Genetic research into ‘schizophrenia’ – how much can it actually tell us?
Lucy Johnstone and Richard Bentall talk about the role of genes in mental health
What causes mental health problems? Following the publication of Oliver James’s book Not in Your Genes last week, the In the Mind BBC series, and the announcement of two recent ‘breakthroughs’ about genes for ‘schizophrenia’ (in Nature, and in Nature Neuroscience), there has been plenty of recent debate about nature vs nurture mental health. From Radio 4 to the Guardian (see here and here) and well beyond, there is robust disagreement about the role that different factors play in our mental well being.
In the attempt to shed some light on the matters, we asked two psychologists to discuss the issue of genetics and their role. Previous DOTW contributor Professor Richard Bentall (of the University of Liverpool) researches social and psychological factors in mental health, but he is also interested in possible polygenetic risk factors for ‘psychosis.’ Dr Lucy Johnstone is a clinical psychologist. Her most recent book, A Straight Talking Introduction to Psychiatric Diagnosis, aims to promote choice by summarising debates about psychiatric diagnosis accessibly.
You know infinitely more than I do about genetics and research methodology, but I’m interested in unpicking the logic of research into genes for ‘schizophrenia.’
No sensible person dismisses the possibility of genetic contributions – presumably genes are distantly related to every form of human thought, feeling and behaviour, including the fact that I am writing this blog post (partly as a result of having genes for the development of hands, eyes and a brain.) And I don’t think any of us would be surprised if a tiny minority of people with a diagnosis of ‘schizophrenia’ (as claimed in the recent Sanger Institute study) turned out to have some kind of genetically-based learning difficulty. Although then of course we wouldn’t call it ‘schizophrenia.’ We’d call it ‘Sanger Syndrome’ or something like that – and there would be no reason to suppose that it necessarily told us anything at all about the much larger group with a ‘schizophrenia’ diagnosis.
But surely the question is – what kind of explanation is likely to result from research into the genetics of ‘schizophrenia’ anyway?
I’m not a geneticist, but I cannot see, logically, how a non-valid construct like ‘schizophrenia’ or ‘psychosis’ can have genetic causes, any more than I can envisage a group of genes that predispose for ‘hysteria’ or a ‘wandering womb’ or any other redundant psychiatric concept. Moreover, the experiences that are grouped together as ‘schizophrenia’ are thoughts, feelings and behaviours, not biological dysfunctions. How can you have genes that increase the ‘risk’ (the wrong word, I think) of certain thoughts, feelings and behaviours, except in an impossibly distant and indirect sense? For example, it’s likely that genes coding for strong legs increase the ‘risk’ that you will later play football. But that really isn’t an explanation for why people run up and down a field kicking a ball, and nor does it imply pathology. (Although perhaps that point is debatable…) And even if it did, what would you do about it? It seems to me that any such findings would relate to such general traits – e.g. temperamental sensitivity – that it would be neither possible nor desirable to eliminate them.
I really think we need to let go of the idea that there is such a thing as ‘schizophrenia’, or indeed ‘psychosis’, and then see what exactly needs explaining that isn’t already adequately accounted for by the numerous relational and social factors we know about. The emerging trauma-informed approach already incorporates the mediating role of biology via the neurodevelopmental impacts of attachment relationships and evolved threat responses. Obviously, these capabilities are genetically transmitted at some level. But what else are gene studies likely to tell us, apart from confirming the fact that humans have evolved to be sensitive to abuse, trauma, neglect and social exclusion?
Anyway, Richard, you know I have a great many genes that predispose me to being argumentative, so I hope you’ll forgive these comments! But I would be genuinely interested in your take on it.
It seems to me that there are several issues that need disentangling here.
1. A strategy issue: I’m anxious not be open to the accusation of genetic denialism. The point is that we need to restore balance to the nature vs nurture (actually nature x nurture) debate and, to many people, we look silly if we deny that genes play any role (which I know is not your position). It seems to me that the quantitative genetic evidence (from family, twin and adoption studies) all show an association between genes and mental health, and this is supported to some extent by molecular genetic evidence, so we should acknowledge this. The question, as you say, is what does this mean?
2. The science issue: But as you say, from a scientific point of view, it’s important to establish how genes are involved.
Of course you don’t need to tell me that schizophrenia is not a useful scientific construct to work with when attempting this and, indeed, the genetic data is often used to make precisely that point these days (the Cardiff group, for example, do not believe in ‘schizophrenia’ as an entity any more than you or I do; they argue – correctly, in my view – that the genetic data is incompatible with the idea that psychiatric disorders segregate into discrete categories like schizophrenia.)
A frequently neglected issue is that genetic research is entirely correlational, and therefore every bit as vulnerable to confounding as any other kinds of correlational research. Hence, just showing an association between genes and mental health does not imply that genes are causal in any simple sense. The recently reported association between SETD1A and schizophrenia, announced as a fundamental breakthrough by researchers at the Sanger Institute, illustrates this perfectly (and raises important ethical issues about the way that genetic researchers report their findings to the public). The gene was found in 10 of about 8000 psychotic patients, but 7 also had learning disabilities, so only 3/8000 non-cognitively disabled psychotic patients had the gene. A separate sample found 4/4000 people with neurodevelopmental problems had the gene. So the gene has a tiny effect in terms of risk of psychosis and is almost certainly really a gene which affects cognitive development. Of course impaired cognitive development is already known to be a small risk factor for psychosis for reasons unknown (maybe lack of cognitive resources impairs the ability to cope with stress, maybe it is a risk factor for experiences of powerlessness and victimization). So it seems very unlikely that SETD1A is causal in psychosis in any simple sense, but this did not stop Sanger Institute Researchers claiming it is ‘conclusively implicated in schizophrenia’, and that its discovery could lead to major therapeutic benefits. To my mind, these kinds of claims are blatantly dishonest and would not be tolerated in other areas of science.
What is really interesting to me is that, properly interpreted, the genetic data does not point to anything like the ‘genes for schizophrenia’. The data is entirely consistent with a more nuanced understanding of psychosis as the end point of a continuum of human variation (the polygenic load concept, for example, is consistent with the continuum model but would be very hard to square with a categorical approach to diagnosis). It is also entirely consistent with the idea that the social environment plays an important role although, egregiously, this has often been denied by some biological psychologists and psychiatrists who don’t seem to understand their own data (the claim that ‘schizophrenia is 80% heritable’ is still often used to promote the idea that psychosis is largely determined by genes whereas, even if this figure can be believed, heritability means nothing of the sort – for a discussion of this issue, see my book Doctoring the Mind).
3. The philosophy of science issue: Filippo Varese and I recently penned an editorial about the different standards by which biological and environmental studies are evaluated. Having thought about this issue over the last few days, I’ve come to the conclusion that this problem is much deeper than we then realised. It’s clear to me that biological papers routinely get away with stuff which would consign papers to the dustbin if they were about environmental determinants. For example, in the SETD1A paper referred to above, the researchers pooled data from studies using different methods and, when they realised that they lacked statistical power, added some data that had previously been published by another research group. But it’s not just genetics that’s the problem – a couple of years ago John Ioannidis (who is a bit of a hero of mine) published a meta-analysis of the statistical power of neuroimaging studies, finding that they averaged out at 7% (for those who have not had the benefit of a statistical education, this calculation concerns whether a study sample size is sufficient to detect a true effect, and is normally expected to be above 70%).
Why does this happen? Part of the problem seems to be a cultural (maybe even universal) human propensity for accepting essentialist ideas (there’s some interesting work by developmental psychologists on this) but it seems to me that this is no excuse: the idea of the journal refereeing system is that it is supposed to eliminate natural biases in the evaluation of research, and it plainly fails in this regard. And the problem is wider than psychiatry – there is a real crisis about replication in social psychology at the moment as well.
Science is above all a social process, and a whole host of social pressures affect what scientists do. These days it is routine, for example, for senior academics to be threatened by severe sanctions (redundancy) if we fail to deliver highly cited papers and large grants. And then there’s the power issue. Who gets appointed to grant awarding panels and journal editorships? And what kind of research do they select for further advancement (funding or publication in a high impact journal)? As the philosopher of science Philp Kitcher noted in his 2003 book Science, Truth and Democracy, science has been captured by an undemocratic elite which routinely and without even thinking about what they’re doing hands out large rewards to their mates or people who look like their mates. And, of course, their mates in return hand out prizes back to them.
Finally, there is the public understanding of science issue: Given all of the above, it seems to me that the public is up against it when it comes to understanding the nature x nurture issue. Which is where we come in, pushing back the tides…
We are pretty much in accord about the conceptual issues. What you are saying – and I heartily agree with you – is that these are fundamentally issues not (just) of science, but of power – who has it, in whose interests it operates, and how it is maintained by marginalising accounts that might challenge the dominant ones. Any remotely objective scientific perspective would take a 3% genetic risk – which seems to be what the studies are showing – as strong evidence that the thoughts, feelings and behaviours we call ‘schizophrenia’ are almost entirely a response to the environment – and as we know, those environments typically consist of multiple relational and social adversities.
But given all this, why are we still pursuing the genetic route? We wouldn’t, I hope, search for genetic reasons to explain why people are distressed by homelessness or debt. As you know, I have recently been working in the Valleys of South Wales, one of the most deprived areas in the UK. Just down the road in Cardiff, millions of pounds are being poured into yet more genetic studies – even though their own research fails to support the validity of the concepts they are investigating. Each glowing account of the latest ‘breakthrough’ starts with a statement to the effect that: ‘For the first time, we have evidence that……’ Somehow I must have missed all the retractions. But I know what would have made most difference to the lives of my clients, and it wasn’t genetic research. Should it still have a role?
This may be a point where we disagree. The 3% you refer to is, I think, the amount of variance accounted for in genome-wide association studies (actually, some estimates are a bit higher but nowhere near as high as the heritability estimates from quantitative genetics, an embarrassing difference that genetic researchers call ‘missing heritability’). But variance is a correlational concept and therefore does not address the confounding issues referred to earlier.
The problem is not genetic research: it’s naive genetic research. Geneticist have failed to realise that they cannot hope to learn anything by studying genes in isolation. For reasons already explained, showing a (typically tiny) association between a gene and mental health is completely uninformative, unless we explore the pathway by which that gene exerts its small effect. This means addressing what geneticists call the phenotype problem (in other words, looking at patients’ experiences in detail and not just lumping them together into meaningless categories like ‘schizophrenia’), which they talk about but do nothing about in practice. But it also means measuring the social environment and the psychological mechanisms that link risk factors to outcomes, which they never talk about at all. Whereas it is very unlikely that genes ‘cause’ psychosis, they may well modulate the pathways between risk factors and outcomes (think of them as dimmer switches attached to the psychological mechanisms that determine how we react to adverse environments). Research of this kind may well lead to a truly bio-psycho-social understanding of mental health, point the way to personalised psychological and medical therapies, and also help inform public mental health policy.
I have been trying to interest geneticists in this kind of research for a while, and some have been at least willing to talk about it, but my applications for funding for these kinds of projects have all been unsuccessful despite consistently very positive reviews. There is an ideological resistance to this kind of research amongst the senior biologists and doctors who hold power within organizations like the Medical Research Council, who would rather stick to their test-tubes and their machines that go ‘ping’ than support the kinds of studies that might actually help people.
We agree about the need for a sophisticated approach to genetic research, and it’s a disgrace that it seems so much harder to get it funded. But I return to my earlier point. We’re not funding genetic research into genes that may modulate the pathways between the ‘risk factor’ of bereavement, and the ‘outcome’ of weeping and despair. Why then do we feel the need for an extra genetic layer of explanation for responses to bullying, sexual abuse, racial discrimination, and all the other adversities implicated in ‘psychosis’? And I think we have to be careful about what we mean by ‘biopsychosocial’ models. Unless very carefully defined, they can be used to privilege biological causal factors that have never actually been identified. I’ve always thought this was highly insulting to service users. Breaking down after a horrific series of events isn’t evidence of biological ‘vulnerability’ – with the implicit message that others without this genetic make-up would have coped just fine. It’s a sign of being human.
Using pseudo-medical terms like ‘psychosis’ – a mysterious entity that apparently exists somewhere between the adversities and the response to the adversities, and then becomes an object of investigation in its own right – distracts us from seeing what is in front of our noses. A mountain of research tells us that reducing social inequalities, implementing preventative public health measures, and asking what has happened to you instead of what is wrong with you, is what is urgently needed. Gene studies are not a priority, and may even be a distraction.
We’ve laid out our cases. What do others think?