Introduction

Alcohol consumption is internationally recognized as one of the compelling risk factors for breast cancer. Ethanol contained in alcoholic beverages and its metabolite acetaldehyde have been experimentally shown to be carcinogenic [1]. However, the increased risk of breast cancer due to alcohol consumption does not necessarily correlate with the prognosis of breast cancer patients. Even in breast cancer patients, there are many occasions where alcoholic beverages are preferred in daily life or social gatherings. It is a modifiable factor through individual effort, and accurate information is required.

In June 2022, the Japanese Breast Cancer Society practice guidelines, the 2022 edition, was published by the Japanese Breast Cancer Society. In the guidelines, we addressed the clinical question, “Is alcohol consumption associated with prognosis in breast cancer patients?” To make recommendations for this clinical question, we assessed the benefits and harms of alcohol consumption after patients were diagnosed with breast cancer by performing a systematic review and meta-analysis of the relevant literature. The results of this systematic review and meta-analysis are reported.

Methods

A quantitative and qualitative systematic review was performed according to the Minds Handbook for Clinical Practice Guideline Development Ver. 3.0, 2020 [2]. Regarding alcohol consumption in breast cancer patients, the importance of outcomes was examined based on the Minds Handbook. Breast cancer recurrence, breast cancer mortality, and all-cause mortality were judged to be important as harms, and the cardiovascular disease reduction effect was judged to be important as a benefit.

Search strategy and selection criteria

The PubMed, Cochrane Library, and Ichushi-Web databases were searched using the terms “breast neoplasms”, “breast cancer”, “alcohol drinking”, “alcohol consumption”, “prognosis”, “recurrence”, “survival”, and “mortality”. The search was limited to articles published up to March 2021.

According to these criteria, two authors (TN and YK) independently reviewed all titles and abstracts of the retrieved articles. The full articles were evaluated for relevance. Systematic reviews were eligible when they were judged to be of moderate quality according to AMSTAR [3] and included studies fulfilling the inclusion criteria. Authors were allowed to independently search the articles and add them to the list when they met all inclusion criteria (hand-searching). Reviews, articles on breast cancer incidence, very low population numbers, and articles with missing outcome data were excluded.

Quality assessment and data extraction

The qualities of the trials selected through the process described above were evaluated according to the Minds Handbook for Clinical Practice Guideline Development Ver. 3.0, 2020 [2].

Statistical analysis

Review Manager software 5.4 (https://review-manager.software.informer.com/5.4/) was used for statistical analysis. The summary of relative risk (RR) (hazard ratio or risk ratio) and 95% confidence intervals (95% CI) for the highest and lowest categories of alcohol intake were estimated using the inverse variance method.

RR < 1 favored alcohol intake. The heterogeneity of the trials’ results was assessed by inspecting graphical presentations; the Chi-squared test was used to evaluate heterogeneity, and I2 statistics were used to evaluate inconsistency. Significant heterogeneity was defined as I2 statistic > 50%. All analyses were performed using a random effects model. A funnel plot estimating the precision of a trial was examined for asymmetry to estimate publication bias. The outcomes of this meta-analysis were breast cancer recurrence, breast cancer-specific death, all-cause and cardiovascular disease mortality.

Results

Literature search results

The database search identified 725 potentially relevant articles, and hand-searching was used to identify six studies. In the 1st screening, 668 studies were excluded, leaving 63 systematic review articles. In the 2nd screening, 31 papers met the criteria and were judged eligible (Fig. 1). Alcohol consumption and breast cancer recurrence, breast cancer-related morality, all-cause mortality in breast cancer patients and cardiovascular disease mortality in breast cancer patients or healthy participants were assessed.

Fig. 1
figure 1

Results of the search

Full size image

Overview of included studies

Table 1 presents an overview of the characteristics of the included studies of pre- and post-diagnosis alcohol consumption and/or breast cancer recurrence, breast cancer-related mortality, all-cause mortality, and cardiovascular disease mortality. All studies meeting the inclusion criteria were observational, with 32 being cohort studies and 1 a case–control study. Data collection ranged between 1982 and 2013. Effect measures used were the hazard ratio or the risk ratio. The studies included in this systematic review presented data from 4638 cases of recurrence, 12,209 cases of breast cancer-related mortality, 21,945 cases of all-cause mortality, and 7727 cases of cardiovascular disease mortality. The participants were women diagnosed with in situ or invasive breast cancer (TNM Stage 0-IC). Median or mean follow-up was 31.56 months to 11.3 years. Exposure times were pre-diagnosis 24, post-diagnosis 7, and pre- and post-diagnosis 2. Alcohol exposure varied from over 4 g/day to over 30 g/day.

Table 1 Characteristics of included studies of pre- and post-diagnosis alcohol consumption and/or breast cancer recurrence, mortality (breast cancer, all-cause, cardiovascular disease)
Full size table

Evaluation of the risk of bias

The Cochrane risk-of-bias assessment was performed to evaluate the quality of the 33 included studies (Table 2). Non-exposed cohort bias was present in all studies because these studies were observational and not randomly assigned. On the other hand, there was a low risk of allocation concealment, blinding of participants and personnel, and selective reporting. Incomplete outcome data were rated as unclear risk of bias in 6 cases with an observation period of less than 5 years and as high risk of bias in 1 case with an observation period of less than 3 years. Regarding the adjustment of confounding factors item, there were 7 studies in which adjustment for confounding factors was hardly performed, and 5 studies in which it was evaluated as inadequate.

Table 2 Risk of bias summary
Full size table

Association between alcohol consumption and breast cancer recurrence

There were 7 pre-diagnosis studies [4,5,6,7,8,9,10] and 3 post-diagnosis studies [11,12,13] of alcohol consumption and breast cancer recurrence in breast cancer patients. The risk of recurrence increased in two pre-diagnosis studies [4, 9] and in one post-diagnosis study limited to postmenopausal cases [12], but the risk was not significant in other studies.

In the meta-analysis of seven studies, the RR for pre-diagnosis alcohol intake was 1.02 (95% CI 0.77–1.37, p = 0.88), showing high heterogeneity in the effect (I2 = 69%), and no significant risk was observed. The RR for the post-diagnosis alcohol intake was 0.96 (95% CI 0.85–1.10, p = 0.57), showing no heterogeneity in the effect (I2 = 0%), and no significant risk was observed (Fig. 2A, B).

Fig. 2
figure 2

Forest plots of the hazard ratio for alcohol intake and breast cancer recurrence. A Pre-diagnosis; B post-diagnosis

Full size image

Association between alcohol consumption and breast cancer-related mortality

There were 21 pre-diagnosis studies [4, 8, 9, 14,15,16,17,18,19,20,21,22,23,24,25,26,27,28] and 9 post-diagnosis studies [12, 14, 16, 19, 29,30,31,32] of alcohol consumption and breast cancer-related mortality in breast cancer patients. For pre-diagnosis alcohol intake, increased risk was observed in four studies [4, 8, 23, 24] and decreased risk in two studies [26, 27], but the risk was not significant in other studies. For post-diagnosis alcohol intake, increased risk was observed in 1 study [30] and 1 study [12] when limited to postmenopausal cases.

In the meta-analysis of 21 studies, the RR for pre-diagnosis alcohol intake was 1.02 (95% CI 0.93–1.11, p = 0.69), showing high heterogeneity in the effect (I2 = 60%), but not significant. In nine studies, the RR for post-diagnosis alcohol intake was 0.96 (95% CI 0.77–1.19, p = 0.70), showing slight heterogeneity in the effect (I2 = 38%), but not significant (Fig. 3A, B).

Fig. 3
figure 3

Forest plots of the hazard ratio for alcohol intake and breast cancer-related morality. A Pre-diagnosis; B post-diagnosis; C ER-positive cases; D ER-negative cases

Full size image

In addition, five studies of estrogen receptor (ER) expression status were reported in relation to breast cancer-related death [14, 19, 27], three studies of pre-diagnosis alcohol intake [14, 19] and two studies of post-diagnosis alcohol intake [19, 27]. In the meta-analysis, the RR for the ER-positive cases was 0.77 (95% CI 0.59–1.00, p = 0.05), showing no heterogeneity in the effect (I2 = 0%), and the risk was slightly reduced (Fig. 3C). The RR for ER-negative cases was 0.79 (95% CI 0.56–1.11, p = 0.17), showing slight heterogeneity (I2 = 39%), but not significant (Fig. 3D).

Association between alcohol consumption and all-cause mortality

There were 18 pre-diagnosis studies [5, 8, 11, 14,15,16, 18,19,20,21, 26,27,28, 33, 34] and 8 post-diagnosis studies [12, 13, 16, 19, 29, 30] of alcohol consumption and all-cause mortality in breast cancer patients. For pre-diagnosis alcohol intake, decreased risk was observed in three studies [11, 19, 33]. For post-diagnosis alcohol intake, increased risk was observed in one study [30], with decreased risk in three studies [12, 14, 19].

In the meta-analysis of 18 studies, the RR for pre-diagnosis alcohol intake was 0.90 (95% CI 0.82–0.98, p = 0.02), showing slight heterogeneity in the effect (I2 = 37%), and the risk was significantly reduced (Fig. 4A). The meta-analysis of seven studies showed that the RR for post-diagnosis alcohol intake was 0.88 (95% CI 0.77–1.02, p = 0.08), showing high heterogeneity in the effect (I2 = 67%), and the risk was not significantly reduced (Fig. 4B).

Fig. 4
figure 4

Forest plots of the hazard ratio for alcohol intake and all-cause mortality. A Pre-diagnosis; B post-diagnosis; C ER-positive cases; D ER-negative cases

Full size image

In addition, five studies of ER expression status in relation to all-cause death were reported. Risk reduction was observed in one ER-positive study [19] and three ER-negative studies [14, 19, 27], with three studies of pre-diagnosis alcohol intake [14, 19] and two studies of post-diagnosis alcohol intake [19, 27]. In the meta-analysis, the RR for ER-positive cases was 0.79 (95% CI 0.68–0.92, p = 0.003), showing no heterogeneity in the effect (I2 = 4%), and the risk was significantly reduced (Fig. 4c).

The RR for ER-negative cases was 0.68 (95% CI 0.50–0.91, p = 0.01), showing slight heterogeneity (I2 = 32%), and the risk was significantly reduced (Fig. 4d).

Association between alcohol consumption and cardiovascular disease mortality

There were two post-diagnosis studies [12, 16] of alcohol consumption and cardiovascular disease mortality in breast cancer patients. For post-diagnosis alcohol intake, decreased risk was observed in one study [16]. In two studies [12, 16], the RR for breast cancer patients was 0.47 (95% CI 0.28–0.79, p = 0.005), showing no heterogeneity in the effect (I2 = 0%), and the risk was significantly reduced (Fig. 5).

Fig. 5
figure 5

Forest plots of the hazard ratio for alcohol intake and cardiovascular disease mortality

Full size image

Discussion

Alcohol consumption has been reported to increase breast cancer incidence in healthy individuals [35]. However, it is unclear whether alcohol consumption affects breast cancer recurrence, breast cancer-related death, or all-cause death in breast cancer patients. Therefore, a systematic review and meta-analysis of epidemiological studies in breast cancer patients was performed. There were no significant associations between alcohol consumption and breast cancer recurrence [4,5,6,7,8,9,10,11,12,13] and breast cancer-related mortality [4, 8, 9, 12, 14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32]. There was high heterogeneity between pre-diagnosis alcohol consumption and breast cancer recurrence and breast cancer-related mortality. This may be due to the insufficient follow-up period in several cases [6, 8, 10, 18, 34] and the variability in the number of events (breast cancer recurrence: 28–1487 cases; breast cancer-related mortality: 26–824 cases) as shown in Table 1. On the other hand, a significant inverse association between pre-diagnosis alcohol consumption and all-cause morality [5, 8, 10, 14,15,16, 18,19,20,21, 26,27,28, 33, 34] and a significant inverse association between post-diagnosis alcohol consumption and cardiovascular disease mortality were found [12, 16].

Low to moderate amounts of weekly or daily alcohol consumption may be beneficial to cardiovascular health [36, 37]. The reduction in all-cause mortality associated with alcohol consumption after the diagnosis of breast cancer was thought to be related to decreased cardiovascular disease mortality. However, the relationship between alcohol consumption and cardiovascular disease mortality in patients with cardiovascular disease appears to be biphasic and have a J-shaped association [38].

Population level recommendations for upper limits of alcohol consumption exist in many nations as part of public health strategies aimed at reducing adverse health effects. These recommendations vary from country to country. For example, the United States Department of Agriculture recommends women consume not more than 1 standard drink per day (where 1 drink = 14 g of alcohol) [39].

The Ministry of Health, Labour and Welfare of Japan recommends consuming not more than 20 g of alcohol per day. However, this evidence is based on Tsugane’s study [40] and is the result for men. For women, smaller amounts than for men are considered appropriate, because women are smaller than men and are thought to break down alcohol more slowly. Indeed, an increased relative risk of all-cause mortality from alcohol drinking for men is found at 6 or more drinks, whereas for women it is found at 2.0–2.9 drinks [41].

Regarding ER status, we examined it in 5 studies [14, 19, 27]. Regarding the association between alcohol consumption and breast cancer-related death, the HR for ER-positive cases was 0.77 (95% CI 0.59–1.00, p = 0.05), and the HR for ER-negative cases was 0.79 (95% CI 0.56–1.11, p = 0.17). Regarding the association between alcohol consumption and all-cause death, the HR for ER-positive cases was 0.79 (95% CI 0.68–0.92, p = 0.002), the HR for ER-negative cases was 0.68 (95% CI 0.50–0.91, p = 0.01), and the risk was significantly reduced. There was no clear significant difference in risk associated with ER status.

There are several limitations of this systematic review. First, study selection involved filtering for English and Japanese language articles, potentially resulting in the exclusion of relevant, non-English publications. Second, in some studies, analyses were either performed without adjusting for any confounding factors, or analyses did not account for the factors considered most important in regard to the outcome of interest, such as age, BMI, and stage of the initial breast cancer.

Third, since alcohol consumption status is mainly based on questionnaire surveys, the exact amount and frequency of alcohol consumption are not known. In addition, there is the possibility of underreporting.

The present findings led us to create the recommendation for the clinical question in the Japanese Breast Cancer Society Clinical Practice Guideline Committee, 2022 edition, as follows: Whether pre-diagnosis or post-diagnosis, alcohol consumption has a substantial effect on risk and is unlikely to increase the risk of breast cancer recurrence and death from breast cancer.

However, excessive alcohol consumption increases the relative risk of all-cause mortality, and alcohol consumption should be limited to a moderate level as defined by each country.