Will the future look like GATTACA? The application of genomics in Psychiatry
Hospital setting. A child is born and cries at the top of his lungs. The hospital personnel carefully takes the newborn from the mother and collect a blood sample from his heel. One drop of blood is deposited in a machine that few seconds later expels a paper sheet. A doctor reads out loud the output of the genetic analyses: “Neurological conditions 60% probability. Manic depression 42% probability. Attention deficit disorder 89%. Heart disorders 99% probability, early fatal potential. Life expectancy 30.2 years”.
This scene is taken from GATTACA, a 1997 cult science-fiction movie directed by Andrew Niccol and starring Ethan Hawke, Uma Thurman, and Jude Law.
I am a researcher at Department of Psychiatry of Amsterdam UMC, where I lead the genetic research activities. I am also a clinical psychologist by training and sci-fi aficionado. While recently celebrating the 25th anniversary of the release of GATTACA, I asked myself whether the future application of genomics in clinical psychiatry may resemble the one seen in the movie.
GATTACA depicts a dystopian future in which society is rigidly structured in classes demarcated by the results of genetic predictions of the individuals’ potential. But not necessarily of its actual realization as the story will teach us. The main focus is the societal abuse of these predictions to impose an order based on inequalities and discrimination. As the main character narrates “I belonged to a new underclass, no longer determined by social status or the colour of your skin.”
Predicting disorders with genetic data
GATTACA - released 6 years before the completion of the Human Genome Project that provided a first full reading of human DNA - included clever foresights of the potential future application of genomics (the study of the genetic sequence of organisms), including the possibility to predict the risk to develop physical or mental disorders.
This is slowly becoming reality in research with the development of tools called Polygenic Risk Scores (PRS). These are scores which predict the lifetime risk of developing specific disorders. The last two decades of genomic research taught us that the risk for psychiatric disorders is not determined by one or a few genes, but rather from hundreds if not thousands of genetic variants scattered across the entire genome, each one determining a very small increase in the overall genetic risk.
PRS captures the additive effect of this myriad of small risks in the attempt to predict the overall genetic risk of disorders. PRS are built using parameters derived from large studies called GWAS (Genome-Wide Association Studies), in which millions of genetic variants (points in the DNA sequence differing between individuals, also known as single-nucleotide polymorphism or SNPs) are compared across thousands to hundreds of thousands of subjects with a certain disorder versus healthy controls.
Psychiatric genetics has been at the forefront of this field, with the first study applying PRS in 2009.
The limits of genomic prediction
In GATTACA genomic prediction tools were applied at birth in the entire population to derive the probabilities of developing different conditions. Will this be the most rationale use of PRS for psychiatric disorders in our future? Highly unlikely, based on the knowledge we are rapidly accumulating.
There are inherent limits to the predictive capacity of PRS. The risk of developing any disorder is differently distributed in the population: some persons have high risk, others low risk and all the others will lay in the middle of these extremes. This risk distribution is partly due to genetic factors (“heritability” is the technical term indicating the portion of this distribution explained by genetics) and partly due to other non-genetic and environmental factors such as, for instance, trauma or socioeconomic deprivation.
Here we could appreciate the most obvious limit of genetic prediction: a genomic-based tool like a PRS could capture only the genetic portion – the heritability - of the risk, which is only one part of the overall disease risk. PRS for psychiatric disorders currently capture only a very minimal part of this “heritability”, which represent an even smaller proportion (<10%) of the overall risk.
Another limitation is that a PRS built from GWAS performed in subjects of a certain ancestry (e.g. north-European) have lower prediction ability when applied to populations of different ancestral background (e.g. east-Asians). To date the overwhelming majority of GWAS has been performed in subjects of European ancestry, with only less than 5% of the studies including participants of different populations. Application of PRS without realising their skewed ancestry representativeness could not only impair their predictive performance, but also further exacerbate existing health inequalities.
Finally, the public health value of screening an entire population for a certain disorder depends on the number of cases that could be identified. Let’s consider conditions with low population prevalence like schizophrenia or bipolar disorder (~1%); after a (logistically challenging and expensive) full-population screening effort, even among the small subgroup of subjects with the highest PRS, only a very limited number of individuals is expected to develop the disorder.
A more interesting future
Rather than applied to the whole population as in GATTACA, genomic tools may become more relevant in groups at higher risk, such as subjects with family history of psychiatric disorders or those presenting at early stages with undifferentiated prodromal symptoms. For these people, PRS could be helpful in resolving diagnostic uncertainties or as prognostic predictors at different stages of their clinical trajectory.
The emerging field of pharmacogenomics is also showing that genetic data may be crucial for the selection of the appropriate medication regimen aimed at improve clinical effects and reduce adverse drug reactions.
Research on these potential applications has just begun and important issues need to be addressed before this technology will be ready for clinical implementation. The predictive accuracy of PRS for psychiatric disorders must improve trough increasingly larger GWAS testing not only diagnostic status (cases versus controls), but also other key outcomes of clinical relevance such as treatment response.
Furthermore, GWAS studies in samples from diverse populations are highly needed in order to ensure that PRS would be useful for subjects of different ancestry and not be limited to privileged groups (e.g. those with European ancestry).
Finally, a new generation of clinicians need to be properly trained in the use and communication of genetic probabilistic risk predictions to patients. In all these promising future applications, genomics tool will never be used alone, but always as part of a broader clinical appraisal together with all other established environmental risk factors, psychopathological features and biomarkers.
In this context, PRS may have an appealing advantage as compared to other type of biomarkers: genetic data need to be measured only once in life trough affordable technologies (currently <100$ per sample) and from that unique measurement PRS for multiple traits and disorders can be continuously generated and updated.
An alternative scene
The application of genomics in clinical psychiatry will unlikely resemble the one in GATTACA. Could we imagine an alternative scene for an hypothetical sci-fi movie?
Fade in. Visit room in a hospital. A man in his thirties has been referred to a specialized psychiatric evaluation. The psychiatrist connects her terminal to the microchip implanted under the skin of the patient containing his electronic medical records. In the meantime, the man describes the depressive symptoms he experienced in the previous months.
The psychiatrist reads on the terminal that the general practitioner has prescribed antidepressants. However, after several weeks of treatment the symptoms are worsening. The psychiatrist asks additional diagnostic questions to the patient and collect further information on his family history and relevant risk factors.
Then the clinician checks on the patient’s data for available biomarker measurements and activates a newly developed application that allows to generate on the spot - from the encrypted genetic data of the patients stored in his microchip - PRS for different psychiatric disorders based on the latest GWAS studies. She explores the output screen starting from the section on bipolar disorder.
Later. The psychiatrist and the patient sit together discussing the potential treatment options. The doctor explains to the patient her clinical hypothesis: the lack of response to antidepressant and the increased genetic risk for bipolar disorder measured by PRS led her to think that his mood symptoms may represent the onset of such disorder. Together they agree in starting a treatment course tailored for bipolar disorder.
Fade to black.