For many diseases, most notably cancer, drug development has been led by understanding of gene mutations and pathophysiology, but in psychiatry it has tended to be achieved through clinical observation.
That said, the value of clinical observation should not be underestimated as a starting point. After all, the Research Domain Criteria (RDoC) project arises from the idea that symptoms crossing diagnostic boundaries – such as the cognitive dysfunction common to major depression and schizophrenia – share disorders of neural circuitry which reflect common cellular, molecular and ultimately genetic abnormalities.1
Low glutathione levels predict cortical loss
And there are areas of psychiatry where understanding of etiological mechanisms is being used to enrich populations for clinical trials. One example – related to the possible role of oxidative stress in the development of schizophrenia -- is that of glutathione, where low levels seem to predict loss of cortical volume in people vulnerable to early onset psychosis.2
Target enrichment in a trial population is well and good but -- since individual response to treatment is dynamic -- we need to measure factors relevant to response not only at baseline but also during the course of treatment.3
Secondly, we need trials that allow adaptive treatment, so that patients not responding can be re-randomized to an add-on or alternative therapy.
Real-time data not subject to biased recall
Acquisition of real time data by passive monitoring poses problems of integration
And, thirdly, in predicting and measuring outcomes we need more precise measurement -- such as can now be achieved by the high-intensity, real-time monitoring of behavior through wearable devices – rather than episodic recall measures such as PANSS that are so subject to bias and unintended inaccuracy. This poses problems of integrating large amounts of data from multiple sources, but in this task machine learning is coming to our aid.
Those pursuing precision medicine have placed much faith in genetics. But we must not oversell its potential, Professor Arango cautioned.
If the lifetime risk of schizophrenia is around 1%,4 a genetic risk score that confers a twenty-fold increase means that an individual still has an 80% chance of not developing the disorder.
Pragmatic advances precede full understanding
In another example drawn from schizophrenia, there has been interest in the presence of the genetic deletion syndrome 22q11, which identifies a relatively homogeneous group of patients with shared etiology. But only 0.3% of people with schizophrenia have this deletion.5
On the other hand, there is no reason to be too pessimistic. It took twenty years to fully realize the benefits of targeted HER2 therapy in women whose breast tumours over-express the growth factor. It is often the case that advances can be achieved pragmatically, before underlying mechanisms are fully understood.