For reprint orders, please contact: [email protected]

Kenneth J Mukamal Division of General Medicine & Primary Care, Beth Israel Deaconess Medical Center, 1309 Beacon Street, 2nd Floor, Brookline, MA 02446, USA n Tel.: +1 617 754 1401 n Fax: +1 617 754 1440 n [email protected]

Dr Mukamal is an Associate Professor of Medicine at Beth Israel Deaconess Medical Center and Harvard Medical School, MA, USA. Following medical school at the University of California, San Francisco, he completed a residency in internal medicine at Yale-New Haven Hospital and, after 3 years in academic practice, he took a general medicine fellowship and MPH from the Harvard School of Public Health. Dr Mukamal practices general internal medicine at Beth Israel Deaconess Medical Center, with a primary research focus on the relationship of lifestyle factors – particularly alcohol consumption – with the incidence and prognosis of cardiovascular disease. „„How did you become involved in the field of lifestyle factors & cardiovascular disease?

I have a practicing interest as a primary care provider in this field. For a long time, it has been very natural for people in my position to think not just about how to treat coronary heart disease (CHD), but also how to prevent it. Increasingly, over the last 10–15 years, an enormous part of what we do has been risk stratification/identification and treatment of risk factors. In part, that’s been driven by the increasing prevalence of coronary risk factors in places such as the USA and the UK, where the prevalence of diabetes, hypertension and obesity has really skyrocketed. Therefore, an enormous part of what we do these days is measuring these risk factors and trying to identify ways to ameliorate them.

“There were also hints that alcohol consumption may affect other elements of the coagulation cascade in ways that would lead to less thrombosis.” One of the first observations that intrigued me, and one that I haven’t necessarily gotten to fully explore, dates back to the early 1990s, where one of the consistent findings made concerning alcohol consumption was its relationship with various elements of the bleeding and clotting cascades. It was noted, for example, that ethanol can increase bleeding times above and beyond those resulting from aspirin administration, and it has antiplatelet effects in a variety of models. There were also hints that alcohol consumption 10.2217/FCA.09.11 © 2009 Future Medicine Ltd

may affect other elements of the coagulation cascade in ways that would lead to less thrombosis. For example, one of the strongest findings is the association of alcohol consumption and lower levels of fibrinogen, but there are certainly others as well. That’s the backdrop of what was known mechanistically at that time. As part of my training, we took care of a number of individuals with acute forms of CHD, whether that be myocardial infarction (MI) or unstable angina (what we would now call acute coronary syndromes). At some point, I had a ‘lightbulb’ moment, connecting my growing interests in where prevention and diet fits into clinical practice with the training I was getting in acute forms of coronary disease, to ask a question about whether there might be a role for alcohol consumption in the acute side of coronary disease. Put differently, if we think that alcohol consumption has antiplatelet effects and anticoagulant effects, and drugs such as aspirin have similar effects that we use in prevention and for acute MI, then why not alcohol too? The notion that we ask people who come to the emergency room with an acute MI to sip a glass of chardonnay along with the aspirin that we ask them to chew seems fairly preposterous. On the other hand, I had a peculiar kind of vision regarding the mechanisms at play, and to somebody early in his career trying to think out of the box, it seemed like a question that could be approached. That was the first really odd set of questions that I was interested in asking: is there an acute effect of alcohol consumption on cardiovascular risk and if so, how could we use that in acute c­oronary syndrome? Future Cardiol. (2009) 5(3), 219–225

Interview

Kenneth J Mukamal speaks to Christine Forder, Commissioning Editor

Future Cardiology

Alcohol and heart disease: where to next?

part of

ISSN 1479-6678

219

Interview

Mukamal

The problem, of course, is how to study that. We have not yet been able to answer that question, and I have not yet had the gumption to propose a randomized trial of patients presenting with MI, but it would be fun someday. „„What was the first indication that there was a link between alcohol consumption & cardiovascular risk?

Historically, this dates back quite a long way, a good 100 years at least. As far back as the early part of the last century it was observed that individuals that died from alcoholic cirrhosis seemed to have cleaner arteries than they ought to, based on their lifestyle habits. Some of that may have had something to do with their age – as they were dying at younger ages. However, the pathologists of the time felt that alcohol consumption was a cause of atherosclerosis, yet when they performed autopies, the exact opposite seemed to be the case. That was the first hint that there might be a link.

“...individuals who developed myocardial infarctions were more likely to be abstainers and less likely to be moderate drinkers compared with the controls.” More population-based observations date back approximately 30 years to when the Kaiser Permanente group [1] , which was directed by Arthur Klatsky, examined alcohol consumption reports amongst individuals who went on to have heart attacks, as well as amongst controls from the same population. This was based in a large health-maintenance organization in California, USA. They found, in an elegant way that has only been moderately improved upon in subsequent years, that individuals who developed MIs were more likely to be abstainers and less likely to be moderate drinkers compared with the controls. So, that was the beginning of the epidemi­ ology, and Framingham and other groups shortly confirmed these findings. Of course, Klatsky has continued to do a lot of work in this field in the subsequent 30 years. „„Could you briefly describe the mechanisms/relationship behind moderate alcohol consumption & lower coronary risk?

This is one area where we have seen some nice work over the last 30 years or so. Clearly, the best described mechanism for why moderate drinkers might have lower coronary risk involves HDLcholesterol. By all accounts, this is the strongest 220

Future Cardiol. (2009) 5(3)

mechanistically and pathophysiologically, and also the strongest statistically, in observational studies. Both in observational and in interventional studies, where we administer alcohol to individuals over several weeks, alcohol consumption appears to raise HDL-cholesterol in a dose–response manner, and it does so at least as strongly as any drug that we currently have available to us. In randomized trials, a dose of 30 g of alcohol per day (approximately two drinks) administered over several weeks raises levels of HDLcholesterol by approximately 4 mg/dl, or a little less than 10% depending on the population. It is noteworthy that we do not have particularly effective therapies to raise HDL-cholesterol, so this finding therefore represents a large effect, and there is also a strong relationship between alcohol consumption and HDL-cholesterol in population studies. If you combine the effect of alcohol on HDL-cholesterol with the fact that HDL-cholesterol is strongly associated with lower risk of heart disease, how much does HDL account for the relationship of alcohol and CHD? Statistically, it is approximately one-half to twothirds. Therefore, just knowing that alcohol raises HDL-cholesterol in human trials explains approximately half or more of the apparently lower risk of CHD amongst drinkers. There are a few other mechanisms that are more or less well developed. There is the anti­ coagulant/antiplatelet effect, which hasn’t been as well studied statistically, partly because there are few consistent measures of antiplatelet activity in different cohort studies, but there are clear data that alcohol lowers levels of fibrinogen, which contributes, to some degree, to how a­lcohol c­onsumption may relate to heart disease.

“...alcohol lowers levels of fibrinogen, which contributes, to some degree, to how alcohol consumption may relate to heart disease.” Insulin sensitivity is one of the really interesting relationships currently being studied. Although alcohol provides approximately 100 calories in a standard drink, there have now been a series of studies suggesting that alcohol consumption may improve insulin sensitivity. The largest of the studies was performed at the US Department of Agriculture and involved administering a placebo or two concentrations of alcohol for 8 weeks at a time to a group of women. It found that the higher the dose of alcohol that future science group

Alcohol & heart disease: where to next?

the women received, the better their insulin sensitivity 8 weeks later. Likewise, in observational studies, alcohol consumption is associated with a lower risk of diabetes. Therefore, we have both interesting mechanisms from experimental studies and consistent evidence from observational studies that suggest that alcohol consumption improves insulin sensitivity. There have also been two fairly long randomized studies of alcohol consumption in individuals with diabetes. One study administered alcohol for 3 months to one group and placebo to a second group. Another similar trial randomly recommended wine – but did not provide it – to a group of diabetics who had suffered heart attacks. In both cases, the group assigned to alcohol had better glycemic control. This interesting group of findings suggests that part of the mechanism by which alcohol reduces heart disease may be through improved insulin s­ensitivity and better glycemic control. „„ Do these positive effects outweigh any long-term effects & the possible link to cancer?

This is one of the most interesting things about studying alcohol. In studies of exercise or smoking, there are pretty much only beneficial effects for exercise (and the opposite for smoking); it is not very controversial to recommend that people exercise, whereas alcohol is very different.

“...recommended levels of what we might call ‘safe drinking’ are up to two drinks in a given day for a man and up to one drink per day for a nonpregnant woman.” Firstly, the cancer issue: there have been a number of reports concerning alcohol and cancer recently and I think it is important to take a step back and ask what we really know about moderate drinking and cancer. What we have known for close to 30 years is that alcohol consumption is associated with a higher risk of breast cancer. That is essentially a dose–response effect, so even moderate drinking does seem to lead to a higher risk of breast cancer. Data are less abundant but also suggest that alcohol raises the risk of head/neck and esophageal cancer, although those are much less common so the dose–response is less certain. For moderate drinking, that is virtually the whole cancer picture – those are the cancers for which we have fairly consistent data implicating moderate drinking, and of the two, breast cancer is the most important as it is so much more prevalent. future science group

Interview

For at least a few cancers, for example lymphoma and possibly thyroid cancer, there might even be a lower risk with low levels of drinking.

“...even moderate drinking does seem to lead to a higher risk of breast cancer.” When we weigh the effects of cancer versus heart disease, we also need to take into account how strong these effects are. If we look at how strong the effects of alcohol are on CHD, it is typical to see a 25% lower risk amongst drinkers compared with nondrinkers after adjustment for a variety of ways in which drinkers and non­drinkers differ. If one considers, for example, the recent data from the Million Women study [2] owing to its remarkable amount of attention, the higher risk amongst moderate drinkers of any type of cancer was 6%. For breast cancer, the increase in risk was approximately 13%, so it appeared that virtually all of the elevated cancer risk in moderate drinkers was related to breast cancer, probably partly owing to how women were recruited. For a woman thinking about drinking, she may be at high risk for breast cancer but at low risk of heart disease, which would be true for many young premenopausal women. In that case, the balance would be unfavorable for drinking, but for older women and for men, it is much less clear that the risk of cancer would outweigh the potential benefits of alcohol for heart disease since the magnitude of benefit for heart disease is great and because cardiovascular disease is still more common than cancer as a cause of death. Furthermore, the major increase in risk of cancer is based on breast cancer, which is less of an issue for men. It seems reasonable to say that for young people, the benefits do not outweigh the risk, and even more so for women than men, but for the majority of older women I think the data would suggest otherwise. „„Why does there appear to be a difference between men & women in alcohol consumption/heart disease?

Overall, women are more susceptible to the effects of alcohol at lower doses, so one of the big differences between men and women is dosage. In the USA, recommended levels of what we might call ‘safe drinking’ are up to two drinks in a given day for a man and up to one drink per day for a nonpregnant woman. Those differences reflect the fact that women respond to alcohol differently owing to smaller body size and differ­ ent metabolism, which accentuates the effects www.futuremedicine.com

221

Interview

Mukamal

of alcohol at lower doses. In most studies, the overall shape of the relationship between alcohol consumption and heart disease is similar in men and women, but the lower risk of heart disease associated with drinking begins at a lower dose amongst women. However, it also becomes harmful at a lower dose in women. The overall magnitude of the response seems to be relatively similar between the sexes, and so the difference is less about the relative benefit and more about the fact that women are generally at lower risk for heart disease than men, but at higher risk for the harms that alcohol consumption can cause. „„What are your feelings towards the current recommendations for alcohol consumption? Would it not be safer to recommend the consumption of no alcohol?

This can be answered in two ways. On the one hand, the current recommendations are pretty hard to improve upon because they indicate that nobody needs to drink. If it is safe for you to drink, based on a variety of personal characteristics, there is no limitation on the frequency with which one drinks, but there is a limitation on the quantity. The guidelines state this limit as up to two drinks in a given day for a man and up to one drink a day for a nonpregnant women. Importantly, they do not limit the number of days per week, just the number of drinks in any given day. I think this is consistent with current data. Most of the data concerning the harms of alcohol consumption seem to relate to the quantity that is consumed in a given day. In some cases, the risks of drinking alcohol relate to the overall volume of alcohol consumed and not just the quantity on a given day, but if we were able to keep everyone’s drinking within those guidelines, then the amount of harm that we see from alcohol consumption would be d­ramatically reduced.

“Individuals who drink on 3 to 4 days or more per week seem to be at lowest risk for heart disease” Would it be safer to recommend no alcohol? I suppose the answer is yes and no. I do think that for young individuals that is most certainly the case. For college students, for example, there is virtually no way that drinking could be beneficial and there are a number of ways in which it is demonstrably bad. It is easy to hypothesize that it would be safer to suggest that young individuals don’t drink alcohol at all. However, this is not so clear for older adults aged in their 50s or 222

Future Cardiol. (2009) 5(3)

60s where there are large numbers of studies that cumulatively suggest that alcohol consumption is not only associated with less heart disease but, as a consequence of that, also associated with a lower risk of mortality. Would abstention be safer for those individuals? It is possible, but it is also possible that it could be harmful. Therefore, the current guidelines give useful advice – do not drink at all if you cannot control your drinking or if you have contraindications – but if you do not have any of those contraindications, only drink within the guidelines. If we could actually get people to do that, we would be in a much better position. The hard part is achieving that balance. It is fair to note that the best evidence we have for reducing problem drinking tends to be for ways that reduce drinking altogether (e.g., higher taxation or reduced availability). These ways reduce problem drinking and also reduce drinking as a whole. As a primary care doctor, I do not know if it is truly possible to hit the proverbial sweet spot – to get the population to only drink a safe amount.

“...it’s not a specific type of beverage that is most important but rather the frequency that one drinks it.”

The tolerance level culturally for alcohol consumption is also different. Obviously, there has been benefit attributed to alcohol in countries such as France. The French paradox is that the French have consistently had lower rates of CHD that might have been predicted from other features of French citizens. Whether that is really attributable to red wine consumption among the French is a matter of debate, but their frequent red wine intake does reflect the strong cultural differences in alcohol consumption that may have a role in cross-national differences in heart and other diseases. „„What is more important in cardiovascular health: the frequency or the amount of alcohol consumed & why?

It depends on whether one considers the ‘good’ or ‘bad’ effects. In our work in both men and women, it is really the frequency of alcohol consumption that appears to be the most important factor for CHD. Individuals who drink on 3 to 4 days or more per week seem to be at lowest risk for heart disease, and the amount of alcohol consumed per drinking day does not appear to confer any additional benefit. Frequency trumps quantity, as it were. future science group

Alcohol & heart disease: where to next?

This may relate to short-term effects of a­lcohol, such as its effects on platelets. Since it is a shortlived effect, when alcohol is removed the effects go away relatively quickly, and repeated small doses may have the most benefit. A related issue is one of beverage type. Although there is a widespread belief that red wine is best for one’s heart, in our work that does not seem to be the case at all. For example, in our studies of large groups of men in the USA, beer and spirits seem to be most related to lower risk. I think that stems from the importance of drinking frequency, namely that it’s not a specific type of beverage that is most important but rather the frequency that one drinks it. Therefore, when studying Italian or French populations, red wine appears superior as that is what those individuals consume frequently. By contrast, when studied in the USA, beer and spirits are most frequently consumed and those are the beverages associated with lowest risk. „„What is the likelihood that randomized controlled trials of alcohol consumption will be performed to establish a direct link between alcohol consumption & reduction in coronary heart disease?

There have been at least two parallel-designed studies in diabetic adults. In one case, alcohol was administered for 3 months. Another one lasted 1 year, but did not actually provide alcohol; rather, red wine intake was randomly recommended.

“...there has been a longstanding notion that coffee must be bad for you, as if it had a sinful quality to it. In our work, that hasn’t panned out at all.”

Overall, there is beginning to be a sense that this may be possible. However, I’m still not sure it is feasible since the challenges for doing this are much greater than other randomized trials. There are all the issues of who to get involved – trying to minimize heterogeneity, finding people for whom it is safe to drink alcohol and are willing to drink alcohol everyday (if they are randomized to that arm) but also willing not to drink any alcohol outside of the study for potentially years on end. I suspect that many more people would chose not to enroll than would. It’s much more of a commitment than taking a pill every day. In essence, we need people in the middle range – those who drink sometimes and so wouldn’t mind being asked to drink every day, but don’t drink very regularly and would future science group

Interview

not be upset about not being able to drink. Finding people in that range is more challenging than one would expect. For example, in the USA, half of the people report no recent alcohol consumption at any given time. Furthermore, among drinkers, a sizable proportion are binge drinkers or report aspects of unhealthy drinking, who we would want to exclude, leaving few people in the midrange. „„How would you go about individualizing suggested alcohol intake in patients to help avoid cardiovascular disease?

It is important to think about the net risks and benefits of alcohol intake and to try to find individuals who would not come to harm from it. In general, I would not recommend alcohol intake in the young or in those who have a very low underlying risk of heart disease. Individuals with low levels of HDL-cholesterol may see the largest HDL increase owing to alcohol consumption. In a simple, practical way, HDL may be one way in which we could individualize – those who already have high levels will have less to gain from drinking than those who have low levels. Finally, a variety of studies have related the health effects of alcohol consumption – both good and bad – to folate intake, demonstrating that alcohol consumption benefits are maximized and the harmful effects minimized in those with plentiful folate intake. In the USA, we have folate fortification so that mean levels of folate intake have increased dramatically in recent years. As a result, in the USA it is not too difficult to ensure high folate intake, but where folate fortification doesn’t exist, that may be less true. To ensure correct individualization of alcohol consumption, you would need to ensure that individuals also have an adequate intake of folate. „„Which other lifestyle factors have you studied that play a role in cardiovascular disease?

We have performed a few studies rated to tea and coffee and heart disease – we have somehow managed to get stuck on beverages! Beverages are relatively easy to study because people remember them fairly well. It is generally easier to quantify someone’s tea or coffee intake than their broccoli intake, for example. We have looked particularly at tea and coffee as risk factors for both heart disease and prognosis amongst individuals with heart disease. I think that both are interesting areas of study. www.futuremedicine.com

223

Interview

Mukamal

Overall, there may be benefits to regular tea consumption, although this may require high doses at least relative to the amounts people consume in the USA. For example, we found no clear effect of tea on a variety of cardiovascular risk factors in a randomized trial of three cups per day. Although participants did not always consume the full three cups, that dose is pro­bably a minimum starting point for where the health benefits would be. On the other hand, in countries where much heavier levels of tea intake are not uncommon, such as in much of Asia, studies have been very interesting, often demonstrating lower risk of mortality and c­ardiovascular disease.

“...the relationship between coffee and heart disease may relate to what seems to be a substantially lower risk of diabetes associated with coffee drinking.” Coffee is interesting because there has been a longstanding notion that coffee must be bad for you, as if it had a sinful quality to it. In our work, that hasn’t panned out at all. Coffee is a relatively large source of antioxidants in the US diet, and that is at least as true in Europe where intake is even larger. Most studies suggest that coffee intake does not seem to have a major effect on the chances of getting heart disease. While some initial studies suggested that it might increase risk, nearly every study that we have seen over the last 5–10 years has suggested the opposite, namely that coffee does not seem to lead to higher risk. In fact, some studies have found that coffee drinkers have lower risk of mortality and, in some cases, heart disease. We have also seen this in survivors of a heart attack; prognosis seemed to be improved amongst individuals drinking larger amounts of coffee. An interesting area that is currently unfolding is the relationship of coffee consumption with diabetes; the relationship between coffee and heart disease may relate to what seems to be a substantially lower risk of diabetes associated with coffee drinking. „„What is now needed to further understand the area of lifestyle & cardiovascular disease? And where will your research be focused in the next 5 years?

Obviously, there has been an explosion of interest in genetics, and in many cases, I think we have a very different understanding now of how genetics may be playing a role in heart and other 224

Future Cardiol. (2009) 5(3)

chronic diseases. For example, we are now finding areas of the genome that are associated with heart disease that we really had no reason to suspect based on the genes there. As we learn more about how genetics affect heart disease, one of the really important questions is to what degree are the effects of these newly discovered genes modified by our environment. For example, could a gene that affects heart disease do so because it influences how alcohol raises HDL amongst drinkers? Are there genes that influence how much benefit we get from exercise, or are there genes that m­odulate who gets benefit from omega‑3 fatty acids? Hopefully, that will inform us about how our diets impact cardiovascular risk and the mechanisms involved – how do different dietary factors influence heart disease and what genes are involved in the downstream pathways? Frankly, it is an area where the methods have not been very well worked out yet, and it should be a c­hallenging area for the future. Another interesting area for heart disease is the ongoing question of whether we can conduct randomized trials for these many elements of lifestyle. We have focused randomized trials disproportionately on pharmaceutical agents that we can easily manipulate. There have been a few randomized trials of diet and prognosis amongst those with heart disease, and in some cases, they have had extraordinarily large effects, the Lyon study being a good example. Although these studies are always very difficult to do, they raise the possibility that if we can invest the time and effort to undertake more of them, we will see a much larger benefit than we have so far and maybe more than we expect. I hope this is at least one of the areas in which we will see some progress over the next few years.

“...the first step is to determine how successfully we can conduct randomized trials of alcohol consumption...” For our group in particular, we remain very interested in whether we can complete randomized trials of alcohol consumption. The jury is still out in regards to their feasibility, but the first step is to determine how successfully we can conduct randomized trials of alcohol consumption, not just over a week, but over months to years. If we can get people to complete such trials, then we can determine alcohol’s effects definitively, both on the risk of developing heart disease and on people who already have heart disease. future science group

Alcohol & heart disease: where to next?

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. No writing assistance was utilized in the p­roduction of this manuscript.

future science group

Websites 1.

The Permanente Medical Group. Physician home pages. www.permanente.net/homepage/index.html

2.

The Million Women Study: A Confidential National Study of Women’s Health. www.millionwomenstudy.org/introduction

www.futuremedicine.com

Interview

Affiliation n

Kenneth J Mukamal Division of General Medicine & Primary Care, Beth Israel Deaconess Medical Center, 1309 Beacon Street, 2nd Floor, Brookline, MA 02446, USA Tel.: +1 617 754 1401 Fax: +1 617 754 1440 [email protected]

225