Schizophrenia is a brain disease: The fly brain man cometh

On 12^th April, a BBC Radio 4 discussion included a contributor suggesting first that schizophrenia is a brain disease, and secondly that discoveries in the physiology of the fly brain could offer insights into the origin of this ‘disease’.

--------------------------------

An email exchange ensued. It culminated (but possibly hasn't ended) with my colleague, Richard Bentall, offering a concise, but assertive, commentary:   

Dear Dr H,

[My colleague] forwarded me your email to him, which I feel moved to reply to. I am copying in various colleagues who are as concerned as I am about the naive biological reductionism that seems to be dominating media discussions of mental health these days. Briefly, the problems with this view when applied to 'schizophrenia' are:

(i) Schizophrenia is a meaningless construct

There is no syndrome of schizophrenia and nobody can agree on who is schizophrenic. To my knowledge, no statistical study has ever identified a cluster of symptoms that correspond to the Kraepelinian concept or its subsequent revisions. Most recent studies have converged on a multidimensional model that incorporates five dimensions of positive symptoms, negative symptoms, cognitive dysfunction, depression and mania/excitability, or even more complex structural models which encompass patients across the psychosis spectrum (see Demjaha, A., et al. (2009). Combining dimensional and categorical representation of psychosis: the way forward for DSM-V and ICD-11? Psychological Medicine, 39(12), 1943-1955 and Reininghaus, U., Priebe, S., & Bentall, R. P. (in press). Testing the psychopathology of psychosis: Evidence for a general psychosis dimension. Schizophrenia Bulletin, available online). In recent field trials, the proposed DSM-V criteria for schizophrenia generated a derisory kappa of 0.46, showing that clinicians working with a precise definition of the disorder and following a diagnostic interview often could not agree on who was schizophrenic and who was not (Regier, D. A., et al(2013). DSM-5 field trials in the United States and Canada, Part II: Test-retest reliability of selected categorical diagnoses. American Journal of Psychiatry, 170, 59-70)!

(ii) Heritability coefficients are misleading

It is often forgotten that heritability coefficients are, actually, fancy correlation coefficients. We all know, or should know, that correlation does not necessarily prove causality. Heritability coefficients are statements about populations and not individuals so that it is wildly misleading to suggest that high heritability = mostly genetically caused (for a detailed discussion of this, see Bentall, R. P. (2009). Doctoring the mind: Why psychiatric treatments fail. London: Penguin.).

In fact, precisely because heritability coefficients are correlations which attempt to parse up the variance in a trait to genetic and environmental causes, low variance in the environment leads to inflation of heritability. This is why, for example, IQ is highly heritable in middle class families (where environmental variation is low) but very low in working class families (where environmental variation is high) (Turkheimer, E., et al. (2003). Socioeconomic status modifies heritability of IQ in young children. Psychological Science, 14, 623-628). Also, heritability coefficients assume an additive model of genes and environment, which is wildly implausible given what we know know about how genes work. Again, assuming an additive model when there are G x E interactions leads to massive inflation of heritability and an underestimate of environmental effects (Dickins, W. T., & Flynn, J. R. (2001). Heritability estimates versus large environmental effects: The IQ paradox resolved. Psychological Review, 108, 346-369). This is probably why, as you know, molecular estimates of heritability are generally much lower than those based on the methods of classical genetics. The 'missing' heritability in these studies is probably phantom heritability.

Incidentally, you will also know from the genetic studies that you cite, that the consensus amongst geneticists is now that many common alleles (perhaps hundreds) probably each confer a tiny risk of all kinds of psychosis . Although some CNVs have much higher association with psychosis, they account for only a small proportion of patients and are also associated with intellectual disabilities and autism (Owen, M. J. (2012). Implications of genetic findings for understanding schizophrenia. Schizophrenia Bulletin, 38(5), 904-907. doi: 10.1093/schbul/sbs103). This is further evidence, if ever it was needed, that schizophrenia is a meaningless construct and confirms the impossibility of devising a genetic test for the disorder.

(iii) There is massive evidence that environmental factors are causal in severe mental illness

The implications of ii above are that you can't estimate environmental influences from heritability estimates - you have to look for them and measure them. Recent studies have pointed to a wide range of environmental factors associated with psychosis. These include social disadvantage, migration, living in cities and various forms of victimisation. I attach a recent meta-analysis I conducted on the evidence linking childhood adversity to psychosis (Varese, F., et al. (2012). Childhood adversities increase the risk of psychosis: A meta-analysis of patient-control, prospective and cross-sectional cohort studies. Schizophrenia Bulletin, 38, 661–671. doi: 10.1093/schbul/sbs050.) The bare odds ratio between childhood trauma was stable across methodologies (retrospective/prospective) and came in at about 3, much higher than any association with common alleles. More importantly, there is evidence of a dose response effect, with Ors climbing to around 50 for children who have been multiply traumatised. Reaction in the psychiatric community has sometimes been bizarre, with convoluted attempts to explain away the data (see a recent editorial I wrote about this, also attached).

(iv) Brain studies do not provide clear evidence of neurodevelopmental disorder in psychosis

The evidence linking the basal ganglia to psychosis is far from clear cut. The best evidence is from response to antipsychotics, but recent studies suggest that only about 20% of patients show a genuine clinical response (Marques, T. R., et al. (2011). The different trajectories of antipsychotic response: antipsychotics versus placebo. Psychological Medicine, 41(07), 1481-1488). In any case, abnormal basal ganglia activity could just as likely be attributed to environmental factors – animal studies show that chronic victimisation leads to sensitisation of dopamine pathways in this part of the brain (Selten, J.-P., & Cantor-Graae, E. (2005). Social defeat: Risk factor for psychosis? British Journal of Psychiatry, 187, 101-102). Current structural neuroimaging studies of psychosis are probably not to be trusted for a variety of complex methodological reasons (Ioannidis, J. P. A. (2011). Excess significance bias in the literature on brain volume abnormalities. Archives of General Psychiatry, 68, 773-780), not least the emerging evidence that drugs affect brain structure (Ho, B.-C.,et al. (2011). Long-term antipsychotic treatment and brain volumes. Archives of General Psychiatry, 68, 128-137) but, in any case, could also be the consequence of social and environmental factors (Hoy, K., et al. (2011). Childhood trauma and hippocampal and amygdalar volumes in first-episode psychosis. Schizophrenia Bulletin. doi: 10.1093/schbul/sbr085

(v) A narrow neurodevelopmental approach is damaging to patients.

There is little evidence that the biological approach to psychiatry is benefiting patients. Outcomes for patients suffering from 'schizophrenia' have not improved since the Victorian age and an increasing number of people are disabled by psychiatric conditions. This is precisely the opposite to what has happened in physical medicine, where genuine advances have led to improved outcomes and reduced disability (see my Doctoring the Mind, and also Whitaker, R. (2005). Anatomy of an epidemic: Psychiatric drugs and the astonishing rise of mental illness in America. Ethical Human Psychology and Psychiatry, 7, 23-35). Just as importantly, although it is often assumed by doctors that promoting a biological understanding of psychosis will reduce stigma, empirical research provides strong evidence that the opposite is the case (Read, J., et al. Acta Psychiatrica Scandinavica, 114, 303-318; Angermeyer, M. C., et al. (2011). Biogenetic explanations and public acceptance of mental illness: systematic review of population studies. British Journal of Psychiatry, 199, 367-372.)

The claim that biological research (on flies or whatever) will one day lead to a cure for schizophrenia is a common rhetorical trick designed to gain publicity and guarantee grant funding. I have no problem with research on the CNS of flies, which seems valuable in its own right. But linking flies to schizophrenia (whatever that is) is really about self-promotion and is damaging to the interests of patients.

Sincerely etc.