Does Drinking White Wine Give You Melanoma?

Brown University put out a press release this month about some of its soon-to-be-published research. The study, named “Alcohol Intake and Risk of Incident Melanoma: A Pooled Analysis of Three Prospective Studies in the United States,” appears this month in Cancer Epidemiology, Biomarkers & Prevention, a journal of the American Association for Cancer Research. It looks at the drinking habits of more than 210,000 participants, and found bad news for those who enjoy a glass of white wine: the findings suggests that drinking white wine will significantly raise your risk for melanoma! Or so the media reports would have you think. Regular Skeptoid readers will be familiar with the paucity of such science reporting: sensational headlines with unsupported conclusions often dominate such news. Let’s take a close look at this research and determine if this pop-science flavor-of-the-week is in fact science, or just science fiction.

Typical shape and design of a white wine tasting glass. New Zealand wine via wikimedia

In the interest of full disclosure, I should note that one of the principal investigators for this study is from my affiliated university: Eunyoung Cho, Sc.D., an associate professor of dermatology and epidemiology at the Warren Alpert Medical School of Brown University in Providence, Rhode Island.

The results were interesting. Overall alcohol intake was associated with a 14 percent higher risk of melanoma per drink per day. Each drink per day of white wine was associated with a 13 percent increased risk of melanoma. Other forms of alcohol—beer, red wine, and liquor—did not significantly affect melanoma risk.

This is just one piece of data supporting a growing body of evidence indicating that alcohol is associated with several types of cancer. What was surprising about this research is the high association with skin cancer lesions (melanoma) in low-exposure areas—finding cancer on areas of the skin that don’t get a lot of sun. Obviously, that is considered unusual since cancer correlates best with sun exposure/damage. Secondly there was one significant blip that has drawn all the media attention, namely that white wine in particular had a much higher association with melanoma. According to reporting in EurekAlert!:

Cho said that compared with nondrinkers, those who consumed 20 grams or more of alcohol per day were 2 percent more likely to be diagnosed with melanomas of the head, neck, or extremities, but 73 percent more likely to be diagnosed with melanomas of the trunk.

Dr. Cho also commented:

It was surprising that white wine was the only drink independently associated with increased risk of melanoma. The reason for the association is unknown. However, research has shown that some wine has somewhat higher levels of pre-existing acetaldehyde than beer or spirits. While red and white wine may have similar amounts of pre-existing acetaldehyde, the antioxidants in red wine may offset the risks.

Does this study actually say that white wine causes melanoma? No, it doesn’t,there are major flaws that put doubt on that conclusion. Here is why: all correlational studies have one common flaw, it is an association, not a causative link. Lay persons and media outlets fail almost universally on that distinction. A single study suggesting a possible association is no more definitive than saying drinking water causes car accidents. All people who have car accidents drink water; does drinking water cause car accidents? The correlation is very definitive if you see it replicated multiple times from different studies and types of studies, such as the correlation of lung cancer and smoking.

The most surprising finding is that white wine in particular had a significantly higher correlation than other alcohol. There are many reasons why that might be false conclusion, such as confounding cultural variables. What if the study had a high degree of white wine-drinking Scandinavians with albinism? It might be that the white wine drinkers were all fair skinned and passed out drunk in the sun, or they were nudists. Although these suggestions are unlikely they are not impossible. We don’t know because they didn’t really do the study with those types of controls in mind. The original structure was asking a more general question, so the correlation is at best weak, even if it is mathematically strong. It is interesting and worthy of further study, but without those controls in place for those factors we really don’t know what this says.

Additionally the study had a major flaw that is significant for melanoma. It didn’t control for sun exposure. The single biggest factor in skin cancer wasn’t controlled for. That really drives a wedge into any causality statements for white wine. There is no evidence at all for the the statement by the researcher that chemicals in the wine are the causative factor. That is pure speculation by the researcher suggesting a cause of an unknown with another unknown.

Nonetheless, following the press release, white wine has been widely reported as raising the risk for melanoma. This is another example of poor science reporting. Research is often offered by the news media outlets as isolated factual Truth. Skeptoid listeners know that scientific knowledge is rarely revolutionary. Rather, it’s built brick by brick until you have a solid hypothesis with multiple lines of evidence supporting it. This study is just one piece of new evidence; it is an association, and not a very definitive one.

4mm thick nodular melanoma. Via Wikimedia

Press attention is always good when you’re looking to fund further research, but trying to get it can lead to one’s work being wildly misrepresented. This is a common flaw in the game that researchers play with the news media: trying to call attention to their research by expounding on some new finding that makes their research more catchy and newsworthy. It takes a finding, such as here the correlation of increased risk of cancer with alcohol, and puts a headline-grabbing spin on the research. Does it mean that the research is fabricated or false? No, it doesn’t. It is a very subtle discussion that the press extrapolates from freely to make the story more interesting. In practice it is a win-win for researchers and media outlets. But the science-illiterate reporting works to the detriment of the public. In this case, such exaggeration is probably not harmful, but definitely inaccurate.

So what is the definitive answer? Does white wine cause cancer? Truth be told we don’t know. We have a pretty well structured study that suggests a unexpected finding, with a plausible but unsupported guess as its mechanism. Based on this one study I don’t think your dermatologist is going to recommend you stay away from white wine just yet. It is an interesting finding that deserves a focused, controlled, and well-structured scientific evaluation. I suspect in the end that we will find that this was statistical noise generated from the large number of subjects in the study. I hope I am not wrong, I would hate to think that a nice glass of Riesling is as dangerous as getting a sunburn.

That is the beauty of the scientific method, its ability to answer these questions. Despite the news reports, white wine causing melanoma has not been answered. Realistically we just have found the question.

Whenever you see a new fantastic and exciting medical discovery broadly disseminated throughout the media, remember you have good reason to be skeptical.

Take a minute and support Skeptoid. The money doesn’t go to me, but instead goes to keep Skeptoid running as a resource of science and skepticism. Remember: all donations and gifts to Skeptoid Media, Inc. are tax deductible under section 501(c)(3) of the Internal Revenue Code (sections 170, 2055, 2106, 2522).

You can follow me at Twitter @steveproacnp for a daily dose of skeptical nursing.

Disclaimer: This post is my personal opinion, it is not a substitute for medical care. It is for informational purposes only. The information on Skeptoid blog is not intended nor recommended as a substitute for medical advice, diagnosis, or treatment. Always seek the advice of your own physician or other qualified health care professional regarding any medical questions or conditions. This post does not reflect the opinion of my partners, professional affiliates, or academic affiliations. I have no financial conflicts of interest to disclose.

About Stephen Propatier

Stephen Propatier is a board certified acute care nurse practitioner specializing in spine and sports medicine. He is a member of the Society for Science Based Medicine.
This entry was posted in Health, TV & Media and tagged , , , , . Bookmark the permalink.

8 Responses to Does Drinking White Wine Give You Melanoma?

  1. mudguts says:

    Thanx to this week we have decided not to wax, shave or drink bubbly…

    Yeah right!

    Gotta go.. The Douglas’ just arrived for arvo tea..

  2. Michael Forster says:

    I would think correlating the position of the melanoma as being under clothes would be equivalent to a direct correlation of total body exposure as you are commenting. As I understand the ‘causal’ mechanism for sun/melanoma it is by UV DNA damage, so this would point to (a) a different mechanism for melanoma (as noted), and that (b) the sun is not the causal mechanism.

    Why do you believe this is not sufficient as a control group, or in this situation, ‘control skin area’ which should be equivalent.

    • Mike a thoughtful question. There was no experiment or control group. This is retrospective study. Derived from data that is not individualized. Rather pools of data. So it is a statistical analysis of the totality. Only way you could use the self exposure as a control is if you correlated the data in individuals. The melanoma data is from many individuals who may have radical differences in sun exposure. As opposed to saying that a person gets the same amount of sun to head and neck and therefore meaningful. They are correlating two data pools alcohol exposure and melanoma. They found a statistical correlation in the data that suggest white wine correlates. It may be that if you correlated it to individuals you may see that the difference disappears. It might be that white wine correlates even more strongly with sun exposure time. Thereby explaining the statistical anomaly with a much more probable explanation. Namely that white wine drinkers may be in this data set taking more risks than other alcohol drinkers.

      • Michael Forster says:

        Well there were two ‘control’ groups in terms of the data analysis with respect to White wine. These are generated synthetically through the behavior data analysis. The individuals are ‘washed out’ due to the number of people used to generate these pools. Oh, yes, I know, meta studies contain biases due to the availability of databases on a world wide. I assume this is a north american pool with enough datapoints to wash out diet, genetics and pollution.

        a) the grouping of people that consumed no alcohol
        b) the grouping of people that consumed white wine
        c) the grouping of people that consumed ‘other’ alcoholic beverages.

        I use the meaning ‘control’ in a statistical sense, not in the double-blind sense, nor in the sense of applying a ‘factor’ (such as a drug) to an individual and see at what dose a meaningful response is got from the individual. that is a different sort of analysis anyway.

        Clearly I am missing what you are getting at by ‘individual’. When one uses metadata it is true there are correlations that are unknown. One believes (as I understand this), that pooling a large enough pool washes out correlations, in this case for alcohol source and behavior. In addition in the grouping that USES WHITE WINE (ie, a hidden correlation between white wine and something else), the SUB GROUP that is CONTROLLED FOR SUN, has MORE melanomas with the white wine than non-sun locations. This is a geographic location on the same body.

        It was unclear to in the abstract if the ‘sun shielded location’ melanoma had NO ‘Sun affected melanoma’ pair on the same body. This would put 2 melanomas on the same individual, but the abstract seemed to say (single) melanoma, so this detail may be meaningless.

        THIS is the ‘control group’ I am talking about (namely the sun shielded locations). It is hard to believe that there is a correlation between the location of melanomas and their frequency in the same grouping, that is somehow correlated to some other unknown factor, such as Vitamin D for instance. Yes, it would be true that the counts of melanoma underclothes SHOULD be less with the Vitamin D, but this would affect ONLY the frequency and still show up as a correlation.

        I certainly (in light of (pun not intended), don’t understand your sentence “… a person gets the same amount of sun … still are meaningful ” as we are talking about the parts of the body that DONT get the sun. I see 3 data pools (if we use sub-groups), not two, as this was a multi-variate analysis. I also see 3 results in terms of statistics.

        In particular I don’t see how one would correlate a study based upon individuals, as individuals are all in people pools of varying characteristics. Immediately one is pooling the individuals to get the numbers to make the statistics work.. I think I would understand if you put some words around the ‘individual’ in terms of a study design. I can’t see such a design, only pools of characteristics of behavior and body.

        Have patience.

        • Without addressing every point specifically once the full text is available the noted limitations related to sun exposure and specifics related to this data set is reviewed by the researcher. If I was not clear let me apologize. I cannot cut and paste on this site because it is not open source.
          To paraphrase. There was no data specific to location geographically and year round solar radiation levels. Since both of these factors have a large impact both related to level of clothing worn year round, exposure and radiation density, chronological age of exposure. Some of that data was either not correlated or unavailable.
          I am confused by the objection. Are you saying that a large enough group negates the significance of radiation exposure. Or are you suggesting that because the melanoma in a lower risk areas it then supports the assumption that a low risk area cannot be exposed to more solar radiation than a high risk area in any given cohort?
          I admit that on an individual that would most likely be true. Unless someone is wearing a ski mask in a speedo that makes sense. As a statistical group it was not controlled for, so that doesn’t necessarily mean that it is so. If your pulling data without regard to location or sun exposure, it doesn’t invalidate your finding but it makes the correlation weaker. Especially if you think about this temporally. If a significant percentage of the White Wine cohort had a tremendous childhood exposure to solar Rad. The red wine drinkers a much larger number had very little childhood exposure skewing the results.

  3. mudguts says:

    saving an inordinate rant.. pens and papers out, in the comparative cohorts;

    1) What is the increased incidence of melanoma amongst white wine drinkers vs non drinkers
    2) what is the increased incidence of melanoma amongst all drinkers v non drinkers
    c) Is this one of those bacon and pancreas reports?

  4. ScepticalScotty says:

    Cant resist it….despite breaking down in tears every time I hear this song…

Leave a Reply

Your email address will not be published. Required fields are marked *