How Do We Know if Alcohol Is Bad for Us?

There’s a Japanese proverb that says: “The reverse side also has a reverse side.” And sometimes, it feels appropriate in medicine.

NEWS: If you are a drinker, we hate to be the ones to break some sobering news to you. But a major new study says no amount of alcohol is good for your overall health.

NEWS: Exercise or alcohol? When it comes to living a long life, booze may actually help you live longer than hitting the treadmill.

NEWS: A team of international researchers found the increased risks to cancer and heart disease from eating red meat are small and uncertain

NEWS: Eating a cheeseburger could be as bad for you as smoking cigarettes

When people hear conflicting headlines, it makes it hard to trust anything that doctors say. What do we really know, after all? Should you eliminate meat from your diet? Does exercise help you live longer? Is it safe to consume moderate amounts of alcohol? These are actually deceptively difficult questions, in part because it can be hard to do the kind of solid research that’s required to answer them well.

Well, what if there was a better way to get some answers?

From the Freakonomics Radio Network, this is Freakonomics, M.D.

I’m Bapu Jena. I’m a medical doctor and an economist. Each episode, I dissect an interesting question at the sweet spot between health and economics.

Today on the show, we’re going to look at how a research method you’ve probably never heard of can help us figure out how to prevent and treat big health problems, like cancer and heart disease.

SMITH: To try to take that level of certainty, you need to look across the whole field of evidence. I’m interested in the problem of actually identifying things which do cause disease, which genuinely influence disease risk

And then, how do we best use this information?

KATHIRESAN: It’s like surgery without a scalpel.

Americans like to drink alcohol. Surveys have shown that half of the adults in this country regularly consume it. Though, I’ll be honest, I mostly like mocktails. Anyway, we know alcohol can make us feel good in the moment, which is why so many people enjoy it. And we also know that heavy alcohol use can cause serious health problems, like liver disease and addiction. But there’s this huge gray area when it comes to moderate alcohol use. Is it good for us? Or is it bad for us? Here, a lot of the research linking alcohol to certain health benefits or health problems is associational, meaning it seems like alcohol is responsible, but it could also be something else that we can’t measure. So, how do we figure out what alcohol does to our bodies? It turns out there’s a group of people who might offer some clues.

PETERSON: I think alcohol generally makes people feel good, but it doesn’t make me feel good. It makes me feel like I want to lie down and take a long nap.

That’s Lucas Kwan Peterson. He’s a food columnist at the Los Angeles Times.

PETERSON: When I drink, I feel immediate heat in the face, and then it kind of, uh, translates also into a difficulty breathing, almost like there’s a cat in the room. So, I’m allergic to cats and it’s a similar feeling where there’s kind of like a shortness of breath. Sometimes it can come with like a headache

Lucas’s mom is Chinese. And like her, and some of his cousins, and about 560 million other people of East Asian descent, he experiences a flushing effect when he drinks alcohol. This phenomenon is rarely, if ever, experienced by any other group of people. It’s often called “Asian flush,” though Lucas prefers the term “Asian glow.” Whatever you call it, the flushing effect and its associated symptoms make Lucas feel bad.

Lucas, and lots of other people of East Asian heritage, have a genetic variant that lowers their ability to tolerate alcohol. Because of their genes, their bodies don’t detoxify alcohol appropriately, so when they drink, a toxin called acetaldehyde builds up in their blood. Acetaldehyde causes those unpleasant effects, which are nearly impossible to hide.

PETERSON: I definitely remember being at a house party and we were drinking some horrible like malt beverage and turning totally flushed — face hot —and then people pointing that out to me and telling me that my face, uh, looked like a tomato and then going into the bathroom and seeing my face and, and noticing just the splotchiness all over and thinking like, “Oh my God, what’s going on?”

Because Asian flush causes such physical discomfort and embarrassment, the people who carry this genetic variant tend to consume less alcohol than the general population. And that actually presents a rare opportunity in medical research.

It’s a little tricky, but here’s how it works. You’ve heard me talk before about randomized controlled trials. These trials are considered the gold standard of research because they directly compare outcomes between two groups of people. One group is randomly assigned to get a treatment, and the other isn’t. The whole idea behind randomization is that other factors that might affect the outcome, what we call confounders, should be evenly distributed between the two groups in a randomized trial.

It’s a fantastic approach, but it’s also expensive. Randomized trials involve recruiting lots of participants, and sometimes following them up over a long period of time. For this reason, they’re not done as often as they should be — which can have consequences.

SMITH: Those vitamin E papers were presented in such a way that TheNew York Times has its top headline, front page headline, that vitamin E slashed heart disease risk. You know, so people start taking the tablet, the supplements, obviously, because it’s presented in that way. So, that was a big waste of time.

That’s professor George Davey Smith.

SMITH: I’m the director of the medical research council integrative epidemiology unit at the University of Bristol. I’m an epidemiologist, which means studying the causes of disease, trying to identify causes of disease, so that you can prevent them.

The big waste of time George mentions — his words, not mine, by the way — refers to

research published in 1993 in the New England Journal of Medicine. The study, which was not randomized, initially suggested that vitamin E could reduce the risk of cardiovascular disease.

SMITH: And when the randomized trials were done and there were a large number of them were done with very long follow-up, uh, there was absolutely no beneficial effect of taking vitamin E supplements, uh, compared to not taking them.

The initial vitamin E study — the one that was debunked — initially suggested that alcohol could lower the risk of coronary artery disease.

SMITH: This became really popular story because, you know, where do you get this sort of good news, research story? Of course, it gets wide publicity.

George wanted to do his own follow-up research to look at how alcohol may impact the heart. But that’s a situation where you can’t set up a randomized trial. You can randomly assign one group of test subjects to take vitamin E every day, and another group not to. But you can’t easily assign a random group of people to drink a lot of alcohol and another group not to. It’s unethical and impractical. But it would be really useful to have two random groups like that because you could learn a lot.

That’s where Asian flush comes in. Because you can take a group of people whose genes have randomly assigned them to be especially sensitive to alcohol — and therefore likely to drink less of it. And researchers can look at that group, and compare them to other people who don’t have the Asian flush gene. It’s a natural experiment of sorts, and it’s also a research method known as Mendelian randomization.

Professor George Davey Smith happens to be an expert in it. We asked him how he’d explain Mendelian randomization to someone who didn’t know anything about it, like a seven-year-old.

SMITH: So — I wouldn’t try explaining it to a seven-year-old. I’d find a different job.

I can’t say that I blame him. Anyway, Mendelian randomization is named for Gregor Mendel, who is widely considered the founder of modern genetics. It’s a pretty reliable way to draw conclusions about the causes of particular health problems or the effects of specific habits, like smoking, and diet. Or drinking alcohol. In his own research on alcohol and heart disease, George and his colleagues used Mendelian randomization to analyze data from a group of people — like Lucas Kwan Peterson — whose genetic sensitivity to alcohol tended to dictate their drinking habits.

SMITH: So you get groups of people, who, for all of the factors you can see are similar, but they drink very different amounts due to their genetic makeup, and using that method, demonstrates that sadly, there appears to be no beneficial effect on coronary heart disease of alcohol consumption, and a substantial adverse effect on overall cardiovascular disease. So, the observational association suggesting benefit are simply not demonstrated in the Mendelian randomization studies.

The key idea in Mendelian randomization studies is that groups of people differ only because of certain genetic variants. These variants predispose them to higher or lower levels of exposure to a risk factor, in this case alcohol. Otherwise, these groups are similar. So, the presence of one gene variant may randomly make one group of people drink very little and another group randomly drink more. But factors like race, age, other medical conditions, where they live, what kinds of work they do — those should be similar between the two groups for the assumptions of Mendelian randomization to hold.

Of course, alcohol affects more than just cardiovascular disease. Cancer has also been strongly linked to alcohol intake, and a group of researchers from China and England recently looked at whether this link may be more than just an association. Could they demonstrate that alcohol is a direct cause of cancer? Again, we know they’re correlated — but maybe those heavy drinkers have other unhealthy behaviors, too. Maybe they smoke more, they exercise less — and those confounders, not all of which we can account for, are what causes the cancer.

That’s where Mendelian randomization comes in. Here’s what the researchers did for this new study. They analyzed D.N.A. samples from approximately 150,000 people in China, and also had participants fill out questionnaires about alcohol use. For cultural reasons, women are much less likely to consume alcohol in China, so the study focused on men. Men with the Asian flush gene were less likely to drink alcohol, and less likely to develop any cancer — in particular, they were 31 percent less likely to be diagnosed with some cancers, like esophageal cancer, that have been linked to alcohol use in associational studies.

The study was published in January of 2022 in the International Journal of Cancer. Applying Mendelian randomization allowed researchers to reliably determine that it was alcohol intake, and not other factors — like diet or smoking — that caused elevated cancer rates in these men. Of course, this kind of randomization is complex, and demands that researchers use it well and properly.

George argues that for research to be truly reliable, though, it has to go even further.

SMITH: To try to obtain that level of certainty, you need to, you need to look across the whole field of evidence that is available from all different sources and put that together. Then that is what strengthens your belief that something is causal. So, in the case of cholesterol and coronary heart disease, we have Mendelian randomization studies. We have randomized control trials. We have animal studies. They all point in the same direction. There’s a different story with the genes. There’s a different story with the drugs. There’s a different story with the animal studies. Each one of them could be biased. The fact that one’s biased would not influence the result you would obtain from the other study.

This approach George describes is called triangulation of evidence. You draw on three different types of research — like points of a triangle — to strengthen the results of each. It sounds logical, sure. It’s time-consuming though, and in most cases requires a lot of resources to answer one question. All of this discussion of how to get better answers — it does make you wonder why there isn’t more good research. George has a theory.

SMITH: People have very strong attachments to their studies or study designs. You know, if you have a hammer, everything appears to be a nail. So, if you’ve got a vast cohort study, which you’ve set up to investigate dietary influences on health, then the fact that what you find actually had, doesn’t turned out to be very useful, you know, isn’t a welcome message. People are very strongly committed to hypotheses that they’ve put forward or notions they’ve put forward and they’re very committed to the studies that they’ve set up.

Researchers, including me, become attached to their work, like anyone else. Except, in medicine, the stakes of publishing subpar research are considerably higher. If you tell people that something like a supplement, that they can easily buy in a drugstore, will dramatically reduce their risk of heart disease, they’re likely to try it, which can lead to other unintended effects.

SMITH: People doing something which they think might be beneficial to reducing heart disease risk might make them focus less on doing things like stopping smoking and losing weight and reducing their cholesterol levels.

In other words, thinks that actually lower the chances they’ll develop heart disease. Randomized controlled trials are powerful, and their impacts on health can be wide and long-lasting. Which is why nature’s very own randomization offers so many opportunities for discovery and advancement. George Davey Smith isn’t the only one who sees its potential. So, how do we use the information Mendelian randomization gives us to prevent and treat diseases? That’s coming up, after the break.

*          *          *

At the end of every show, I always ask you to share your thoughts. Well, this episode was the result of two bright, young listeners — Tynan Friend and Hunter Chang — who emailed me a crazy idea about how Asian flush could be a natural experiment to study the impact of alcohol on our health. I got really excited, only to find out that others had already looked into this exact same question using the tools of Mendelian randomization, which, by the way, itself borrows from a statistical tool that economists use quite often, called instrumental variables. But the studies and the approach were so cool that I wanted to bring them to you. On that note, back to the show.

When the Human Genome Project was completed in April of 2003, it provided the world with a map of all human genes — and offered medicine incredible opportunities to advance. Without this elaborate sequencing, it would be a lot more difficult to conduct studies using Mendelian randomization, and to answer some of the most critical questions about health and disease.

One of those is heart disease, which, like cancer, is an enormous burden in the United States, and around the world. One person dies every 36 seconds from cardiovascular-related issues in the U.S., and more than 18 million adults in this country live with coronary artery disease. Our next guest knows this all too well.

KATHIRESAN: My name is Sekar Kathiresan. I’m a C.E.O. of a biotech company called Verve Therapeutics. Prior to this role, I was a professor of medicine at Harvard Medical School. Ran a research lab that spanned, uh, Mass General Hospital and the Broad Institute, studying the genetic basis for heart attack. I have, uh, a strong family history of heart disease. And as part of the reason, I decided to go into cardiology — my father’s brother passed away at 42, uh, of an M.I. And my dad, when I was a resident, um, had a bypass surgery at the age of 54.So, all this family history, um, kind of got me into cardiology, And then my brother in 2012, living in Pittsburgh had gone for a run, uh, came back, um, to his house was dizzy, collapsed at home, um, and unfortunately suffered a heart attack and had a fair amount of brain injury during the resuscitation at home, uh, by the E.M.T. and never, uh, survived. He was hospitalized and died 10 days into the hospital course, uh, due to anoxic brain injury. So, he was 42, and a very sad, dramatic event. And that really led me to think hard about —we’ve done all this research. We’ve found these important insights. How can we use this information to avert similar tragedies?

At the time of his brother’s death, Sekar — who goes by Sek — was already studying the genetic underpinnings of heart disease and risk for heart attack. A lot of the research in this area has looked at associations, like the Vitamin E study I discussed earlier. But figuring out the causes of heart disease can be difficult because of the sheer number of variables that can influence someone’s risk for it. Through his research using genetics, Sek has been able to overturn long held beliefs about one of those variables — cholesterol.

Let me walk you through it. Historically, L.D.L., or low-density lipoprotein, has been known as the bad cholesterol. For a while, H.D.L., or high-density lipoprotein, enjoyed a somewhat better reputation. I talked to Sek about his work with L.D.L. and H.D.L.

JENA: So, L.D.L. now is well-established as the bad cholesterol. You’ve done some work on H.D.L. as well. Can you tell me a little bit about that?

KATHIRESAN: So, this was the research finding that was most surprising. When I was in medical school, we knew that high H.D.L. correlated in the population with lower risk of heart attack. So, I was taught that H.D.L. was the good cholesterol and that anything that raised your H.D.L. cholesterol must be good for you. That’s what was taught, but it wasn’t actually clear whether that was a true statement. Is it just correlation or is there a cause and effect relationship? And as you know, based on the observational epidemiology, of higher H.D.L. correlated with lower risk of heart attack, many drug companies assumed that if they could develop a medicine that raises that H.D.L. cholesterol, it would be good for you — that it would lower risk of heart attack. So, there are many companies back in 2005, 2006, 2007 — that were developing medicines that raised H.D.L. cholesterol, with the expectation that they would lower risk of heart attacks. We decided to look at this question and cause and effect question using human genetics,

Their genetic analysis suggested that around 3 percent of people had a particular mutation that raised their H.D.L. cholesterol. They used Mendelian randomization to figure out the rest.

KATHIRESAN: So, we said, “Okay. If higher level of H.D.L. cholesterol was causally protective for heart disease — for a heart attack, then the 3 percent of people that carry that mutation should have lower risk of heart attack. And to our surprise Bapu, the individuals who had the H.D.L. raising mutation had the same risk of heart attack as those who didn’t carry the mutation.

JENA: In this scenario here, all the other things that we might expect are different between people who have high and low H.D.L. in general, those are basically constant — held constant because of the randomization.

KATHIRESAN: Exactly. Everything is the same. All the measured variables were the same, except the H.D.L.. So, this was a shock to us, and it actually took us, Bapu, four years to publish this finding, because the whole field thought that H.D.L. was causally protective.

Sek’s research on H.D.L. was published in The Lancet in 2012, and it was a game changer. Research on drugs that would raise H.D.L. was halted. Sek’s approach showed the power and the potential for genetics — and the randomization it permits — to help us to better, more accurate work. Sek hasn’t stopped there, though. He and other researchers are also using genes to offer insights that are otherwise nearly impossible to achieve. And they’re trying to determine if changing certain genes might prevent heart disease — or other conditions — before they start. With heart attacks, that means targeting the cholesterol component that should never be ignored: L.D.L.

KATHIRESAN: There are two major insights that we got from the human genetics, uh, over the years. One is that if one’s L.D.L. cholesterol is low, lifelong, it’s really hard to get a heart attack. The second insight that we got was that there are some people walking around who naturally have their L.D.L. cholesterol very low, lifelong, because they have a cholesterol raising gene switched off. These are so-called human knockouts. These are people who completely have a cholesterol-raising gene turned off. And these are very special folks. Uh, it took a while to kind of identify them. And what we learned from them is that, first of all, it’s feasible to turn off a gene and be healthy and have very low L.D.L. And then we learned that these people are resistant to heart attack. So, this gave us the idea, Bapu that if, if we could develop a medicine that would mimic that natural resistance mutation to these people carry and switch off at cholesterol-raising gene, then this could be a definitive treatment for heart attack.

Sek’s company, Verve Therapeutics, is developing what he thinks of as a “one-and-done” treatment for heart disease that works by changing one single letter in the human genome from an A to a G.

KATHIRESAN: And the way it works is that it essentially makes a, the medicine makes a single spelling change in the DNA of the liver and turns off a cholesterol raising gene, permanently. We’ve gotten this whole approach of a one and done treatment to work in monkeys. So, a one-time intravenous infusion of the drug. the cholesterol raising gene is switched off. The L.D.L. cholesterol comes down by 60 to 70 percent at two weeks. And then now about a year and a half later, the L.D.L. cholesterol is still down 60 to 70 percent, and we expect this to be a treatment durable for the lifetime of the person.

JENA: How long does it take for the encoding to occur? So, if you infuse, you know — if you infuse a monkey with the medicine, the DNA changes happen within hours? Days? What?

KATHIRESAN: Yeah, it’s remarkable. So, it’s a one-time infusion, Bapu, into the bloodstream. All of the stuff goes to the liver. All of the editing — in essentially every liver cell — is done in the first 48 hours. And then all of the drug components are gone from the body within the first few days. And then you’re just left with the clinical consequence of that, which is lowering the blood L.D.L. cholesterol permanently.

JENA: Would you think of this as a therapeutic or sort of a preventative medicine?

KATHIRESAN: Both.

Sek’s work focuses on heart disease — an interest that was sparked because of the inherited risk in his own family. His research using genetics has led him to seek treatments that change people’s genes. It sounds pretty sci-fi, and in a way, it is. But the potential for this gene-editing approach — if done right — is sort of limitless.

KATHIRESAN: It’s like surgery without a scalpel. The kind of human genetics work that we did for coronary heart disease — for heart attack — very similar work is being done for a range of other diseases. You name it: diabetes, Alzheimer’s, C.O.P.D., kidney disease. The genetic underpinnings of all those diseases have been decoded. So, the Mendelian randomization approach has been in is being applied, to try to understand, which biomarkers, for example, cause Alzheimer’s or cause kidney disease. And so, it would be an appropriate target for therapy. And then the final piece of this, is that, can you directly intervene at the D.N.A. level like we’re trying to do for other diseases? And the answer there also is yes.

It’s hard to know what the future holds. For a treatment that’s safe and effective, many people would probably jump at the chance to lower or eliminate their risk for conditions like Alzheimer’s, heart disease, and diabetes. But this kind of treatment changes someone’s genetics — D.N.A.. Would you be comfortable with that?

In the meantime, what about other, less dangerous, but still sort of annoying, genetic quirks — like Asian flush? This mutation, and others, might incidentally help some people avoid certain problems. Here’s Lucas Kwan Peterson again.

PETERSON: I have wished that I was different, but on the other hand, I think about possible health problems that I’ve dodged. I think about other possible situations I’ve dodged or stupid things that I could have done when I was drunk. I think about money that I’ve saved — tens of thousands of dollars over the years on alcohol. It’s just sort of a message like, “Hey, your body can’t handle this. You shouldn’t be doing this.” I’ve ultimately tried to put a positive spin on it and think, “Okay, well maybe this is just kind of the universe’s way of trying to protect me.”

I’m not sure we’ll have any randomized controlled trials involving the universe any time soon, but I guess you never know.

Look, at its core, a lot of medical research comes down to what we know — what we can prove — about the definitive causes of disease. Because if you know what causes a disease, that gives you a better chance of figuring out how to treat it — including by modifying certain behaviors, like drinking alcohol. Mendelian randomization helps us get one step closer, to go beyond associations, and to get to a level of certainty that we couldn’t really have had before the early 2000s, when the human genome was fully sequenced.

But that was nearly 20 years ago. So, why don’t we have even more solid research steeped in genetics?

In the end, I think it’s a question of economics. We know certain medications work, or don’t, because of randomized controlled trials. A lot of those trials occur because of funding from companies with a financial interest in knowing the outcome, one way or the other. The same isn’t true about things other than drugs that can affect our health — like what we eat, what we drink, or how much we exercise. We might be really interested in knowing how alcohol impacts our risk for certain health problems, but most companies wouldn’t have a reason to study that issue. These are public health problems with not a lot of incentive for private investment.

There’s a lot we still don’t know in medicine and a lot we may never know. The way we design studies, though, matters. It can help us come closer to the answers we crave, and can have real impacts on people’s lives. We have to make sure to do it better, and get it right.

That’s it for today’s show. If you want to hear more about the research I talked about today, that’s at freakonomics.com. Also, let me know what you thought about this episode. I’m at bapu@freakonomics.com. Thanks for listening.

Freakonomics, M.D. is part of the Freakonomics Radio Network, which also includes Freakonomics Radio, No Stupid Questions, and People I (Mostly) Admire. All our shows are produced by Stitcher and Renbud Radio. You can find us on Twitter and Instagram at @drbapupod. Original music composed by Luis Guerra. This episode was produced by Julie Kanfer and mixed by Eleanor Osborne. Our staff also includes Alison Craiglow, Gabriel Roth, Greg Rippin, Rebecca Lee Douglas, Morgan Levey, Zack Lapinski, Mary Diduch, Ryan Kelley, Jasmin Klinger, Emma Tyrrell, Lyric Bowditch, Jacob Clemente, Alina Kulman, and Stephen Dubner. If you like this show, or any other show in the Freakonomics Radio Network, please recommend it to your family and friends. That’s the best way to support the podcasts you love. As always, thanks for listening.

JENA: I’ll tell you this. If you can package that mR.N.A. into a tub of Ben and Jerry’s, sign me up, okay? I will be good to go, and I’ll watch the L.D.L. melt away. You have no idea how much compensating behavior I will take part in

KATHIRESAN: There is a moral hazard we’ve been told about.