≥23	66	1.36 (0.93, 1.99)
Lee et al. (2023)	Korean Genome and Epidemiology Study (KoGES) [South Korea]	Baseline: 2004 to 2013 End of follow‐up: 2018	68,130	Abstainer	33	1.00	Education, smoking status, TE, family history of cancer
				>0 to 14	239	0.83 (0.57, 1.19)
				>14	590	0.81 (0.57, 1.16)

Abbreviations: BBD, benign breast disease; BMI, body mass index; HRT, hormone replacement therapy; OC, oral contraceptives; PreM, premenopausal; PostM, postmenopausal; PYs, pack years; SES, socio‐economic status; TE, total energy.

^a Average lifetime alcohol intake.

FIGURE 2.

Scatter plot of studies that investigated the relationship between alcohol consumption and breast cancer incidence.

Relative risks for standard drink per day, categories of consumption, and HED

The meta‐regression estimates of overall breast cancer indicated a nonlinear association with alcohol consumption (see Table S6); RRs were 1.05 (95% CI: 1.04, 1.06), 1.10 (95% CI: 1.08, 1.12), 1.18 (95% CI: 1.15, 1.21), and 1.22 (95% CI: 1.19, 1.25) for 0.5, 1, 2, and 3 standard drinks per day, respectively. Two pooled analyses and five individual studies reported RR estimates for daily consumption of more than three drinks per day (i.e., more than 30 g/day). Among those seven studies, the meta‐regression estimated RRs for 4, 5, and 6 drinks per day of 1.24 (95% CI: 1.20, 1.28), 1.26 (95% CI: 1.21, 1.31), and 1.28 (95% CI: 1.21, 1.35), respectively. The sensitivity analysis revealed that the RR increase per gram of alcohol significantly changed at 20 g/day. Increases in alcohol consumption above this threshold were significantly associated with breast cancer incidence (RR: 1.02 [95% CI: 1.00, 1.03] per 10‐g increase in alcohol consumption). However, this association was weaker compared to the increases in alcohol consumption from 0 to 20 g/day (RR: 1.09 [95% CI: 1.08, 1.11] per 10‐g increase).

The association between alcohol consumption of less than one standard drink per day (<10 g/day) and breast cancer incidence was assessed in nine studies; the meta‐RR was 1.04 (95% CI: 1.01, 1.07) (see Figure 3 and Figure S1). A sensitivity analysis that used a meta‐regression to control for the correlation between multiple RR estimates from the same study and different categories of alcohol consumption also found a significant association (RR: 1.04 [95% CI: 1.02, 1.05]). A second sensitivity analysis, which used a meta‐regression to control for multiple RR estimates from the same study and differences in alcohol consumption categorizations (among studies examining the association between alcohol consumption of less than one standard drink per day and breast cancer incidence), also found a significant association (RR: 1.06 [95% CI: 1.04, 1.08] per 5 g/day increase).

FIGURE 3.

Forest plot of studies which investigated the relationship between consuming less than 10 g of ethanol per day and breast cancer incidence.

A total of nine studies reported the risk of breast cancer among former drinkers (compared to lifetime abstainers) (see Tables S7 and S8 and Figures S2 and S3). A meta‐analysis of RR estimates for former drinkers (compared to lifetime abstainers) showed a nonsignificant increase in breast cancer risk for former drinkers compared to lifetime abstainers (RR: 1.11 (95% CI: 0.99, 1.25)).

The association between HED and breast cancer risk was assessed in four studies (see Tables S9 and S10). Two studies observed that engaging in HED in the past year increased breast cancer risk (compared to not engaging in HED) (Sánchez‐Bayona et al., 2020; White et al., 2017). When combined through meta‐regression, these two studies indicated that engaging in HED in the past year increased the risk of breast cancer by 1.40 (95% CI: 1.22, 1.63). Two studies observed a positive trend with higher numbers of drinks consumed on a drinking occasion in relation to breast cancer risk (Chen et al., 2011; Mørch et al., 2007). Furthermore, one study observed that the frequency of HED (in the past year and lifetime) was also associated with breast cancer risk; however, there was not a consistent increasing RR with increasing frequency of HED (White et al., 2017). Two studies that examined the association between HED and breast cancer risk adjusted for the average volume of alcohol consumption. In these studies, a significant increase in breast cancer risk was observed among individuals who engaged in HED in the past year (Sánchez‐Bayona et al., 2020). Additionally, a significant association was found between the largest number of alcoholic drinks consumed in 1 day in a typical month and breast cancer risk (Chen et al., 2011).

Relative risks according to age at initiation of drinking

The association between age at initiation of drinking and breast cancer risk was assessed in six studies (see Tables S11 and S12). No significant association was observed between age at initiation and breast cancer risk in all six studies. A meta‐regression of five studies that used never drinkers/nonregular drinkers as a reference found that age of initiation was not statistically associated with breast cancer risk (see Table 2).

TABLE 2.

Results of meta‐analyses and meta‐regressions: former drinking, heavy episodic drinking, age of alcohol consumption initiation, alcohol consumption before first pregnancy and their association with breast cancer risk, as well as alcohol consumption and breast cancer risk by menopausal status, hormone receptor status, and luminal type.

Analysis	Number of publications, Included in the meta‐analysis	Relative risk (per 10 g/day increase)
		Point estimate	Lower 95% CI	Upper 95% CI	p‐value
Former drinkers (compared to lifetime abstainers)	9	1.11	0.99	1.25	0.088
Heavy episodic drinking (past year)
No	2	REF	–	–	–
Yes	2	1.41	1.22	1.63	<0.001
Age of initiation (alcohol consumption)
18 years of age (compared to never drinkers)	5	1.07	0.81	1.35	0.195
Per year increase (in age of initiation among drinkers)	5	0.99	0.96	1.02	0.438
Alcohol consumption before first pregnancy (10 g/day increase)	2	0.99	0.97	1.02	0.531
Menopausal status
Premenopausal (10 g/day increase in alcohol consumption)	11	1.04	1.01	1.07	0.002
Postmenopausal (10 g/day increase in alcohol consumption)	8	1.12	1.1	1.14	<0.001
Hormone receptor status
ER+ (10 g/day increase in alcohol consumption)	6	1.08	1.07	1.1	<0.001
ER− (10 g/day increase in alcohol consumption)	6	1.03	0.99	1.07	0.112
PR+ (10 g/day increase in alcohol consumption)	4	1.08	1.06	1.11	<0.001
PR− (10 g/day increase in alcohol consumption)	4	1.06	1.03	1.09	<0.001
Luminal type
A (HR+/HER2−/low ki67) (10 g/day increase in alcohol consumption)	3	1.12	1.02	1.23	0.016
B (HR+/HER2+ or HR+/HER2−/high ki67) (10 g/day increase in alcohol consumption)	3	1.10	0.98	1.25	0.116
HER2 type (ER−/PR−/HER2+) (10 g/day increase in alcohol consumption)	3	1.02	0.56	1.83	0.956
Triple negative (ER−/PR−/HER2−) (10 g/day increase in alcohol consumption)	4	1.02	0.88	1.19	0.761

Abbreviations: ER, Estrogen receptor; HER2, human epidermal growth factor receptor 2; HR, Hormone receptor; PR, Progesterone receptor.

Three studies assessed alcohol consumption between menarche and first birth (see Tables S13 and S14). Liu et al. found that the average volume of alcohol consumed between menarche and first birth was statistically significantly associated with higher breast cancer risk. In contrast, Jayasekara et al., who assessed the average volume of alcohol consumed per day between menarche and first birth, observed both positive and negative associations between alcohol consumption between menarche and first birth (measured in grams per day) and breast cancer risk (Jayasekara et al., 2016; Tjønneland et al., 2004). Tjønneland et al., who assessed age of drinking initiation (before or after first birth), observed a negative but nonsignificant association between consuming alcohol between menarche and first birth and breast cancer risk (Jayasekara et al., 2016; Tjønneland et al., 2004). A meta‐analysis of two studies found that alcohol consumption between menarche and first birth was not statistically associated with breast cancer risk (see Table 2) (Jayasekara et al., 2016; Tjønneland et al., 2004).

Relative risks stratified by menopausal status

The associations of alcohol consumption with premenopausal and postmenopausal breast cancer incidence were assessed in 8 and 11 studies, respectively (see Table 2 and Tables S15 and S16). The meta‐RRs of premenopausal and postmenopausal breast cancer for each additional standard drink per day were 1.04 (95% CI: 1.01, 1.07) and 1.12 (95% CI: 1.10, 1.14), respectively. One study examined the associations of alcohol consumption with perimenopausal breast cancer. No significant association was observed between perimenopausal breast cancer and alcohol consumption.

Relative risks stratified by receptor status and luminal type

A total of six, four, and one study examined the associations of alcohol consumption with breast cancer risk stratified by estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status, respectively (see Tables S17 and S18). The meta‐RRs for each additional standard drink per day were 1.08 (95% CI: 1.07, 1.10) for ER+, 1.03 (95% CI: 0.99, 1.07) for ER−, 1.08 (95% CI: 1.06, 1.11) for PR+, and 1.06 (95% CI: 1.03, 1.09) for PR− breast cancers. The one study that examined HER2 status found that alcohol consumption was associated with HER2− breast cancer (RR: 1.07 (95% CI: 1.03, 1.12) for each additional standard drink), but not HER2+ cancers (RR: 1.02 (95% CI: 0.91, 1.14) for each additional standard drink).

A total of two studies examined the association between alcohol consumption and risk of luminal A (HR+/HER2−/low ki67), B (HR+/HER2+ or −/high ki67) and HER2 enriched‐(HR+ or −/HER2+) breast cancers (see Tables S18 and S19). The meta‐regression of results from these studies indicated that alcohol consumption was significantly associated with an increased risk of luminal A breast cancer (RR: 1.12 (95% CI: 1.02, 1.23) for each additional standard drink). Similar results were observed for alcohol consumption and risk of luminal B breast cancer (RR: 1.10 (95% CI: 0.98, 1.25) for each additional standard drink) but not with HER2‐enriched breast cancer (RR: 1.02 (95% CI: 0.56, 1.83) for each additional standard drink).

A total of three studies examined the association between alcohol consumption and risk of triple negative (ER−/PR−/HER2−) breast cancer. A meta‐regression of results of these studies indicated that alcohol consumption was not significantly associated with triple negative breast cancer (RR: 1.02 (95% CI: 0.88, 1.19) for each additional standard drink) (see Table 2).

The percentage of cases ascertained for ER/PR/HER/luminal status varied by study. The percentage of cases which were not ascertained for ER status ranged from 12.2% to 56.9%, and the percentage of cases that were not ascertained for PR status ranged from 12.2% to 55.7%. The percentage of cases that were not ascertained for HER status in the study by Romieu et al. was 79.5%. The percentage of cases that were not ascertained for luminal status ranged from 3.2% to 10.0% (see Tables S19 and S20).

DISCUSSION

This review updates previous reviews on the association between alcohol consumption and breast cancer incidence. Similar to the pooled analyses by Jung et al., the present review found that the relationship between alcohol consumption and breast cancer risk is nonlinear (Jung et al., 2016). Based on the fact that the association between alcohol consumption and breast cancer has been known for decades, in this review we explore this association in more detail using the available evidence published up to November 15, 2023.

This meta‐analysis corroborates previous evidence that consumption of 1 drink per day is associated with a higher risk of breast cancer (Jung et al., WCRF). The conclusion that low amounts of alcohol consumption are associated with a higher risk of breast cancer also aligns with previous meta‐analyses. Choi et al. reported a meta‐RR of 1.09 (95% CI: 1.06, 1.12) for consuming <12 g/day (Choi et al., 2018; Seitz et al., 2012), and Seitz et al. reported a RR of 1.05 (95% CI: 1.02, 1.09) for consuming <12.5 g/day. However, the meta‐analysis by Choi et al. included retrospective cohort studies, and the meta‐analysis by Seitz et al. included both breast cancer incidence and mortality as outcomes. The observation that the impact of each additional gram of alcohol consumed on breast cancer diminishes after low volumes (20 g/day) has been previously observed (Jung et al., 2016); however, the potential biological mechanism for this observation is currently unknown.

A younger age of initiation may lead to a longer exposure period and to an increased exposure between menarche and first birth. An association between alcohol consumption between menarche and first birth and breast cancer risk is hypothesized to be due to a greater susceptibility to the accumulation of molecular damage that can lead to breast cancer. In particular, the rapid cellular proliferation of breast tissue that occurs during puberty leads to an increase in susceptibility to carcinogenesis. Furthermore, it is hypothesized that breast epithelial tissue, which differentiates during pregnancy and lactation, is more susceptible to genotoxic exposures. This increased susceptibility is because the terminal ductal–lobular units of the breast do not fully differentiate until the end of gestation (Russo et al., 2005). Pregnancy leads to a decrease in the number of hormone‐sensitive luminal cells, down‐regulation of the Wnt signaling pathway in basal stem/progenitor cells, and reduction of prolactin and estrogen concentrations; this series of events decreases a woman's susceptibility to carcinogenesis (Bernstein et al., 1985; Colditz & Frazier, 1995; Medina, 2013; Musey et al., 1987). Furthermore, studies of mice have shown that dietary exposure to ethanol during puberty leads to changes in the mammary glands (e.g., ductal branching and epithelial proliferation and density) (Masso‐Welch et al., 2012; Singletary, 1997). No study included in this review controlled for age of smoking initiation.

The impact of age at initiating alcohol consumption on breast cancer risk was found to be nonsignificant in all studies. Mixed findings were found with regards to the impact of alcohol consumption between menarche and first birth on breast cancer risk. As smoking initiation before first full‐term pregnancy has been shown to increase breast cancer risk, this lack of control for smoking initiation may confound the relationship between age of initiation and alcohol consumption between menarche and first birth and breast cancer risk (Bjerkaas et al., 2013). The inconsistencies, nonsignificant findings, and methodological limitations in prospective cohort studies highlight that more research is needed to comprehensively assess these risk relationships.

This review strengthens the evidence that alcohol consumption increases the risk of premenopausal breast cancer. The WCRF CUP systematic literature review for breast cancer (World Cancer Research Fund/American Institute for Cancer Research, 2018b) reported a RR of 1.03 (95% CI: 0.99, 1.07) per 10 g increase of alcohol consumed based on an analysis of data from the Pooling Project (Jung et al., 2016) combined with nonoverlapping studies in the CUP. Our findings, which include five additional analyses published after the WCRF CUP systematic literature review was performed (Arriaga et al., 2019; Arthur et al., 2020; Iwase et al., 2021; Rainey et al., 2020; Viner et al., 2019), were similar but with greater statistical precision (RR, 1.03 (95% CI: 1.01, 1.06)). One study was found on the RR of alcohol consumption and perimenopausal breast cancer. This RR was nonsignificant; however, as only one study was found on this topic, further research is needed.

Based on the results of four studies, HED is positively associated with breast cancer risk. Therefore, there may be an effect of drinking pattern on cancer risk, where acute, heavy alcohol exposure impacts breast cancer risk differently than the same amount of alcohol consumed over a greater number of occasions. There are multiple biological rationales for this difference. The enzymatic pathways responsible for the elimination of ethanol and its metabolite acetaldehyde may be saturated during HED, leading to an increase in carcinogenicity (Seitz & Maurer, 2007). Furthermore, binge or heavy drinking behaviors may trigger other biological mechanisms, including increased oxidative stress, inflammation, and insulin resistance, all of which are known pathways in the development of breast cancer (Lindtner et al., 2013; Pichard et al., 2008; Rojdmark et al., 2000; Ward et al., 2009).

ER, PR, and HER statuses, and luminal subtype are diagnostic and prognostic biomarkers of breast cancer (Nicolini et al., 2018). These markers also help in the identification of the most appropriate adjuvant systemic therapy (Nicolini et al., 2018). Consumption was associated with a higher risk of ER+, PR+, PR−, HER2−, and luminal A breast cancers and less strongly with luminal B breast cancer, and no association was observed with triple‐negative breast cancer. There were fewer cases of luminal B, HER2 type, and triple negative breast cancers compared to luminal A breast cancers, which may have impacted the power to detect an association between alcohol consumption and luminal B, HER2 type, and triple negative breast cancers. A limitation of these findings is the lack of data on breast cancers by ER, PR, and HER statuses and luminal subtype in some studies. The missingness of data on ER, PR, and HER statuses and luminal status may lead to selection bias and affect the results of these analyses.

The finding that alcohol consumption is more strongly associated with ER+ than with ER− breast cancer may be due to the mitogenic activity of ethanol in ER+ human breast cancer cells only and to the fact that alcohol stimulates estrogen receptor signaling (Fan et al., 2000; Singletary et al., 2001). The same biological pathway may explain the association between alcohol consumption and luminal A breast cancers, which are HR+. The findings from individual and a pooled analysis of cohort studies were inconsistent with regards to the association between alcohol consumption and ER− breast cancer risk. The consistency of these findings has implications for the carcinogenic mechanism of how alcohol consumption causes breast cancer. The significant finding for ER+ breast cancers and the positive but insignificant finding for ER− breast cancer may indicate combined effects of alcohol consumption promoting the development of breast cancer through modifying hormone levels and alcohol consumption leading to breast cancer initiation. The finding that alcohol is linked to ER+ but not ER− breast cancers also may have implications for the advice given to people with ER+ breast cancer. In a study, ethanol was found to enhance cell proliferation of ER+ human breast cancer cell cultures (as compared to ER− human breast cancer cell cultures) (Singletary et al., 2001). Accordingly, women with ER+ breast cancers or in remission from ER+ breast cancer may benefit from abstaining from alcohol consumption or reducing their alcohol consumption.

Limitations

The systematic review is limited by the search engines utilized. PubMed and Embase may not have indexed all relevant English language studies. Furthermore, the China National Knowledge Infrastructure (CNKI) indexes articles in Chinese that are not included in PubMed and Embase (Cohen et al., 2015); therefore, this systematic review may have excluded relevant studies from China. Additionally, the exclusion of theses may have led to relevant articles not being captured in the systematic search.

The measurement of alcohol consumption is limited by multiple biases in cohort studies due to both question design and reporting biases. Respondents may underreport alcohol consumption due to recall bias and social desirability bias (Stockwell et al., 2018). Measures of quantity and frequency also may underrepresent alcohol consumption, as infrequent HED occasions are not taken into consideration when participants are asked to report the usual volume of alcohol they consume. While the underreporting of alcohol consumption is nondifferential as everyone reports consumption before diagnosis in cohort studies, the RR estimates for breast cancer may overestimate the risk for given alcohol consumption amounts as such amounts may be underestimated.

There are limitations in the measurement of the biologically relevant exposure period for the risk relationship between alcohol consumption and breast cancer incidence. Because the mechanisms underlying the alcohol‐breast cancer association are not entirely clear, the latency period for alcohol and breast cancer is not known. Based on current plausible mechanisms for carcinogenesis, alcohol consumption may increase cancer risk in the short term and long term. Specifically, if alcohol consumption increases the risk of breast cancer initiation through mechanisms including DNA damage, then long‐term alcohol consumption affects breast cancer risk. Furthermore, if sex hormones play a role, alcohol consumption promotes the growth of some breast cancers, and recent alcohol consumption affects breast cancer risk. A total of 19 analyses did not separate lifetime abstention from cessation when assessing whether alcohol consumption increases the risk of breast cancer. A total of five studies assessed former drinking and lifetime abstention separately. Of these, only one study assessed lifetime alcohol consumption. Individuals who formerly drank may be at an increased risk of breast cancer when compared to lifetime abstainers, especially among individuals who previously consumed large amounts of alcohol, but have reduced their alcohol consumption or have quit due to adverse health diagnoses and/or health complications associated with alcohol consumption (i.e., sick quitters) (Sarich et al., 2019). As a result, the RR of breast cancer reported by studies that used current abstention as the reference group may underestimate the association between alcohol consumption and the risk of breast cancer when compared to the results reported by studies that used lifetime abstention as a reference group (see: Sarich et al., 2019).

Some meta‐analyses were impaired by heterogeneities in the alcohol exposure measurement, and, therefore, some meta‐analyses did not include all cohort studies that measured the association between alcohol consumption and breast cancer risk. Furthermore, there was heterogeneity in the manner in which RR estimates were controlled for across studies. With the exception of Arriaga et al., which only adjusted for age (Arriaga et al., 2019), most studies adjusted for age, smoking history, body mass index, oral contraceptive use, and use of hormone replacement therapy, and some studies also adjusted for other risk factors such as family history of breast cancer, history of breast cancer screening, parity, breastfeeding, diet (studies controlled for variables such as total dietary score, scored in accordance with dietary guidelines such as the WCRF guidelines for the intake of key food groups, fat intake, and total energy intake), and physical activity, all of which have been shown to be associated with breast cancer risk. The effect of the lack of adjustment for confounding variables on the risk relationship between alcohol consumption and breast cancer incidence is currently unclear, and further investigations are needed to clarify how the adjustment, or lack of adjustment, for confounders like smoking impacts the risk relationship between alcohol consumption and breast cancer (Chu & Wallach, 2020). A list of the range of confounding factors adjusted for across all studies can be found in the Supplement (Table S21).

The current review examined the significance of the risk relationship between alcohol consumption and breast cancer incidence by menopausal status, hormone receptor status, human epidermal growth factor receptor 2 status, and molecular subtype. The review did not examine if the risk relationship between alcohol consumption and breast cancer incidence differed significantly between these factors.

Public health implications

Alcohol consumption increases breast cancer risk, even for women consuming less than one drink per day and at all periods of life (i.e., both pre‐ and post‐menopause). Accordingly, to help prevent breast cancer, women can avoid consuming alcohol or reduce the amount of alcohol they consume. One of the major barriers to women drinking less is their limited awareness of the relationship between alcohol consumption and the risk of developing breast cancer. Indeed, only 10% to 20% of the European Union's general population is aware of the link between alcohol and breast cancer (Kokole et al., 2023); only 11% to 17% of people in the United Kingdom are aware of this relationship (Agabio et al., 2022); and 25% of women 15 to 44 years of age in the United States are aware of this relationship (Khushalani et al., 2020).

To improve awareness of this relationship, cancer‐specific health warnings on labels (and advertisements in countries where ads are allowed) to inform consumers about the risks of alcohol consumption (including, but not limited to, breast cancer risks) may be an effective policy. Indeed, studies have shown that labels on alcoholic beverages informing consumers of the link between alcohol consumption and cancer risk can increase knowledge about this association (Correia et al., 2024; Weerasinghe et al., 2020). Other methods of health promotion campaigns, including mass media campaigns such as the European Code Against Cancer, which recommends “If you drink alcohol of any type, limit your intake. Not drinking alcohol is better for cancer prevention,” may also be effective ways of increasing the proportion of women who are aware that alcohol consumption can cause cancer (International Agency for Research on Cancer, 2024; Young et al., 2018). Targeted health promotion campaigns, such as those taking place at school and workplaces, may also increase knowledge of alcohol's impact on cancer risk (Agabio et al., 2015; Ames & Bennett, 2011). Furthermore, routine assessments of alcohol consumption by health care professionals and integration of evidence on the association between alcohol and breast cancer into screening and brief intervention guidelines provide an opportunity to discuss the increased risk of breast cancer at even low levels of alcohol consumption and how to reduce that risk by reducing alcohol consumption (McKnight‐Eily et al., 2017).

The implementation of alcohol policies that aim to reduce the burden of breast cancers attributable to alcohol consumption should take into consideration the biologically relevant exposure period between alcohol consumption and the incidence of breast cancer for public health planning purposes and to manage expectations of policy makers before and after policy changes are implemented. Based on the plausible biological mechanisms of how alcohol consumption affects breast cancer risk, the biologically relevant exposure period between alcohol consumption and breast cancer incidence may translate into a delay in terms of a full return on investment of alcohol policies aimed at addressing breast cancer incidence (i.e., differing alcohol policies may take differing lengths of time to affect breast cancer rates) (Michels et al., 2020; Thygesen et al., 2008).

CONCLUSIONS

This systematic review is consistent with previous studies, showing that consuming less than one standard drink per day is associated with a higher risk of breast cancer compared with no intake. In addition, this systematic review provides further evidence that alcohol consumption is associated with a higher risk of both premenopausal and postmenopausal breast cancer. To inform the public about the deleterious health effects of alcohol consumption, compulsory warning labels and public health information campaigns should be implemented, as well as increased screening for alcohol‐related disorders and integrating evidence on the association between alcohol consumption and breast cancer into brief intervention guidelines.

FUNDING INFORMATION

The research leading to these results or outcomes received funding from EU4Health under contribution agreement SANTE/2022/SI2.883729 (addressing alcohol harm: capacity building, raising awareness, and implementation of best practices in the Union). The contents of this manuscript are the sole responsibility of the authors and do not necessarily reflect the views of the U.

CONFLICT OF INTEREST STATEMENT

GS, DC, CF‐B, GG, and MN are staff members of the World Health Organization. BL‐S, PF, HR, and Isabelle S are staff members of the International Agency For Research on Cancer. DK, KS, and JR are World Health Organization consultants. Where authors are identified as personnel of the International Agency for Research on Cancer/World Health Organization, the authors alone are responsible for the views expressed in this article, and they do not necessarily represent the decisions, policies, or views of the International Agency for Research on Cancer/World Health Organization.

Supporting information

Data S1.

Sohi, I. , Rehm, J. , Saab, M. , Virmani, L. , Franklin, A. , Sánchez, G. et al. (2024) Alcoholic beverage consumption and female breast cancer risk: A systematic review and meta‐analysis of prospective cohort studies. Alcohol: Clinical and Experimental Research, 48, 2222–2241. Available from: 10.1111/acer.15493

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.