Using neuroimaging to predict clinical outcomes in the early phase of psychosis
Poster C93, Saturday, October 22, 11:30 am - 1:00 pm, Le Baron
philip mcguire1, paul allen2, howes oliver1, egerton alice1, kempton matthew1, demajaha arsime1, juhar sameer1, valmaggia lucia1, dazzan paola1, bossong matthias3, modinos gemma1, mechelli andrea1; 1Department of Psychosis Studies, Institute of Psychiatry, Psychology & Neuroscience, King's College London, 2Department of Psychology, University of Roehampton, 3University Medical Center, Utrecht
A fundamental problem in the management of psychiatric disorders is that it is difficult to predict important clinical outcomes on the basis of the patient’s clinical features and history. For example, it is not possible to predict whether an individual at high risk for psychosis will subsequently develop the disorder or will recover. Similarly, when a patient presents with first episode psychosis, there is no way of knowing if treatment with antipsychotic medication will be effective. However, data from neuroimaging studies suggest that measures of brain structure, function, and chemistry may be able to differentiate subjects with similar presenting features but distinct clinical outcomes. In subjects at high risk, the later onset of psychosis has been linked to alterations in hippocampal volume and activity, and to subcortical dopamine function. In patients with first episode psychosis, a poor response to antipsychotic treatment has been associated with relatively normal subcortical dopamine function but elevated regional glutamate function. A key challenge in translating these research findings into clinical tools is to develop statistical methods that permit predictions about outcomes on the basis of neuroimaging data from an individual subject. Machine learning is one approach that has the potential to address this issue, and clinical tools that employ such methods are currently being evaluated in large, multi-centre studies in high risk and first episode patients. These tools may also incorporate genomic and cognitive data, as well as proteomic, metabolomics, inflammatory and immune biomarkers from peripheral blood.
Topic Area: Translational Research