T90. NEUROBIOLOGICAL SIGNATURES OF ANTIPSYCHOTIC TREATMENT RESPONSE – A COMBINED MAGNETIC RESONANCE SPECTROSCOPY AND RESTING STATE CONNECTIVITY STUDY IN SCHIZOPHRENIA

BACKGROUND: Approximately 30% of patients with schizophrenia (SZ) do not improve with antipsychotic drug (APD) treatment and 60% show sub-optimal response. Our prior studies have showed that hippocampal regional cerebral blood flow (rCBF) is elevated in unmedicated patients and that rCBF reduction is correlated with good treatment response (Lahti, 2009). In addition, we found elevated hippocampal glutamate in unmedicated (Kraguljac, 2013) but not in medicated patients (Kraguljac, 2012). To characterize this pattern, we conducted a longitudinal study where hippocampal neurochemistry and function were measured while patients were unmedicated and after a 6-week course of APD. METHODS: Thirty-four unmedicated SZ were enrolled in this multimodal neuroimaging study. They were scanned while unmedicated and after a 6-week treatment with risperidone. We obtained resting state fMRI scans (TR/TE= 2100/30msec, flip angle= 70, 4mm slice thickness, 1mm gap, 150 acquisitions) and MR spectroscopy scans from a left hippocampus voxel (TR/TE= 2000/80ms; 640 averages, 2.7x 1.5x 1cm). Treatment response criteria were based on a >50% improvement on the Brief Psychiatric Rating Scale (BPRS) psychosis subscale. Seventeen patients met this criterion (10 male; mean age 28.53 years), and 17 did not (14 male; mean age 29.94 years). Using a left hippocampus seed region, functional connectivity maps to the whole brain were generated in the CONN toolbox. MRS data were quantified using jMRUI. We compared hippocampal connectivity between good and poor responders at baseline and examined whether hippocampal functional connectivity was predictive of subsequent treatment response. A group x time (baseline vs week 6) interaction was used to assess group differences longitudinally. We examined the correlation between treatment response and change in hippocampal connectivity. Connectivity analyses were corrected for age as appropriate and corrected for multiple comparisons using voxel height threshold (p