What does the recent ‘discovery of schizophrenia genes’ in Nature really tell us?
A group of academics recently published the results of a large scale study into the genetics of ‘schizophrenia’ in the prestigious journal Nature. O’Donovan et al reported the discovery of 108 genes that appear to increase the risk of developing the disorder. Some of these genes are involved in brain chemical messaging and others in the immune system of the body.
Following the publication, the media declared a schizophrenia genes ‘breakthrough’ and an uncovering of the ‘biological basis’ for schizophrenia. Academic psychiatry has not discouraged this speculation, instead has proclaimed new insights into the causes of ‘schizophrenia’, and the possibility of revolutionary new medications targeted at faulty brain molecules.
Others, however, warn that this may be one of a long line of studies that ultimately do not justify the hype they create, and end up contributing little to our understanding of psychotic illness. They warn that decades of exaggeration of evidence in favour of genetic factors and significant problems with methodology in genetic research have not been fully acknowledged. Taking this view further, some conclude that the emphasis placed on searching for genetic causes of ‘schizophrenia’ is a mistake, and even that this issue is symptomatic of fundamental problems at the heart of the psychiatric profession.
‘The heritability of schizophrenia’
From evidence inferred from historical investigations, ‘schizophrenia’ is thought by many to have a strong genetic component. The predominant idea is that the illness results when a genetic predisposition is combined with environmental ‘triggers’. In practice, this leads to an implied inevitability of illness in those rendered vulnerable by their inherited genetic code, and a relegation of the importance of environmental causes, such as childhood trauma, leaving home, or family strife.
If a condition or attribute has a strong genetic component, we tend to say it has a high heritability. Strictly speaking, heritability is an estimate of the degree to which variation of a characteristic is due to genetic variation, as compared to the degree that can be attributed to environmental reasons. This assumes that genetics and environment are entirely separate. Given that genetic factors influence a person’s effect on the environment (think, for example, of how a person’s facial symmetry might affect their experience at a party), and that the environment can affect alter genes (epigenetics), this seems implausibly simplistic.
O’Donovan et al state at the onset that “schizophrenia is a highly heritable disorder”, and that “high heritability points to a major role for inherited genetic variants in the [causes] of schizophrenia”. However, the two sources cited to support this statement actually appear to contradict it.
Problems with existing genetics studies
The first referenced paper, by Lichtenstein et al, is a family study, in which the prevalence of schizophrenia in relatives of those already diagnosed is compared to the prevalence in the general population. The authors’ results reinforce previous findings that ‘schizophrenia’ occurs more commonly in people who have a first degree relative with the same diagnosis. The obvious flaw with such studies is that this recurrence is not necessarily down to genes, but could well be related to a myriad of other factors that are shared by relatives.
The second reference, by Sullivan et al, is a meta-analysis of all of the existing twin studies of schizophrenia. Twin studies compare shared ‘schizophrenia’ diagnoses in identical twins and non-identical twins. As identical twins theoretically share 100% of their DNA, and non-identical twins only 50%, if more identical twins both have the diagnosis than both non-identical twins, the disorder is said to have a genetic influence. A meta-analysis looks at a number of independent studies, in order to produce a more accurate and less biased overall result.
The quality of a meta-analysis is dependent on the quality of the studies that it incorporates. The authors of this particular paper admit that “methodological quality of the published twin studies of schizophrenia was not uniformly high”. They go on to say that in most of the included studies the researchers were not ‘blinded’. Without blinding, there is a danger that conscious or unconscious bias in the mind of the researchers will skew results. This risk is made vivid by reports that one of the leaders of an influential early study was quoted ordering that “cases which present a schizophrenic picture clinically, but lack hereditary predisposition, must be excluded…”
Others have documented many other potential issues affecting the reliability of previous twin studies, including the lack of a shared, standardised diagnostic tool, distinguishing identical twins by appearance only, biasing statistical methods, small sample populations, and the fact that ‘identical’ twins can have differences in genes.
Sullivan et al attempt to defend these deficiencies by explaining that the studies were carried out “before the importance of these methodological features was widely recognised and viewed as essential”, and bizarrely appear to suggest that as these early studies “represented monumental and even heroic efforts by individual psychiatrists”, their problems can be ignored.
Another issue is that the authors of twin studies assume that twins, identical and non-identical, share environmental factors to the same degree. In fact, identical twins have been shown to share more environmental factors than non-identical twins, for example, they are treated in a more similar fashion, encounter more similar environments, and experience a higher degree of identity confusion.
That environmental factors are key to causing psychosis, or that ‘people can go mad when bad things happen to them’ is a common belief amongst the public, but apparently difficult for academia to grasp according to a curious discussion in the Sullivan et al paper. They write: “We also determined that there are significant common environmental effects on liability to schizophrenia…This is unexpected and of considerable interest…it is unusual to find a…disorder with significant common environmental influences…and schizophrenia is often described as one of the more “genetic” psychiatric disorders”.
Adoption studies, which compare illness in identical twins who have been raised in different homes, allow some teasing apart of genetic and environmental factors. Six major adoption studies have been carried out, two of which have been particularly influential. Important problems with both of these have been identified. For example, neither study showed evidence of a genetic link until the definition of ‘schizophrenia’ was expanded, or the study design changed retrospectively. In addition, many participants had been separated late from their biological parents, or were adopted late, which means that their experiences up to that point (e.g. living with a dysfunctional family) could have influenced the results.
Another overlooked factor was that of the widespread political and societal views toward the mentally ill when the adoption studies were carried out. Adoptees were studied in Denmark, the USA, and Finland during the mid-20th century. At this time, all of these countries had laws permitting the compulsory sterilisation of people diagnosed with ‘schizophrenia’. It is quite possible that children with a family history of ‘schizophrenia’ were seen as less desirable adoptees, and may have experienced rejection, and placement in more dysfunctional families. Further damning evidence of poor reliability in adoption studies is the revelation that interviews in one of these studies were fabricated.
O’Donovan et al used a molecular technique known as a Genome Wide Association study (GWAS). In this, the DNA of a group with the illness is compared with that of a control group. Each DNA sample is analysed to look for the presence of millions of potential variations. If these variants are found to be more common in the DNA of the illness group, they are said to be associated with the illness.
So far, there have been over 1700 molecular genetics studies into ‘schizophrenia’, without any especially useful results. Often, associations found in the first experiment are not found again with a repeat test. Time will tell whether results in the O’Donovan et al paper will also be found to be yet more false positives. Even if genes are confidently identified, they have extremely small individual effects in reality, and finding a gene associated with a disease is not the same as the genes being the cause of the disease.
The problem with ‘Schizophrenia’
The search for genetic factors in ‘schizophrenia’ has been carried out for over 40 years, with little success.
Some suggest that the search is futile, and that no genes for ‘schizophrenia will be found. Not because it is technically too difficult, but because there are no such genes. A key reason for this opinion is that ‘schizophrenia’ does not appear to exist as a discrete disease. It is a term used to describe a wide variety of symptoms and signs, and is likely to consist of many different conditions. As a disease, it is neither valid nor reliable.
The validity of a disease concept in medicine rests on identifying a clear cause. This is not the case for ‘schizophrenia’.
Reliability is the extent to which clinicians agree on the same diagnoses. The ICD-10 and DSM-V are diagnostic manuals which aimed to increase reliability of schizophrenia diagnosis, but reliability of diagnosis with their use has been shown to be poor. The diagnostic criteria in these manuals is disjunctive – i.e. one person can be diagnosed with schizophrenia if he has symptoms A and B, and another if he has symptoms C, D and E, which has led people to attack the very logic of schizophrenia.
Clearly, searching for genes related to a disjunctive set of symptoms with no single clear cause is unlikely to result in success.
What drives our devotion to DNA?
Why then, is so much effort, money and resource used to continue this endeavour? Perhaps the answer is found by considering what is not looked for.
Trauma, societal and family based factors, amongst others, have been implicated as causes of psychosis. These are much less straightforward to understand and research, being as they are ambiguous and loaded with conflicting moral and cultural meaning.
That psychosis can affect anyone, including an academic psychiatrist working in a genetics lab, may be terrifying. Imagining that it only affects the ‘other’ with their defective genes and brain chemistry can relieve anxiety.
That poverty, dysfunctional families and isolation from society may cause psychosis could worry politicians and policy makers. If so, it would be much more convenient for the majority of blame to lie deep within genetic codes.
A place at the Top Table
An illuminating quote from an author of the study in the publicity surrounding this paper, suggests a further reason for the focus on reductive biomedical causes:
“This study puts psychiatry into the same category as other parts of medicine”.
This seems to betray a feeling within psychiatry of unfairly not belonging to the top tier of medical specialties. Presumably, this reflects individual psychiatrists’ feelings of inadequacy at their perceived lower status and underachievement, compared, say, to those basking in the glamour of neurosurgery.
This appears to have been the way since modern psychiatry emerged from the asylums in the 19th century. At this time, great progress was being made in medicine. Psychiatry as a profession attempted to compete with medicine, adopting a biomedical approach that reduced our thoughts, feelings, behaviours and relationships to things that could be measured and altered in the same way as livers and kidneys.
Emil Kraepelin is regarded as a father of modern psychiatry and ‘discover’ of dementia praecox, which became today’s ‘schizophrenia’. Kraepelin attempted to group unusual behaviours into diseases caused by a physical pathology. Psychiatry has persisted down Kraepelin’s path, but a century later, is no closer to emulating medicine in the way desired by most, and the evidence suggests that outcomes for those affected by psychosis have not improved.
Promoting biogenetic models increases stigma and reduces sufferers’ feelings of control over their symptoms, which can lead to depression. On the contrary, ‘soft’, non technological, human, approaches – such as attempts to listen to and understand the the story of the individual behind the illness, developing a ‘therapeutic alliance’, building self esteem, and fostering control over symptoms, have been shown to be beneficial.
Perhaps we would all benefit from finding the courage to change direction in this field. We would need to accept, to quote Robert Gaupp, one time Professor of Psychiatry University of Tübingen, that:
“…the greater part of all genetic work in psychiatry would immediately collapse like a house of cards if Kraepelin’s theory was shown to be altogether mistaken”.
However, perhaps psychiatry doesn’t need to imitate medicine’s methods. With the right direction, it could show medicine the way forward in a more person-centered approach.