Predicting drug response in schizophrenia: are clinical & cognitive tests a cost-effective alternative to neuroimaging?

Schizophrenia is a severely debilitating disorder affecting 1% of the population. It is characterized by a range of symptoms including false beliefs, false perceptions and disorganized thinking. Medication can reduce these symptoms, but the response varies from one individual to another, in an unpredictable way. As a result, finding the most useful treatment for a given person involves trying s series of different medications, which often takes a long time and prolongs patient suffering. The aim of this research is to find out whether it is possible to predict how well a given patient will respond to a particular medication, without having to go through a trial period of treatment. This will be achieved by measuring a number of things before treatment is given, and then assessing which of these measures best predicts how effective the treatment is. The researchers will carry out brain scans and acquire demographic, clinical and psychological information in a large group of patients who have not been treated before. The researchers expect that these measures can help predict how well a patient will respond to antipsychotic medication, and that some quite simple things that are easy to assess may be as useful as more complicated measures like brain scans. The research should help doctors choose the most suitable treatment for each patient they see.

Although antipsychotic drugs are effective in the treatment of schizophrenia, the response to medication varies considerably from one patient to another, both in terms of therapeutic and adverse effects, and is unpredictable. As a result, finding the most suitable antipsychotic for a given patient often involves a trial-and-error approach which prolongs patient suffering and increases health care costs. There is therefore an urgent need for a better understanding of how we can best predict response to treatment in a given patient. Neuroimaging measures may be useful in predicting antipsychotic response, but scanning is a relatively expensive investigation that is not routinely used in clinical practice. Demographic, clinical and cognitive measures may provide a more practical, cost-effective alternative, but their predictive value at an individual level has yet to be tested. The aim of the present project is to examine how to best predict the response to antipsychotic medication in a given patient with first episode schizophrenia. We will compare three classes of predictors: (i) structural MRI scans; (ii) demographic and psychopathological measures which are routinely acquired in a clinical setting; (iii) cognitive measures which are not part of a routine clinical assessment, but could be acquired at relatively little extra cost. Two hundred patients with first episode schizophrenia will undergo a structural Magnetic Resonance Imaging (MRI) scan and receive a comprehensive demographic, clinical and cognitive assessment. All of them will be given the same antipsychotic (amisulpride) and will be clinically assessed over a period of 4 weeks; this will allow us to examine treatment response, both in terms of symptom remission and side-effects. We will then use Support Vector Machine to investigate the value of pre-treatment neuroanatomical, demographic, clinical and cognitive information for predicting treatment response. Our main hypothesis is that the combination of demographic, psychopathological and cognitive data will provide comparable predictive power to neuroimaging data. If so, the former would represent a practical, cost-effective alternative to brain scans. The results will have direct implications for clinical practice and their translational implementation will be greatly facilitated by King s Health Partners, a recently formed Academic Health Science Centre. The project will provide excellent value for money to the MRC as the costs of subject recruitment and data acquisition are already covered by an ongoing project.