My wife and I annually host Thanksgiving dinner for extended family and friends.  It’s a big affair and since we live in a fairly remote, albeit stunningly beautiful, part of northern Michigan, and our guests have to travel hundreds of miles to get here, the event stretches over multiple days.  One of my responsibilities is to ensure we have sufficient quantity and diversity of alcoholic and non-alcoholic beverages to appeal to all of our guests who span four generations and have varied tastes. Something that gives me particular angst each year is choosing the right wines to pair with the turkey feast and leftovers.  Perhaps you can relate.

In any event, I provide this as context, because you can imagine my discomfort when I opened the newspaper the other morning to see a headline warning that consuming even a “light” amount of alcohol can cause cancer.  Maybe you, too, read or heard about this latest report from medical experts.  What was your reaction?  Did it scare or concern you?  Or, did you simply shrug your shoulders and file it away as just another in a long line of seemingly conflicting messages and health advice from the medical community?  After all, many of us have been led to believe that an occasional drink, particularly red wine, was healthy for us. So what gives with this new report?

As a trained epidemiologist and risk assessor, who also enjoys imbibing now and then, I decided to dig into this story a little further to better understand the science behind it.  Below I’ll share what I learned in the hopes of helping you make better informed decisions about your own consumption of alcohol.  My personal conclusion is that I will continue to enjoy a glass of wine with dinner and the occasional cocktail with friends and serve them without guilt.  As in the past, I will refrain from and discourage overindulgence, not just to avoid small risks of cancer, but because it leads to so many other problems as well.  Of course, you should come to your own conclusions.

Recent ASCO Report on Alcohol and Cancer

The recent news stories on alcohol and cancer were the result of a November 7 statement issued by the American Society of Clinical Oncology (ASCO) which represents physicians who diagnose and treat cancer.  I can’t imagine that they were motivated by anything other than an altruistic desire to help people avoid this dreaded constellation of diseases.  Certainly a noble cause.

Prompted by my special interest in how the news media reports on science, one of the first things I did was search for and inventory the various news articles that reported on this recent ASCO statement.  What I observed was disappointing, but not surprising.  Some news outlets described the work of ASCO as a “new experiment”, others muddled and confused the definitions of “light” and “moderate” alcohol consumption, and still others opined that it is the first report to link alcohol and cancer.  All wrong and all misleading.  Such mistakes in reporting science are all too common, but in my experience they are often the fault of both the scientists and journalists.  This might serve as excellent fodder for a future blog post from me. So stay tuned.

I read the full text of the ACSO statement multiple times and found it to be thoroughly researched, and well-written.  It is not without its shortcomings as I’ll explain later on, but it does provide a compelling case that “moderate” and “heavy” alcohol consumption causes increased risk of some cancers with the evidence strongest for mouth, voice box and throat cancer, and cancers of the esophagus and liver.  The evidence is also compelling for smaller increases in risk for breast cancer and colon cancer.

The statement includes no new research; instead the article summarizes past studies and analyses, and principally relies on a 10-year old systematic review and meta-analyses of the scientific literature reported to that point in time.  This report has been continuously updated with subsequently reported studies but without change to the original conclusions.

Role of Epidemiology Studies in Systematic Reviews 

In a previous blog I chronicled the virtues and limitations of systematic reviews.  An obvious advantage is their improved transparency and objectivity in assessing the totality of relevant evidence applied to resolving important health questions.  However, they are still largely subjective exercises and as has been pointed out by others, various groups of experts examining the same evidence may arrive at different and even opposite conclusions.  Moreover, as they have thus far been applied to the exploration of evidence linking low level chemicals in the environment to various health effects, systematic reviews are ascribing far too much weight to just a few observational epidemiological studies with small sample sizes, and inconsistent results.  Incredibly, some of those reviews did not even consider whether there was animal toxicology or mechanistic evidence to establish biological plausibility to support the epidemiology findings.  By contrast, the systematic reviews and meta-analyses conducted on alcohol and cancer relied upon by ASCO for their statement were far more robustly conducted and were based on hundreds of epidemiology studies with aggregate sample sizes in the tens of thousands.  Moreover, those systematic reviews included and integrated the findings from literally hundreds of animal toxicology and mechanistic studies, thus strengthening their case for establishing causal relationships.  They also thoroughly considered the contributions of other risk factors for those cancer types, such as tobacco use, and the potential for them to confound or modify the effects of alcohol on cancer rates. 

Justification for the ASCO Statement

Given it has been more than 10 years since the scientific community concluded that alcohol causes some cancers, one might question why is ASCO speaking out now?  The report cites rising drinking rates in America as the principal reason.  Moreover, ASCO conducted a survey and found that only a third of Americans knew alcohol was a risk factor for some cancers, and that many practicing physicians were also unaware and, even if aware, were often reluctant to warn their patients.  The statement says "The importance of alcohol drinking as a contributing factor to the overall cancer burden is often under-appreciated.”  ACSO estimates that as many as 5.5% of cancers may be attributable to alcohol.

The ASCO statement also argues that people can be poor judges of how much they drink, so the safest course is to reduce consumption or not drink at all.  The ASCO statement concludes by calling for more research on alcohol and cancer links, for oncologists to pay more attention to alcohol consumption when treating patients (alcohol drinking can reduce the effectiveness of treatments and can cause second cancers).  ASCO’s public policy recommendations included increasing counseling for heavy drinkers; limiting the number of retail outlets for alcohol sales and their hours; keeping state-controlled liquor stores under government control; reducing youth exposure to alcohol ads; and, raising excise taxes on alcohol. "Because alcohol use is quite common, an initiative to address alcohol use (particularly high-risk alcohol use) is a potential preventative strategy to decrease the burden of cancer," reads the statement.  These are strong and aggressive policy recommendations that should cause policy makers and the public at large to take notice.

Light vs. Moderate Drinking

Despite the strengths of the scientific review that underpins the ASCO statement, there are some niggling yet important issues.  The first relates to the definition or lack thereof of “light” drinking and to the actual weight of evidence linking “light” drinking to an increased risk of cancer.  The ASCO statement never unambiguously defines “light” drinking, sometimes inferring it is anything less than “moderate” drinking and other times describing it as less than one drink per day without referencing the size and alcoholic content of the drinks.  The sole table they include to demonstrate dose-response relationships does not present very compelling evidence of increased risk associated with “light” drinking.  Instead it reveals very small elevated relative risks in the “light” drinking category (not defined at all in this instance) for only three of the six cancers examined (mouth and voice box, esophagus and breast), with only two of them barely reaching statistical significance. Hardly convincing.

  

Contributing to my unease is the fact that the original observational epidemiology studies on which the systematic reviews and meta-analyses are based, relied upon interviews with participants and thus self-reports of their consumption of alcoholic drinks.   Self-reporting is liable to underestimate consumption, sometimes grossly, because alcohol is widely known to be unhealthy and undesirable, and is sometimes drunk surreptitiously. Heavy and even moderate drinkers usually underestimate and underreport their consumption, as do drinkers of illegal or unregulated alcoholic drinks. None of the studies adjusted for such underreporting and thus likely ascribed higher risks to lower doses of alcohol than were actually consumed.  Those studies also estimated grams of alcohol consumed by multiplying the numbers of various types of alcoholic drinks reportedly consumed (e.g., beer, wine, spirits) by an estimated size and average alcohol content of each type of drink.  In recent years, the strength and serving size of some alcoholic drinks have increased which can lead to a further underestimation of dose.  Although the ASCO statement concluded that “Even modest use of alcohol may increase cancer risk, but the greatest risks are observed with heavy, long-term use”, and made no specific reference to “light” alcohol consumption, the body of the statement does discuss “light” consumption as a risk for breast cancer, and thus many news reports regrettably chose to trumpet that tenuous observation.

Relative Risk vs. Absolute Risk

A second problem with the ASCO report that has also adversely influenced news stories was its exclusive focus on relative risk and an absence of any discussion of absolute risk.  This has been eloquently discussed in op-eds written by Drs. Aaron Carroll and David Katz and in a blog post from Kevin Lomangino.  Absolute risk is critical. Small relative risk values, when consistent, are important when the number of people affected is large. A large relative risk of a rare type of cancer amounts to only a small absolute risk, which may reasonably be considered not significant, either by public health planners or by individuals assessing their own choices. By contrast, a small relative risk may amount to a large number of cases for a common type of cancer. For example, as Aaron Carroll pointed out, the ASCO statement implies that “if 1,667 40 year-old women became light drinkers, one additional person might develop breast cancer.  The other 1,666 would see no difference.”  This is not to imply that this increased risk isn’t important and shouldn’t be addressed with messages intended to prevent it, but rather that the magnitude of risk may be perceived by individuals much differently when they make personal decisions about whether to have an occasional drink of alcohol.

Possible Health Benefits

A third problem with the ASCO statement that was also addressed by Drs. Carroll and Katz in their op-eds is that it considered, but roundly dismissed any possible health benefits of consuming alcohol, including lower risks of certain other types of cancers, heart disease, diabetes and dementia.  Recent systematic reviews/meta-analyses conducted by others, have shown either no effect or even possible protective effects of “light” alcohol consumption and all cancers combined, pancreatic cancer , lung cancer, bladder cancer, gastric cancer, and kidney cancer.  The all cancers combined review and meta-analysis even reported no effect associated with “moderate” drinking.  I’m not suggesting that these findings are sufficiently compelling to endorse drinking to prevent cancer, but rather that the ASCO statement provides only a partial glimpse into what is actually a very complicated story.

The ASCO statement’s overwhelming focus on cancer and dismissal of studies showing possible benefits of alcoholic drinks was based on an argument that those results may have been tainted by non-drinkers who once drank heavily.  And while it is understandable that ASCO would focus on cancer, their report ignored other studies, including a recent one that looked at lifelong abstainers and former drinkers separately and found that moderate drinkers still experienced lower rates of heart disease.  It also ignores the findings of a randomized controlled trial (a study design that provides much more convincing evidence compared to the observational studies relied on by ASCO) which showed that light to moderate wine drinking can lead to a reduction in some risk factors for heart disease, diabetes and stroke.   As Dr. Katz points out, there are even established mechanisms for this, from elevations of High Density Lipoproteins, to elevations of a clot-dissolving substance called endogenous Tissue Plasminogen Activator, to the concentrated antioxidants like resveratrol in red wine.  Because the incidence of heart disease, diabetes and dementia far exceed the incidence of the several cancer types implicated, light to moderate drinking may confer greater net health benefits compared to harm.  

Drs. Katz and Carroll also remind us that occasional light to moderate drinking can provide pleasure and enjoyment, some quality of life enhancement and indirect health benefits in terms of reduced stress.  

I join Dr. Katz in reminding readers that the evidence of health benefits may not be sufficiently compelling to actually recommend drinking alcohol to prevent disease, but it does provide reassurance that the occasional drink may be worth the risks.

In conclusion, the ASCO statement is an important reminder that alcohol carries cancer risks; people need to know that, but they also need further information such as absolute risks and possible benefits to put the risk into context so they can make informed personal decisions.  

So, I for one will be enjoying a glass of wine with my Thanksgiving dinner and will be serving our guests without guilt, while also strongly discouraging over-indulgence.

Cheers and best wishes to you and yours for the holidays!