Glioma is a devastating tumour of the central nervous system, accounting for approximately 80 % of adult malignant brain tumours(Reference Ostrom, Gittleman and Liao1). It is reported that the global incidence rate of glioma is 3·7/100 000 for males and 2·6/100 000 for females(Reference Bondy, Scheurer and Malmer2). Despite the low incidence rate, glioma is associated with high mortality and poor prognosis(Reference Aminianfar, Vahid and Shayanfar3). Indeed, apart from few established risk factors, such as exposure to ionising radiation, White race/ethnicity, little is known regarding the effect of modifiable risk factors (e.g. diet and alcohol intake) on glioma(Reference Li4).Therefore, identifying the relationship of alcohol intake with glioma is valuable.
Over the past decades, alcohol intake has been recognised as an important risk factor for several types of cancer, including breast cancer(Reference Brennan, Cantwell and Cardwell5), colocteral cancer(Reference Feng, Shu and Zheng6) and liver cancer(Reference Kim, Ko and Han7). Alcohol is neurotoxic and can traverse the blood–brain barrier. A previous study has described the short- and long-term effects of excessive alcohol consumption on brain function and pathology(Reference Harper8). To date, substantial epidemiological studies have explored the relationship between alcohol consumption and the risk of glioma(Reference Braganza, Rajaraman and Park9–Reference Burch, Craib and Choi13). But, results from these studies have been inconsistent. In the NIH-AARP Diet and Health Study, Braganza et al. found the significant inverse associations between alcohol and beer intake and glioma risk(Reference Braganza, Rajaraman and Park9). In addition, a recent report from three prospective cohort studies also found a significant inverse association between alcohol intake and glioma risk in both men and women(Reference Cote, Samanic and Smith11). However, in a hospital-based case–control study, Burch and his colleagues found that wine consumption was associated with an elevated risk of glioma(Reference Burch, Craib and Choi13). Furthermore, a previous meta-analysis from nine observational studies has shown no material association between alcohol consumption and risk of glioma(Reference Qi, Shao and Yang14). Therefore, to clarify the exact association between alcohol intake and glioma risk, we conducted this systematic review and updated meta-analysis to summarise the evidence from observational studies published up to December 2021.
Material and methods
This systematic review and meta-analysis has adopted the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines(Reference Moher, Liberati and Tetzlaff15) and was written according to the Meta-analysis of Observational Studies in Epidemiology proposal(Reference Stroup, Berlin and Morton16).
Literature search strategy
A comprehensive literature search was performed using the PubMed, Web of Science, CNKI and Wan Fang databases to identify relevant articles written in the English and Chinese languages published through December, 2021, with the following search terms: (‘alcohol’ OR ‘ethanol’ OR ‘alcohol drinking’ OR ‘alcohol intake’) AND (‘glioma’ OR ‘gliblastoma’ OR ‘brain cancer’ OR ‘brain tumour’). The search was restricted to human studies. Moreover, we also reviewed the computer-retrieved studies for reference lists by hand-searching.
Studies included criteria
Two independent reviewers (Shu L and Jin FB) read the abstracts of articles retrieved in the initial search to identify human studies that examined the relationship between alcohol intake and the risk of glioma. Differences between the two independent reviewers were resolved by consensus and referred to the third reviewer if necessary. When all agreed, the full-text versions of articles were reviewed against inclusion and exclusion criteria for this meta-analysis. Studies were included if they met the follow criteria: (1) an original study reporting the association between alcohol intake and glioma risk; (2) used a case–control, nested case–control or cohort design; (3) estimates of relative risk (RR) (OR, hazard ratio and rate ratio) with corresponding 95 % CI were provided (or sufficient data to calculate them); and (4) if the data in original publication lacked sufficient detail, the corresponding author of the study was contacted for additional information by email. Studies were excluded if they met one of based on the follow criteria: written in a language other than English or Chinese; not performed on humans; reviews and letters; and studies with insufficient data. Finally, eight studies reported the association between alcohol intake and the risk of glioma.
Data extraction
Two reviewers (Shu L and Jin FB) independently extracted the following data from included studies: the first author’s last name, publication year, geographic, study design, age for cases and participants, number of cases and controls or participants, type of controls, methods used for collection of data on exposure, exposure classification, confounders adjusted for, and the OR, RR or HR estimates with corresponding 95 % CI for the heavy drinking, low–moderate drinking v. non-drinking. Any discrepancies were resolved with a group discussion with a third investigator (Yu D).
Definition of ‘high intake’ and ‘moderate intake’
The different forms of alcohol intake were converted into grams of ethanol per d (e.g. 1 drink = 12·5 g, 1 ml = 0·8 g, 1U = 8 g, 1 oz = 28·35 g of ethanol)(Reference Galeone, Malerba and Rota17). Alcohol intake >25 g/d (or two drinks/d) for men or >12·5 g/d (or 1 drink/d) for women was defined as high intake of alcohol or heavy alcohol drinking; alcohol intake<12·5 g/d for men and <7·5 g/d for women was defined as low intake of alcohol or low alcohol drinking, and alcohol intake of >12·5/d and <25 g/d for men or >7·5 g/d and <12·5 g/d for women was defined as moderate alcohol drinking(Reference Zhang, Shu and Si18).
Assessment of heterogeneity
We performed the Cochran’s Q test and I 2 statistic to test and quantify the heterogeneity among the included studies. A P value of Q test >0·10 indicated an absence of heterogeneity between included studies, and the fixed-effects model was used to calculated the pooled RR. If a P value of Q test ≤0·10 indicated a high degree of heterogeneity among studies, then the random-effects model (DerSimonnian and Laird method) was used(Reference Higgins, Thompson and Deeks19).
Quality assessment
The Newcastle–Ottawa Quality Scale (NOS) was applied to assess the quality of the included studies in this meta-analysis(Reference Stang20). This scale includes four points for selection, two points for comparability and three points for the assessment of outcomes. Finally, studies with a score of greater or equal to 7 were identified as high-quality studies(Reference Zhong, Zhu and Li21). Disagreements were resolved by discussion to reach a consensus.
Data synthesis and statistical analysis
To identify the relationship between alcohol intake and glioma risk, we used meta-analysis to summarise the risk estimate for the heavy drinking, low–moderate drinking v. non-drinking using OR, RR, and HR and corresponding 95 % CI for the included studies. Given the prevalence of glioma was relatively low, OR and HR were directly considered as RR(Reference Greenland22). Multivariable adjusted OR, HR and RR with corresponding 95 % CI from individual studies were combined to produce an overall RR. Publication bias was assessed by inspection of the funnel plot and by formal testing for ‘funnel plot’ asymmetry using Begg’s test and Egger’s test(Reference Begg and Mazumdar23). Moreover, sensitivity analysis was carried out to determine whether sex, study design, geographic area and study quality affected study conclusions. All statistical analyses were carried out using the STATA software, version 12 (Stata Corp.). Statistical tests were two-sided with P value<0·05 accepted as statistically significant.
Results
Overview of included studies for the systematic review
An electronic literature search in PubMed, Web of Science, CKNI and Wan fang database identified 401 studies, 393 of which were excluded based on the reasons listed in Fig. 1. Eight articles(Reference Braganza, Rajaraman and Park9–Reference Baglietto, Giles and English12,Reference Hu, Little and Xu24–Reference Ryan, Lee and North27) met the inclusion criteria and were included in this meta-analysis, including five cohort studies(Reference Braganza, Rajaraman and Park9,Reference Cote, Samanic and Smith11,Reference Baglietto, Giles and English12,Reference Efird, Friedman and Sidney25,Reference Benson, Pirie and Green26) and three case–control studies(Reference Hurley, McNeil and Donnan10,Reference Hu, Little and Xu24,Reference Ryan, Lee and North27) .The characteristics of the included studies were summarised in Table 1.
Flow chart of article screening and selection process.
Characteristics of studies on alcohol intake and risk of glioma (–2021)
(Risk ratio, hazard ratio, odd ratio and 95 % confidence intervals)
Branganza 1, Hurley 1, cote 1, et al. represent the data for men. Branganza 2, Hurley 2, cote 2, et al. represent the data for women.
Alcohol drinking
The heavy alcohol drinking was characterised by high intakes of alcohol-containing beers, wines and spirits. Pooled results form six articles (including eight original studies) identified a heavy alcohol drinking (Fig. 2). Fig. 2 showed no evidence of an increased risk of glioma in the heavy alcohol drinking v. non-drinking (RR = 0·89; 95 % CI (0·67, 1·18); P = 0·404). Data from these studies were assessed using random-effects model, and there was significant heterogeneity (I 2 = 43·7 %, P = 0·087). Eight articles reporting eleven original studies identified a low–moderate alcohol drinking in this meta-analysis (Fig. 3). There was evidence of a reduced risk of glioma in the low–moderate alcohol drinking compared with non-drinking (RR = 0·87; 95 % CI (0·78, 0·97); P = 0·014). A fixed-effects model was used to assess the data, and there was no evidence of heterogeneity (I 2 = 0·0 %, P = 0·656).
Forest plots for RR of heavy alcohol drinking v. non-drinking. RR, relative risk.
Forest plots for RR of light–moderate alcohol drinking v. non-drinking. RR, RR, relative risk.
Publication bias
Inspection of funnel plots did not reveal evidence of asymmetry (Fig. 4 and 5). Egger’s test for publication bias was not statistically significant (heavy alcohol drinking v. non-drinking: P = 0·536; low–moderate alcohol drinking v. non-drinking: P = 0·458).
Funnel plots analysis to detect publication bias in heavy alcohol drinking v. non-drinking.
Funnel plots analysis to detect publication bias in the light–moderate alcohol drinking v. non-drinking.
Quality assessment
The quality of included studies using Newcastle–Ottawa criteria is detailed in Appendix 1. When included studies received a score of 6 or higher, they would be deemed to be of relatively higher quality(Reference Braganza, Rajaraman and Park9–Reference Baglietto, Giles and English12,Reference Begg and Mazumdar23–Reference Benson, Pirie and Green26) .
Sensitivity analysis
The sensitivity analysis revealed that differences in age, sex, ethnicity and study design had an effect on the relationship between alcohol intake and glioma risk. When moderate alcohol drinking was compared with non-drinking, the alcohol intake/glioma association was stronger when subjects were women, White and more than 50 years old, and study design was cohort. As these variables have a strong effect on relationship between alcohol intake and glioma risk, their differences may partially explain the heterogeneity between studies (Table 2).
Alcohol intake and glioma: sensitivity analysis
(Risk ratio and 95 % confidence intervals)
Discussion
Existing evidence on the role of alcohol intake and the incidence of glioma is limited and inconsistent. To the best of our knowledge, this is the latest systematic review and meta-analysis on the effect of alcohol intake on glioma. In this study, we found a significant inverse association between low–moderate alcohol drinking and the risk of glioma. Meanwhile, no significant association between heavy alcohol drinking and the risk of glioma was observed. Data from eight articles involving 2706 glioma cases and 2 189 927 participants were included in this meta-analysis. Our findings provide further evidence on the role of alcohol intake and the risk of glioma, though the lack of a dose–response relationship suggests caution in the interpretation of results.
In our analyses, the significant inverse association was identified between low–moderate alcohol drinking and the risk of glioma. Our findings are inconsistent with a previous meta-analysis of alcohol consumption and the risk of glioma(Reference Qi, Shao and Yang14). Qi et al. reported no material association between alcohol consumption and risk of glioma (total alcohol drinks v. non-drinks: RR = 0·96, 95 % CI (0·89, 1·04))(Reference Qi, Shao and Yang14). In their meta-analysis, the main analysis is ‘ever and alcohol drinkers v. nondrinkers’. Qi and colleagues did not analyse the relationship between different levels of alcohol consumption and the risk of glioma. Thus, a lack of consideration for the association between different drinking group and glioma could contribute somewhat the variance in results. In a comprehensive meta-analysis of alcohol consumption and risk of brain tumours, Galeone and colleagues also found that alcohol drinking did not appear to be associated with adult brain cancer(Reference Galeone, Malerba and Rota17). The difference to our study is that Galeone et al. did not analyse glioma or glioblastom separately from other brain tumours. This analysis of combination of glioma and other brain tumours could make these findings more confounding. Also, inclusion of eight articles reporting eleven original studies with a larger sample size might explain the modest stronger association observed in our meta-analysis. In contrast, a more recent report from three prospective cohort studies found a significant inverse association between alcohol intake and glioma risk in both men and women(Reference Cote, Samanic and Smith11). Cote et al. estimated HR of glioma and 95 % CI by category of alcohol intake and adjusted the covariates including BMI, smoking status and total energetic intake. In the NIH-AARP Diet and Health Study, an analysis including 704 glioma cases also identified significant inverse, dose-dependent associations between alcohol and beer intake and risk of glioma, but no associations for wine or liquor(Reference Braganza, Rajaraman and Park9). In short, the evidence linking alcohol consumption with glioma is inconsistent. Although ethanol has been classified as carcinogenic to humans by the International Agency for Research on Cancer (IARC)(28), there are several plausible explanations for this favourable effect of low–moderate alcohol drinking on glioma. First, xanthohumol, a flavonoid present in beer, has exhibited its anticancer properties via inhibition of various signalling pathways, for example, disruption of the activation of transcription factors, suppression of multiple protein kinases and regulation of the expression of genes which related to cell proliferation, angiogenesis and apoptosis(Reference Festa, Capasso and D’Acunto29,Reference Zajc, Filipič and Lah30) . Second, laboratory evidence has also shown that certain components of red wine, such as phenols, may play an important role in reducing the growth and development of glioma(Reference Arranz, Chiva-Blanch and Valderas-Martínez31). These mechanisms mentioned above could explain the observed association between low–moderate alcohol drinking and risk of glioma. Meanwhile, in this study, we observed no significant association between heavy alcohol drinking and glioma risk. Our results are inconcordant with a previous study(Reference Baglietto, Giles and English12), which suggests that alcohol consumption increases the risk of glioblastoma consistent with a dose–response relationship. Alcohol drinking has been consistently considered as an important risk factor for cancers(Reference Shu, Wang and Wang32). Data from the Melbourne Collaborative Cohort Study including sixty-seven glioblastoma cases showed no significant differences for those drinking<20 g/d of alcohol, but a higher risk for those drinking 40–59 g/d (HR = 3·07, 95 % CI (1·26, 7·47)) and ≥60 g/d (HR = 2·54, 95 % CI (0·92, 7·00)), compared with lifetime abstainers(Reference Baglietto, Giles and English12). Although we observed no significant association between heavy alcohol drinking and glioma risk in this study, several plausible mechanisms have also been proposed. First, alcohol is an identified human carcinogen that penetrates the blood–brain barrier and thus may play an important role in the development of glioma(Reference Cote, Samanic and Smith11). Second, acetaldehyde is an intermediate product of alcohol metabolism, which have been shown to induce DNA lesions, generate free radicals and damage enzymes involved in DNA repair and antioxidant protection(Reference Seitz and Stickel33). Third, animal studies have also shown that N-nitroso compounds contained in alcohol can result in brain tumours(Reference Harper8). In a hospital-based case–control study, Hurley et al. found no significant association between heavy alcohol consumption and risk of glioma in both men and women(Reference Hurley, McNeil and Donnan10). However, the risk estimate was only adjusted for age and reference rate, and residual confounding was possible. In case–control studies of alcohol consumption in particular, the risk of recall bias may be substantial, and this bias may affect the relationship between alcohol intake and glioma risk(Reference Barry34). There are several possible explanations for the null association. First, alcohol intake might have changed during the follow-up, such as after a diagnosis of glioma. This change in alcohol intake could attenuate the association between heavy alcohol drinking and the risk of glioma. Second, we were unable to analyse the effect of specific alcohol types on glioma, because limited data were available. Finally, to the above-mentioned, some constituents in alcoholic beverages, for example, beer, red wine have been reported to have anticancer properties(Reference Zajc, Filipič and Lah30,Reference Arranz, Chiva-Blanch and Valderas-Martínez31) .
Strengths and limitations
This systematic review and meta-analysis had its own strengths and limitations. First, this is the latest systematic review and meta-analysis on alcohol intake in relation to the risk of glioma. We not only have an update on the earlier meta-analysis (Qi et al. in 2014)(Reference Qi, Shao and Yang14) but also further clarify the relationship between heavy alcohol drinking and low–moderate alcohol drinking and glioma risk. Second, the cases of glioma have been diagnosed through clinical manifestations, pathological section or endoscopic ultrasonography, avoiding misdiagnosis. Third, no signs of publication bias were evident in the funnel plot, and the statistical test for publication bias was non-significant. However, several limitations should be noted in this study. First, due to this meta-analysis was based on observational studies (i.e. case–control or cohort design), confounding factors are often of concern. Thus, we cannot rule out the probability that these findings were susceptible to recall and selection bias. Second, there was an inconsistent adjustment for potential confounders in the included studies. As a result, the data included in our analysis might suffer from differing degrees of completeness and accuracy. Third, because of scanty data be available in included studies, we were unable to assess separately various types of glioma, for example, glioblastoma and oligodendroglioma. Finally, the potential publication bias may distort the relationship between alcohol intake and glioma risk.
Conclusion
In conclusion, this systematic review and updated meta-analysis suggests an inverse association between low–moderate alcohol drinking and the risk of glioma. However, the lack of a dose–risk relationship for these findings indicates caution in their interpretation. Our findings need to be affirmed in further randomised controlled trials or large prospective studies.
Acknowledgements
The authors thank all participants from Department of Nutrition and Endocrinology, Zhejiang Hospital, for their assistance and support. Besides, the authors also acknowledge Dr Yu for their important contributions to data collection and analysis in this study.
This study was supported by Traditional Chinese Medicine Research Project of Zhejiang (No.2020ZB009, 2021ZB010), and Medical and Health research fund project of Zhejiang Province (No. 2022KY006).
The authors’ responsibilities were as follows: L. S. and F.-B. J. took responsibility for data integrity and the accuracy of data analysis; L. S. was responsible for study concept and design; F.-B. J. and D. Y. acquired the data; L. S. and D. Y. were responsible for analysis and interpretation of the data; L. S. performed the statistical analysis; and F.-B. J. and L. S. drafted the manuscript. All authors critically revised the manuscript for important intellectual content.
The authors declare no conflict of interest associated with this paper.
