Consumption of Red Versus White Wine and Cancer Risk: A Meta-Analysis of Observational Studies.

A comprehensive search of PubMed and Embase was conducted for studies published up to December 2023 using the Medical Subject Headings (MeSH) terms or keywords including cancer, tumor, carcinoma, malignant neoplasm, alcohol, wine, red wine, and white wine ( Supplementary Table S1 ). Three independent reviewers (RL, JR, and MH) screened the titles, abstracts, and full texts based on pre-defined eligibility criteria. Any disagreements regarding study inclusion were discussed and resolved by the three investigators (RL, JR, and MH). Additionally, the reference lists of identified articles and relevant meta-analyses or reviews were examined to locate additional relevant studies. The Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines were adhered to in the design, execution, analysis, and reporting of this meta-analysis [ 50 ]. We included prospective cohort and case–control studies that examined the association between red or white wine and cancer risk. We restricted our analysis to studies providing the type of wine (red or white, rather than wine as a whole) and relevant measures of association, relative risk (RR) or odds ratio (OR) estimates with corresponding 95% confidence intervals (CIs), to perform a meta-analysis. Articles not published in English, literature reviews, abstracts, posters, case reports, and experimental studies were excluded ( Figure 1 ). Studies with more than three categories of wine consumption, along with data on the case and non-case numbers, RR or OR estimates, and 95% CIs for each category, were eligible for inclusion in the dose–response analysis. From each included study, we extracted the following details: author names, publication year, study design (cohort or case–control), type of wine (red or white), number of cases and non-cases, sample size, the most fully adjusted estimates of association and corresponding 95% CIs, study region, measurement of wine consumption (validated or not), measurement of outcomes (confirmed by medical record or not), and any adjustments for potential confounders ( Table 1 ). For the dose–response meta-analysis, we collected category-specific doses of wine consumption (range, median) and the most adjusted RRs or ORs and their corresponding 95% CIs. The lowest wine intake category (nondrinkers) was used as the reference group. For each study, we assigned the midpoint of each wine intake category (range) as the corresponding RR or OR for that category. If the highest category was unbounded, we estimated the same range as the other categories for the highest category (e.g., if categories are 3, we assumed 3–5 for the last category and assigned 4 as the midpoint). Wine intake measurements were all converted to grams of ethanol per day for dose–response analysis. The standard drink definition established by the government of the study’s country of origin was utilized to calculate conversions [ 51 ]. All extracted data were independently reviewed and cross-verified by the authors at least twice. A meta-analysis for highest versus lowest exposure levels was conducted. The summary RRs and 95% CIs were calculated by using a random-effects model. We performed subgroup analyses by type of wine (red, white) and cancer type for those having more than three studies (skin, colon/rectum, kidney/urinary tract, lung, ovary, female breast, and prostate). Then, meta-regression was used to assess whether the risks of cancer overall or site-specific cancers differ by type of wine ( p -value for difference). We also conducted analyses restricted to cohort studies, which are less susceptible to recall or selection bias than case–control studies. A random-effects exposure–response meta-analysis was performed to examine a linear relationship between wine intake and cancer risk. Potential publication bias was assessed through visual inspection of funnel plot asymmetry and by the p -value from Egger’s tests [ 64 ]. Sensitivity analyses were conducted to evaluate the cancer risk, excluding studies that only had one type of wine intake (red or white) [ 18 , 43 , 46 , 47 , 61 ]. Additionally, the analysis was repeated after removing the most influential study with results that significantly deviated from the summary estimates. Between-study heterogeneity was evaluated using Cochran’s Q statistic and quantified with Higgins I 2 statistic and associated 95% CI [ 65 ]. Sources of heterogeneity were investigated through meta-regression and subgroup analyses, considering variables such as study design (case–control or cohort), geographic region (US/Canada or other regions), sex, publication year (1990s, 2000s), exposure measurement assessment (validated or not), and the level of confounder adjustment [smoking and body mass index (BMI)] ( Table 1 ).

Alcoholic beverages have been classified as Group 1 carcinogens (i.e., carcinogenic to humans) by the International Agency for Research on Cancer (IARC) based on sufficient evidence in humans for cancers of the oral cavity, pharynx, larynx, esophagus, liver, colorectum, and female breast [ 1 ]. According to the IARC Global Cancer Observatory database, 741,300 cancer cases in 2020, which is 4.1% of total global cancer cases for that year, were attributed to alcohol consumption. Ethanol in alcohol metabolizes into acetaldehyde, which readily forms Schiff-base adducts with DNA and cellular proteins [ 2 ], resulting in point mutations and harmful DNA–protein and DNA–DNA crosslinks [ 3 ]. It has also been found that some alcoholic beverages contain significant levels of preexisting acetaldehyde [ 4 , 5 ]. Despite these findings, the consumption of wine in particular has been rising [ 6 ]. It is not clear whether cancer risk differs by consumption of different types of wine, in particular red versus white. Red wine has been considered a healthier option because it contains a higher number of antioxidants, including flavonoids and polymeric tannins, than white wine [ 7 ]. In fact, a recent Canadian survey showed that 41% of respondents felt uncertain about whether or not red wine reduced cancer risk, while 54% agreed that alcohol consumption increased cancer risk [ 8 ]. Numerous experimental studies found that resveratrol, a polyphenol stilbenoid in red wine, inhibits the proliferation of several different types of cancer cells [ 9 , 10 , 11 , 12 , 13 , 14 , 15 ]. Epidemiologic studies have primarily examined the relationship between overall alcohol intake and cancer risk [ 16 , 17 ]; however, studies that have separately analyzed red and white wine have reported inconsistent findings. Several studies have observed inverse or no associations between red wine intake and cancers of the breast, skin, prostate, ovaries, colon, and lungs [ 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 ], while other studies have shown increased risk of aforementioned cancers [ 19 , 37 , 39 , 44 , 45 , 46 , 47 ]. Similarly, white wine, which generally contains a lower amount of resveratrol than red wine (0–1.09 mg/L vs. red wine: 0.36–1.97 mg/L [ 48 ]), has been positively [ 29 , 32 , 33 , 35 , 36 , 41 , 44 ], inversely, or not [ 34 , 42 ] associated with cancers of the breast, skin, and prostate. A systematic review and meta-analysis summarized the relationship between wine consumption and cancer risk but did not evaluate red and white wine separately [ 49 ]. To our knowledge, there are few studies evaluating the difference in cancer risk by type of wine. To address this research gap, especially given the vast and often contradictory literature on the carcinogenicity of red and white wine separately, we conducted a meta-analysis to investigate the association between red versus white wine consumption and the risk of all and site-specific cancers. We aimed to clarify the potentially differing carcinogenic effects of red and white wine and contribute to a more nuanced understanding of the role of wine type in cancer risk. Ultimately, this research may provide valuable insights to inform public health guidelines and individual lifestyle decisions regarding wine.

The searching scheme resulted in a total of 252 published articles, and we selected 42 articles (20 cohort [ 20 , 21 , 22 , 25 , 26 , 29 , 31 , 32 , 35 , 36 , 37 , 38 , 41 , 43 , 44 , 45 , 53 , 56 , 60 , 63 ] and 22 case–control studies [ 18 , 19 , 23 , 24 , 27 , 28 , 30 , 33 , 34 , 39 , 40 , 42 , 46 , 47 , 52 , 54 , 55 , 57 , 58 , 59 , 61 , 62 ]) to conduct a meta-analysis ( Figure 1 , Table 1 ). Of the 42 studies selected, 25 used validated methods of dietary assessment [ 18 , 20 , 21 , 22 , 25 , 26 , 27 , 28 , 29 , 30 , 32 , 34 , 35 , 36 , 37 , 38 , 40 , 41 , 43 , 44 , 45 , 53 , 59 , 60 , 63 ] ( Table 1 ). A total of thirty-seven studies included measures of association for site-specific cancer incidence and consumption of both red and white wine, and five studies included data for red wine intake only [ 18 , 43 , 46 , 47 , 61 ]. We included cancers of the skin (number of studies = 6) [ 29 , 32 , 36 , 41 , 42 , 45 ], prostate ( n = 6) [ 20 , 34 , 35 , 37 , 38 , 53 ], female breast ( n = 5) [ 19 , 25 , 30 , 39 , 44 ], colon/rectum ( n = 5) [ 24 , 28 , 31 , 43 , 59 ], ovary ( n = 5) [ 23 , 26 , 27 , 52 ], lung ( n = 4) [ 21 , 22 , 33 , 46 ], kidney/urinary track ( n = 3) [ 18 , 56 , 57 ], pancreas ( n = 2) [ 54 , 55 ], brain ( n = 2) [ 62 , 63 ], lymphoma ( n = 2) [ 60 , 61 ], stomach ( n = 1) [ 58 ], and mouth/pharynx ( n = 1) [ 46 ]. When a study presented effect estimates for multiple cancers (different sites or a subtype) or sex-specific estimates, we included all estimates and considered them as separate studies in our analyses. The summary RR for the overall cancer risk, comparing the highest versus lowest level of wine intake, was 0.98 (95% CI = 0.87 to 1.10) for red wine and 1.00 (95% CI = 0.91 to 1.10) for white wine ( Table 2 ). We observed no difference between red and white wine consumption on overall cancer risk (P difference = 0.74). After excluding five studies [ 18 , 43 , 46 , 47 , 61 ] that only included measurements of association for red wine, we still observed no difference in overall cancer risk between the two types of wine (P difference = 0.38). When restricted to cohort studies, the association with white wine intake became stronger and significant (RR: 1.12, 95% CI: 1.03, 1.22), while the null association remained for red wine intake (RR: 1.02, 95% CI: 0.96, 1.09), and the difference became significant (P difference = 0.02). The difference between red and white wine consumption on overall cancer risk was not significant in men (P difference = 0.37) but was significant in women (P difference = 0.03). The eligible studies for linear dose–response analyses were three cohort studies [ 22 , 25 , 38 ] and twelve case–control studies [ 18 , 19 , 23 , 25 , 30 , 34 , 39 , 47 , 52 , 54 , 57 , 58 , 59 ]. Based on these studies, we observed that every additional 10 g of estimated ethanol (e.g., about one glass) from red wine per day was associated with a 5% increase in overall cancer risk (summary RR = 1.05 [1.03, 1.08]); however, this association was null when restricted to cohort studies (1.01 [0.97, 1.04]) [ 21 , 24 , 37 ]. We found no significant dose–response relationship for white wine (RR = 1.02 [0.98, 1.05]), which was consistent when restricting to cohort studies (RR = 1.00 [0.96, 1.04]). When we evaluated the association between wine intake and cancer risk by cancer site, we observed a significant difference in skin cancer risk associated with red versus white wine (P difference = 0.0003). White wine intake was associated with a 22% increased risk of skin cancer (RR = 1.22, 95% CI = 1.14 to 1.30, Table 2 , Figure 2 ), while red wine consumption was not associated with skin cancer (RR = 1.02, 95% CI = 0.95 to 1.09). Studies of skin cancer were not eligible to assess the dose–response relationship. An increased risk of breast cancer was observed for both wine types (red: RR = 1.17, 95% CI = 0.97 to 1.42, white: 1.12, 95% CI = 1.05 to 1.20) with no significant difference (P difference = 0.61). Other than skin cancer, we did not observe a significant difference between the associations of red and white wine consumption and the risk of site-specific cancers. The test of heterogeneity resulted in a moderate level of heterogeneity (I 2 = 63.9%, P heterogeneity < 0.0001) ( Table 3 ). After excluding the most influential study [ 39 ], there was no change in directionality and significance of the summary RRs of high vs. low meta-analysis for overall cancer risk (red: 0.96, 95% CI: 0.86, 1.08; white: 1.00, 95% CI: 0.91, 1.10) and no significant difference between red and white wine ( p = 0.63). We observed possible publication bias through visual inspection of the funnel plot ( Supplementary Figure S1 ) and Egger’s tests ( p = 0.03). We used Duval and Tweedie’s trim and fill method [ 66 ] to estimate what the summary effect size would be if there was no publication bias. Briefly, this method omits small studies until the funnel plot is symmetrical (“trimming”), uses the trimmed plot to estimate the “true” center of the funnel plot, and then puts back the omitted studies, as well as their theoretical missing “counterparts”, around the new center (“filling”). We observed slightly increased summary RRs for both red and white wine on overall cancer risk after conducting trim and fill analysis (red: 1.00, 95% CI 0.91, 1.09; white: 1.13, 95% CI 1.03, 1.24) ( Supplementary Figure S2 ).

We conducted a meta-analysis to investigate the association between red or white wine consumption and cancer risk. Although red wine has been considered healthier compared to white wine [ 67 , 68 , 69 , 70 ], there was no difference in cancer risk between the consumption of the two types of wine. When we restricted analysis to cohort studies, we observed a significant increase in overall cancer risk with white wine consumption. White wine intake was significantly associated with an increased risk of skin cancer compared to red wine intake. A recent UK cohort study evaluating alcohol consumption patterns and health outcomes found no significant increase in the overall cancer risk and alcohol-related cancer (colon, rectum, breast, liver, esophagus, and larynx) incidence with white wine consumption compared to that of red wine, which is in line with our findings [ 71 ]. Red wine contains resveratrol, a natural stilbene and a non-flavonoid polyphenol, that possesses antioxidant, anti-inflammatory, cardioprotective, and anti-cancer properties [ 72 ]. Resveratrol has been investigated extensively as a potential chemopreventive agent because it has been shown to inhibit the proliferation of cancer cells in the breast, colon/rectum, skin, stomach, and kidney [ 9 , 10 , 11 , 12 , 13 , 14 ]. Many in vitro studies found that resveratrol influences tumor initiation and cancer progression pathways; however, in vivo studies have yielded mixed results [ 73 ]. For example, resveratrol can promote cell-cycle arrest leading to apoptosis of tumor cells, prevent tumor-derived nitric oxide synthase expression to block tumor growth and migration, as well as act as an antioxidant to prevent DNA damage that can lead to tumor formation [ 73 ]. Wu et al. [ 74 ] found that resveratrol is more toxic to cancer cell lines than to normal cell lines by comparing the activity of both types of cells following treatment with resveratrol. Additionally, it has been observed that combination therapy of resveratrol and other cancer drugs kills breast cancer cells more dramatically than either treatment in isolation [ 75 ]. Although there is some evidence of an anticarcinogenic effect of resveratrol in experimental studies, we did not observe that red wine consumption is associated with a reduced risk of cancer in our meta-analysis. In addition, there was no difference in the association between red vs. white wine and cancer risk, except for skin cancer. A potential reason why we did not observe a difference in cancer risk between red and white wine may be that resveratrol has a fast metabolism [ 76 ]. Although concentrations of resveratrol are greater in red compared to white wine [ 48 , 77 ], studies have demonstrated that resveratrol is metabolized quickly, with nearly 75% excreted via feces and urine [ 76 ]. Several clinical studies have shown that the peak levels of resveratrol in plasma were low, given its extensive metabolism and poor bioavailability, after single or repeated oral administration [ 78 ]. A study involving 15 healthy participants who were given a single dose of 500 mg resveratrol observed a peak plasma concentration of 71.2 ng/mL [ 79 ]. Another study of 40 healthy subjects who received daily oral resveratrol of 500–5000 mg/day for 29 days reported a peak plasma concentration ranging from 44 to 967 ng/mL between days 21 and 28 [ 80 ]. Daily consumption of two glasses of red wine (375 mL) is equivalent to a dose of ~27 µg/kg body weight of resveratrol for a 70 kg individual, which leads to detectable concentrations of derivatives but not free resveratrol [ 81 ]. These low levels of resveratrol after consuming red wine may not result in bio-effective concentrations that may eventually make differences in health outcomes compared to white wine consumption. We observed that white wine was associated with an increased risk of skin cancer. It has been hypothesized that alcohol consumption can promote skin carcinogenicity through the intermediate byproducts or metabolites of alcohol, such as acetaldehyde and reactive oxygen species, that have photosensitizing effects [ 82 ]. Also, drinking wine may be more common among whites who are more susceptible to skin cancer than other racial groups [ 68 , 83 ]. However, it may not explain why white wine specifically was associated with skin cancer risk. In addition, heavy wine consumption may be related to high-risk behaviors such as sunburn, indoor tanning, and lack of use of sun protection [ 84 ]. In fact, earlier studies have demonstrated an association between alcohol consumption and increased prevalence of severe sunburn [ 82 , 85 ]. Given that all six studies properly adjusted for detailed information on the risk factors of skin cancer, e.g., skin type, sun exposure, and sun protection habits, a positive association we found in the meta-analysis was most likely the true detrimental effect of white wine [ 29 , 32 , 36 , 41 , 42 , 45 ]. The sample size for skin cancer was the largest among individual cancer sites and potentially provided enough statistical power to detect the difference. Wine intake (both red and white) was associated with increased breast cancer risk, which may suggest that resveratrol is not a significant factor in breast carcinogenesis. There are limitations to this meta-analysis. First, wine intake was obtained from a self-reported food frequency questionnaire (FFQ). While most of the included cohort studies used validated FFQs (18 out of 19), only six out of twenty-two case–control studies used validated FFQs. However, because the intake information on red and white wine was collected in each study, the misclassification would similarly affect the intake assessment of both wines. Nevertheless, the misclassification of exposure might have occurred randomly regardless of cancer status, which may attenuate our findings. In addition, exposure validation was not a source of heterogeneity in our summary estimates for both red and white wine. Additionally, case–control studies may be more susceptible than cohort studies to recall bias when evaluating wine intake; however, we also presented results restricted to cohort studies. For some cancer sites including the pancreas, lymphatic system, brain, mouth, and pharynx, there was a limited number of studies (<3) to explore the risk difference between red and white wine. In addition, our dose–response analyses were limited to a smaller number of studies, largely case–control studies. Potential regression dilution bias in the included cohort studies might have also played a role in underestimating summary risk estimates [ 86 ]. However, we added an important finding to the controversial conversation about whether red wine intake reduces cancer risk. The quality of outcome measurements in all included studies was reliable (from a cancer registry or confirmed by medical professionals). Although we found evidence of potential publication bias, we found similar summary estimates after conducting trim and fill analysis.

To our knowledge, we conducted the first meta-analysis and the largest investigation of red versus white wine consumption and cancer risk and found no difference between the two types of wine. However, we observed that white wine had a significantly stronger association with cancer when the analysis was restricted to cohort studies. Furthermore, white wine intake, but not red wine intake, was associated with an increased risk of skin cancer. Our findings provided a critical public health message that drinking red wine may not be any better than drinking white wine in terms of cancer risk.