Autism spectrum disorder, schizophrenia, and bipolar disorder may superficially appear to be wholly distinct from one another, however these disorders often share overlapping clinical symptoms, cluster within the same families, and in more recent years have also been show to share genetic risk factors. In a Genome Medicine review article, Michael Owen and Joanne Doherty from Cardiff University, Wales, discuss the implications of growing genomic evidence for the overlap between psychiatric disorders, both for future research and clinical practice. Here Owen explains how psychiatric disorders are currently classified, how genetic studies have provided some surprising insights, and what impact this will have for diagnoses, treatment and further research.
How did your interest in psychiatric genomics begin?
I became fascinated by psychiatric disorders especially schizophrenia when I was a medical student. I did a PhD in neuroscience then I realised that the only robust aetiological factor for schizophrenia was a family history (this was the late 1970s). So I trained in psychiatry and genetics hoping that the new molecular genetic techniques would allow us to identify risk genes as a way into understanding the neurobiology. It’s been a long hard road but with improving genomic technology and analytical sophistication, progress has been tremendous over the past five years.
How are the major psychiatric disorders currently diagnosed and what is the importance of DSM-5?
As we understand so little about disease mechanisms, current approaches to diagnosis are largely descriptive. What we call disorders are actually syndromes: constellations of signs and symptoms that tend to occur together.
Work in the 1960s showed that there were significant differences in how diagnoses were being made between countries and pointed to the need for diagnostic standardisation. This led to the development of operationalised classifications that were designed to provide a reliable way of assigning a patient with a particular constellation of signs and symptoms to a diagnostic category. This works by defining a set of inclusion and exclusion criteria for each category. This greatly improved diagnostic reliability (different clinicians should make the same diagnosis of a given case) but at the expense of validity (the degree to which diagnoses defined specific pathological entities). Most Western clinicians use either the Diagnostic or Statistical Manual of the American Psychiatric Association (DSM), or the International Classification of Diseases (ICD) of the World Health Organization. These are updated every decade or so and the latest update of the DSM criteria (DSM-5) has recently been produced. DSM-5 is important because it is used clinically in the US and forms the basis for private healthcare funding.
The intention has always been to modify these diagnostic schemes in the light of new insights into aetiology and pathogenesis, but in psychiatry the weight of evidence has largely been insufficient to justify radical overhaul. Concerns about the degree to which current diagnoses map onto the underlying biology have been around for years, but recent findings in genetics showing extensive pleiotropy at the level of clinical diagnoses have been the straw that broke the camel’s back. There is an increasing concern that current diagnostic approaches are actually hindering research into aetiology and pathogenesis and that, at least for research, we need to rethink how we classify patients.
What are the main surprises from recent studies of the genetic relationships between major psychiatric disorders?
The main surprise in my view is the very broad range of psychiatric outcomes associated with rare disruptive mutations (both copy number variations and point mutations). These outcomes include disorders such as intellectual disability, autism, attention deficit hyperactivity disorder and schizophrenia, as well as people who are apparently asymptomatic. This is similar to the range of outcomes that result from early environmental insults affecting brain development such as birth trauma and in utero infection. I think this is telling us that brain development can buffer itself to an extent against early disruption and this will perhaps suggest new potential therapeutic approaches.
How could the diagnosis of psychiatric disorders be improved, in light of the recent genomic findings?
Recent genomic findings support the view that we are not dealing with a set of one to one relationships between genotype and phenotype but rather that we are dealing with many risk alleles, many of which will be associated with a variety of outcomes as determined by the combinatorial effects of other risk alleles, the environment and doubtless chance. These observations lead to the conclusion that, as well as being insufficiently grounded in disease biology, a major weakness of our current approach to classification may be that it is categorical rather than dimensional. [A categorical approach assesses whether an individual has a disorder or not based on symptoms and characteristics described as typical of the disorder, whereas a dimensional approach quantifies a person’s symptoms or other characteristics of interest along a continuum, rather than designating them to a particular category*].
As far as the clinic is concerned we are not yet in a position to throw out DSM and ICD and replace them with something better. As far as research is concerned, it seems absolutely clear that we cannot continue to rely on simply classifying patients according to current DSM/ICD categories. First, there should be an increasing emphasis on studies that do not stratify solely by current diagnostic categories. These might select cases that share specific clinical symptoms or risk factors but who have different categorical diagnoses or select those who form a subgroup of a current diagnostic category as defined by particular features. Second, in view of the continuous nature of psychiatric phenotypes, there should be an increasing focus on dimensional measures of psychopathology as well as of underlying brain dysfunction. Third, in relating psychopathology to underlying mechanisms, we should seek so-called ‘intermediate phenotypes’ that provide measures of underlying pathophysiology and where possible these should access established brain-behaviour relationships. For instance these might involve measures of neurocognition, neuroimaging or electrophysiology.
Do the recent genomic findings have implications for the treatment of psychiatric disorders?
Indirectly recent findings suggest that we should probably place less emphasis on treating ‘disorders’ and more on treating symptoms and better still underlying disturbances of neurobiological or psychological function. This probably applies to current therapies as well as to those developed in the future. In fact psychiatrists often treat symptoms or groups of symptoms (e.g. depression, psychosis) rather than targeting disorders per se. But drug licensing is still very much based on current diagnoses and this policy should be reviewed.
It’s early days to expect new treatments from recent genetic findings, but they certainly point to specific areas of biology and in particular proteins associated with synaptic function. We can expect these, and other emerging findings in genetics, to become the focus of mechanistic research in the next few years. The complexity and heterogeneity of psychiatric syndromes suggests that we will need biomarkers to stratify patients for future therapies and these will likely emerge in part from research into intermediate phenotypes.
What are the main challenges to future studies of the genomics of psychiatric disorders?
The immediate goals are fairly clear. First we will need to continue to genotype and, increasingly, sequence very large samples to identify more risk genes and, more importantly, reveal more biological insights. The main obstacle here is funding. Sample collection and sequencing are expensive but genetics needs to go further. We have only revealed the tip of the iceberg but this has already given us profound new insights. We need to finish the job. This doesn’t mean carrying on indefinitely but at least until the gains in terms of biological insights start to plateau.
Second, we need to tackle the problem of how new approaches to phenotyping can be applied to sufficiently large samples. Some traction might come from access to routinely collected clinical data, which can increasingly be mined electronically. Here ethical concerns will have to be addressed, safeguards put in place and the public convinced of the crucial importance of this work. There will also be an increasing use of smart devices to measure phenotypic variation over real time in real life settings. Some intermediate phenotypes will be highly scalable, such as measures of cognition, which can be applied over the internet. Others, such as neuroimaging, will likely require large samples to be assembled through collaboration and co-operation between researchers of the sort that has been so fruitful in genomics and through collaborative projects such as the Human Brain Project. Finally, we need to start considering how best to access data on environmental risk in large-scale studies.
What are the remaining obstacles to the translation of genomic findings to clinical psychiatry?
Apart from the need for new ways of stratifying patients for research, translation of genetic findings faces a major challenge from the high degree of genetic complexity. Standard approaches to animal and cellular models in which single genes are manipulated might take us so far, but it is clear that much risk in psychiatric disorders comes from the combined effects of many hundreds of risk alleles. Turning these findings into biological insights will require much improved annotation of the genome and proteome in the brain allowing us to identify the biochemical pathways and protein networks upon which they impinge. To bridge this annotation gap we need shared resources that will provide information about transcriptomes, proteomes and epigenomes at different stages of development and about protein interactions.
There will need to be an increasing focus on networks rather than single genes and larger scale and more quantitative approaches in neuroscience. This will in turn require scalable measures and more collaboration.
How do you see the field of psychiatric genomics developing in the next ten years?
We will identify more risk genes, pathways and networks. There will be an increasing focus on relating specific risk genes and networks to specific phenotypic features.