Researchers have analysed clinical and genetic data from over 8,500 adults with psychiatric disorders, to assess whether schizophrenia polygenic risk scores can improve the prediction of poor outcomes compared to standard procedures.
Polygenic risk scores are predictors of how liable an individual is to a disease due to genetic factors. They are calculated using thousands of risk variants, detected by genome-wide association studies (GWAS), which are then weighted and combined to provide a quantitative index relating to the likelihood that a person has, or will develop, a particular disorder. Today, polygenic risk scores are a fundamental component of human genetics research. They are able to give scientists insight into the relationships between different diseases and genomic alterations, but their clinical utility remains questionable.
It is recognised that many psychiatric disorders are heritable, but their genetic complexity means that single gene markers are rarely helpful for diagnosis. Therefore, ever since polygenic risk scores were first applied to psychiatry in 2009, researchers have learned a great deal about the large number and combination of gene variants that are associated with risk for conditions like schizophrenia and bipolar disorder.
However, clinical psychiatry has often lagged behind other fields of healthcare in terms of the application of new technologies and uptake of innovative research methods. In fact, to date, little is known about polygenic risk scores relative to standards of care for mental health. This means that there is uncertainty surrounding exactly how applicable the predictive analysis method is for helping psychiatric patients and how it would be most useful now, and in the future.
Studying schizophrenia polygenic risk scores
Recently, researchers have investigated the potential of contemporary schizophrenia polygenic risk scores as a clinical tool for an individualised outcome prediction for patients. The power of schizophrenia polygenic risk scores is among the most formidable as they have been recognised to exceed that of most other common psychotic illnesses. Therefore, it is unsurprising that there has been growing interest surrounding the predictive method emerging as a popular clinical tool for both diagnosis and prognosis of the disease.
The scientists studying schizophrenia prediction analysis analysed clinical and genetic data from over 8,500 patients from different ethnic backgrounds. The polygenic risk scores for schizophrenia were calculated for each individual using a self-completed medical questionnaire and information from BioMe – a biobank that links genetic and electronic medical record data.
Unfortunately, after these comprehensive assessments, it was found that schizophrenia polygenic risk scores did not improve the prediction of poor outcomes relative to clinical features captured in a standard psychiatric interview. These results were generally consistent across different ethnic backgrounds and data collection techniques. Essentially, these findings demonstrated that even the most powerful polygenic risk scores for psychiatric outcome predictions are limited in terms of clinical utility.
Can polygenic risk scores really deliver in the clinic?
Although disappointing, results like these are not uncommon. Recently, a study on coronary heart disease polygenic risk scores found that the predictive accuracy of the condition did not improve when used in combination with clinical features alone. Realistically, it is unlikely that polygenic risk scores will ever be able to definitively predict a diagnosis of a complex condition, such as a psychiatric disorder, mainly because genetic factors only contribute part of the risk.
Nevertheless, the future is still bright for polygenic risk scores in psychiatry, as combining them with information about other risk factors, such as lifestyle choices, has the potential to greatly improve disease outcome prediction and aid clinical decision-making. Moreover, data from larger sample sizes and greater ethnically diverse datasets are needed to improve prediction models, both of which are thought to be increasingly available through new machine learning approaches to GWAS. Therefore, although polygenic risk scores are not routinely used in clinical psychiatry practise at this time, innovations through advancing genomic research and technological breakthroughs are likely to drastically transform the ways that disorders are treated in the coming years.
In the future, ethical issues associated with the use of polygenic risk scores in clinics should be evaluated in detail, and realistic expectations should be set in place regarding what the predictive analysis can and cannot deliver. Psychiatry, in particular, is known for detail-rich narrative style clinical documentation. As a result, perhaps the field should prioritise the development of natural language processing mining of electronic medical records, alongside the development of automated information extraction systems. There is no doubt that this would require heavy data engineering, but it could lead to dramatic improvements in psychiatric disease outcome prediction and transform the delivery of mental health care.
Image credit: Consult QD