Professor Marcus Munafo
Marcus Munafo is professor of Biological Psychology at the University of Bristol. The bulk of his research focuses on understanding pathways into, and the subsequent consequences of, health and mental health. He has particular focus on addictive substances, primarily tobacco and alcohol use.
No single study or article is going to be authoritative
On becoming interested in science
Well, how I got to where I am is a matter more of chance than anything else. So I did science A levels - I did maths, physics, chemistry - and then I wanted to do something slightly different, so I wanted to do philosophy as a degree. But I was applying to Oxford and, at the time certainly, you couldn’t do philosophy on its own, you had to do it with something, so I thought, well, psychology looks interesting. So I did psychology and philosophy there. But by the time I’d finished my degree I’d become very interested in psychology, and so I kind of wanted to keep doing that, not because I wanted to become a psychologist or become an academic, but just because I was still enjoying the subject.
So I did a Masters and that then became a PhD, and so I found myself on an academic career path, if you like. And, actually, I had this good fortune to go from a very experimental first degree to a much more applied MSc and PhD where I was looking at the relationship between anxiety and recovery from surgery and trying to bring some of the laboratory based experimental methods in to a much more applied setting. Then I went to do a postdoc back at Oxford in the Department of Public Health and Primary Care working in a very medical setting and from there moved to the Department of Clinical Pharmacology working on the pharmacogenetics of smoking behaviour. So I was being exposed to all of these different fields, all of these different disciplines, different ways of doing things within biomedical science more broadly. I then spent a short time at the University of Pennsylvania in psychiatry and then moved back to the UK and took up my lectureship here in Bristol, back in experimental psychology. So I’ve worked across a range of different departments and that’s given me a range of different perspectives and what I’ve tried to do is bring the best of each of those disciplines or perspectives into the work that we do here.
On current research
So, my group’s the Tobacco and Alcohol Research Group, which is fairly self-explanatory except that, actually, we do work well beyond that relatively narrow remit. We are interested in lifestyle or health behaviours generally, so tobacco and alcohol use are a core part of that, but we’re becoming increasingly interested in things like caffeine use and dietary behaviours, exercise and so on. And we’re also interested in the links between those behaviours, and not just physical health outcomes but also mental health outcomes. So mental health has become an increasing part of what we do.
So my group’s structured into what I like to think of as a kind of pipeline, if you like. We start with epidemiological work. We have postdocs that focus on just looking at the relationships between X and Y, smoking and mental health outcomes for example, and those methods use genetic information to strengthen the causal claims that we can extract from our data, if you like. So the idea is that we try to, by triangulating evidence from different methods, be more confident that there is a causal relationship between, say, smoking and mental health outcome. Once we’ve identified what we think looks like a causal relationship, we move that along the pipeline into mechanistic studies. We bring people into the laboratory, we look at what happens to them when you put them under stress or give them different pharmacological challenges and look at the impact that that has on their behaviour, or we do that in the fMRI scanner to look at the impact that that has on their brain functioning.
And then once we’ve identified what looks like a plausible mechanism for what we think is a causal pathway, we then have a target that we can take further along that pipeline into the intervention development and try to change those mechanisms or develop interventions that help people to stop smoking, or reduce the impact of alcohol consumption or whatever it might be. So the idea is that the different parts of my research group speak to each other and move those ideas along that pipeline, if possible.
On the possible links between psychotic disorders and tobacco use
So I guess the first thing is that we try to be very clear that no single study or article is going to be authoritative, or that’s going to be true in very few cases. For the most part each piece of work that we conduct is just a piece of a jigsaw and, you know, sometimes the piece goes in the wrong place and you need to take it out again. Science isn’t a sort of steady, linear progression towards the sunlit uplands of truth, it’s quite a messy process; you take the wrong step occasionally, you need to retrace your steps, and so on. So we try to be as open and transparent about that as possible.
The smoking and psychosis schizophrenia question is a really interesting one. One of the methods that we use at the epidemiological end of the pipeline is a method developed in Bristol by George Davey Smith, principally, which uses genetic information to strengthen causal inference in observational data. It’s a technique called Mendelian Randomisation. So if we take the example of smoking and schizophrenia, we know that those two are highly correlated: people who smoke are at high risk of schizophrenia diagnosis and people with a schizophrenia diagnosis are more likely to smoke. It could be that smoking is a risk factor for schizophrenia, it could be that people with schizophrenia are more likely to smoke because they’re self-medicating their symptoms or the side effect of their medication. That’s been the generally accepted model of why that relationship exists. Or it could be that there’s a third factor that independently influences both and there’s no causal relationship directly between the two. And unpicking that is notoriously difficult in observational data.
But if you have genetic information the advantages there are that first of all, your germline DNA - the DNA sequence you inherit - can’t be changed by your environment, so it can’t be subject to reverse causality. And because of the way that your genotype is created through the mixing of the genes that you inherit from your mother and father at conception, the specific genotype that you were born with is essentially allocated to you at random. It shouldn’t depend on where you were born or what time of the year you’re born in or anything like that. So you can use genetic information as a proxy for something like smoking, if smoking is the thing that you’re interested in. So it’s not about genetics directly, it’s using genetic information to indirectly capture your exposure of interest, but do so in a way that’s not subject to reverse causality and not subject to confounding, and then use that to see if smoking is causally related to a range of different health outcomes.
We’ve done that as a sort of proof of principle to show that smoking does indeed make you thinner, that’s it’s causally related to reduced body mass index. And if you look at genetic studies of diseases that we know are related to smoking, for example lung cancer, you find that some of the genes that are identified in that genetic study of lung cancer are genes that influence how much you smoke. So those aren’t genes that are directly influencing whether or not you develop lung cancer, they’re genes that are influencing whether or not you smoke and smoking is what’s causing lung cancer.
So when this large genetic study of schizophrenia came out a couple of years ago, one of the findings that we noticed that was relatively unheralded was that the same genes that we see in that genetic study of lung cancer, we also see in this genetic study of schizophrenia. One possibility is that those genes are having a direct effect on your risk of schizophrenia as well as an impact on your likelihood of smoking. The genes encode nicotinic acetylcholine receptors which are obviously relevant, plausible candidates for smoking, but nicotinic receptors also influence cognition and other mechanisms that might be relevant to schizophrenia. But it’s suggestive, it’s an interesting insight that that genetic study has provided for us, and I think the key thing is that people think of genetic studies as telling us about biology. And, of course, they do, but they also potentially tell us about these other behavioural or environmental risk factors and those might be much more amenable to intervention. Because if you do find good evidence that smoking is a risk factor for schizophrenia then you can do something about that much more directly than by learning about the underlying biology that contributes to risk of schizophrenia.
Around the same time that that genetic study came out there were a couple of other studies. A meta-analysis led by Robin Morris’ group at the Institute of Psychiatry in London and also a study by Ken Kendler using a Swedish data set using different methods that arrived at the same conclusion that maybe smoking is a risk factor for schizophrenia. So it’s not a completely done deal yet, but it certainly is a suggestive finding that I think is worth exploring further.
On the social responsibility of presenting evidence
I mean, ultimately our job is to do the research and to produce findings that we feel comfortable with, that we’re confident about, that we think stand up, and that’s our core business. But we do place a great deal of importance on the need to engage with the public, with likely end users of our research, whether those be people with mental health problems or politicians or whoever it might be…healthcare professionals. And there are a few reasons why we want to do that, I mean, one is all of our research is ultimately publicly funded, whether it be through taxation or through charitable donations. So we have a responsibility to tell those who ultimately fund our research what we’re up to and what we’re finding. There’s also a value in articulating our work to a general audience, if you like, so that we don’t get too bogged down in some of the jargon that science wraps itself in to make itself feel kind of clever and special, if you like. You know, being able to communicate our findings clearly and openly is a healthy practice in its own right.
But then also by engaging with the public, with end users, with policy makers and so on we can refine what we’re doing. We can learn a more appropriate language to couch our research in. We can find out what the needs of those end users and those policy makers might be, what questions they have that we could help answer. So I think that kind of dynamic process of engagement is really health for us even if we’re not directly trying to influence policy ourselves.
On defining addiction
No, I mean, we can use rules of thumb, if you like. So, for example, in the case of smoking, if you’re the kind of person who wakes up in the morning and you have to smoke a cigarette first thing, before you brush your teeth, before you get out of bed, because you’re experiencing such profound nicotine withdrawal that you just can’t function before you have that cigarette. That’s a pretty good marker that that person is, you know…requires a dose of nicotine before they can get on with their day. So that’s a kind of rule of thumb definition that you might use. But you might be excluding other people that don’t smoke in that way, but who could still meaningfully be described as addicted on other measures. So it really depends on exactly why you’re trying to define it for the purposes of a research study or whatever it might be.
On addictive personalities
Well, observationally these things tend to go together. So people who use tobacco tend to use alcohol, tend to use caffeine and so on. Now why that is, is something that we’re still trying to unpick for all the reasons that I talked about previously. It’s not trivial to say whether there’s a common factor that influences all of those different things or whether the use of one substance increases your likelihood of the use of another substance. And we’re trying to get a handle on that using the kinds of methods that I mentioned previously. We know that drugs of abuse tend to share the property of being able to activate the brain’s reward pathways so there does seem to be some common basis to substance use, more generally. But then there are also factors that are specific to different drugs. So, for example, the enzymes that break down nicotine in our body are different to the enzymes that break down alcohol. People differ in the speed at which they break down nicotine and that’s largely unrelated to differences in how people break down alcohol. So there’ll be biological factors that characterise those differences that will be different for different drugs.
On behavioural addictions
There’s a lot of interest in that, particularly in the context of, say, gambling. And I think gambling is a very interesting example where you clearly have a sub-group of people who display highly problematic behaviours that in many respects look a lot like addictive behaviours. I think there’s an ongoing debate about whether we should call that addiction or not. I am wary of using the term addiction too lightly, because you then get into the realms of talking about tanorexia and sun bed addiction and things that I don’t think are helpful uses of the word, but that’s just my opinion. So I don’t do that work on behavioural addictions myself, but I’m aware that it’s out there and it certainly intersects with the work that we do on substance use.