PET Image Acquisition and Analysis

PET images were acquired using a Siemens Biograph HiRez XVI PET scanner (Siemens Healthcare). A low-dose computerized tomography scan was performed for attenuation and model-based scatter correction, followed by the injection of a single intravenous bolus of 0.020–0.029 micrograms/kg [¹¹C]-(+)-PHNO. Dynamic emission data were acquired continuously for 90 min after the administration of the radiotracer. The dynamic images were then reconstructed using a filtered back-projection algorithm into 31 frames (8 × 15, 3 × 60, 5 × 120, and 15 × 300 s) with a 128 matrix, a zoom of 2.6 and a transaxial Gaussian filter of 5 mm.

PET images were analyzed using MATLAB version 2015b (Mathworks, Inc.) and MIAKAT (MIAKAT release 4.2.6, www.miakat.org; Gunn et al. 2016). An automatic pipeline was used to obtain an individual parcellation of the brain into regions of interest in MNI space, including the whole striatum and its functional subdivisions as defined by the Martinez atlas (Martinez et al. 2003). A 0–10 min [¹¹C]-(+)-PHNO binding template was nonlinearly coregistered with the 0–10 min summed PET image of each participant using Statistical Parametric Mapping (SPM8 – Wellcome Trust Centre for Neuroimaging). The template was created from an internal library of [¹¹C]-(+)-PHNO PET scans in healthy volunteers and normalized by individual structural MRI into standard space. A frame-by-frame registration process on a single frame of reference was used for motion correction for dynamic PET images. Individual averaged PET images were then coregistered to the 0–10 min summed PET image using rigid body coregistration. The deformation parameters from each participant’s 0–10 min [¹¹C]-(+)-PHNO binding template were applied to the Martinez striatal atlas, which defines the anatomical extents of the limbic, associative and sensorimotor striatum in MNI space (Martinez et al. 2003), bringing the atlas into the individual participant's space, before it was resliced to PET resolution. Regional time activity curves (TACs) were obtained by applying the individual parcellations to the realigned dynamic images. The whole cerebellum, defined using a standard cerebellum atlas (Tziortzi et al. 2011), was used as a reference region due its low density of dopaminergic receptors (Kumakura and Cumming 2009; Egerton et al. 2010). Our outcome measure of interest was nondisplaceable binding of [¹¹C]-(+)-PHNO (BP_ND):

where is the proportion of dopamine 2/3 receptors available to be bound by PHNO (i.e., the fraction of receptors not bound by endogenous synaptic dopamine), the free fraction of PHNO in the brain, and the affinity of ligand for the target. BP_ND for the whole and functional striatal subdivisions was obtained by kinetic modeling with a simplified reference tissue model (Lammertsma and Hume 1996; Gunn et al. 1997). Once the individual BP_ND maps were obtained, they were then warped back to MNI space using the inverse transformation of the initial nonlinear coregistration.

We defined the limbic striatal subdivision as our region of interest as there is a small amount of evidence from animal studies (Haluk and Floresco 2009; Ding et al. 2014) and human fMRI (Schwartenbeck et al. 2015) that it might be most important in controlling policy precision.