Associations between 11C-UCB-J source networks and fMRI resting-state networks

122 Introduction: Resting-state networks (RSNs) consist of functionally connected brain regions of synchronized activity measured with fMRI. Previously, we found source networks of synaptic density based on regional covariance of 11C-UCB-J PET using independent component analysis (ICA), including distinct sources encompassing striatum and prefrontal cortex (PFC) (Unpublished). Cortico-striatal pathways are well-established neural circuits that play a critical role in modulating regional brain activity and interplay (1). The aim is to investigate whether associations exist between amplitude of RSN activity and sources of synaptic density variability identified with ICA of 11C-UCB-J PET. We investigated potential links between cortico-striatal synaptic density and RSN activity in three functional domains: default-mode, motor, and executive networks. We hypothesize that greater fractional amplitude of low-frequency fluctuations (fALFF) of RSNs would be associated with greater intensity of primary 11C-UCB-J sources encompassing associated circuitry in subcortical and PFC regions. Methods: 24 healthy participants (13M/11F, mean age: 46±15y) completed two 5-min resting-state MR scans (3T Trio, Siemens) using T2*-weighted echo-planar imaging (EPI) (multiband factor=4, TR/TE=1000/30ms, flip angle=62°, resolution=2x2x2mm3, 60 slices) and a standard high-resolution T1-weighted anatomical MPRAGE. . Subjects received an i.v. injection of 11C-UCB-J (588±159 MBq; inj. mass: 1.26±0.66 µg). Dynamic PET scans (HRRT, Siemens/CTI) were acquired in list mode (207 slices, 1.2mm slice separation, reconstructed image resolution ~3mm). Parametric volume of distribution (VT) images of 0-60 min data were generated with a one-tissue compartment model using the metabolite-corrected arterial plasma curve. Motion-corrected EPI and PET VT images were transformed into standard space using the nonlinear registration of T1 to MNI152 and smoothed with an 8mm FWHM Gaussian kernel. ICA was performed using GroupICA Toolbox (2, 3). ICA of fMRI data estimated 30 components, six RSNs linked to default mode, motor, and executive functioning were visually identified based on regional consistency with established networks (4). ICA estimated 18 components for the PET data (Unpublished) and primary sources (i.e., those capturing >2% of total variance) that encompassed striatal and PFC regions were selected. The fALFF ( 0.15 Hz) for each identified RSN of interest was correlated with identified source loadings at pFDR<0.01. Results: Selected RSNs included the two default-mode networks (DMN; anterior, posterior), two motor (sensorimotor, primary motor), and two executive networks (executive control, salience) (Fig. 1A). PET ICA identified a subcortical/striatal 11C-UCB-J source, and two sources encompassing PFC regions: a rostral medial PFC source and a dorsal medial PFC source (Fig. 1B). The striatal source correlated positively with fALFF values of the primary motor and salience networks. The rostral medial PFC source correlated positively with fALFF values of the anterior DMN, sensorimotor, and both executive function networks. No relationships were found between the dorsal medial PFC source and any RSN (Fig. 1C). Conclusions: This preliminary study suggests an association between fALFF (representing RSN activation amplitude) and subject loadings (representing components of synaptic density) that extended beyond regional overlap to implicate circuit-based relationships. Greater coherence of synaptic density in striatum was associated with higher fALFF in cortical RSNs associated with motor and executive functioning, while rostral PFC density was associated with fALFF in the spatially overlapping aDMN but also other cortical networks spanning selected functional domains. These initial findings provide insight into potential links between cortico-striatal synaptic architecture and functional systems related to diverse brain processes.