Feasibility of imaging synaptic density in the human spinal cord using 11C-UCB-J PET

112 Introduction: In vivo measurements of synaptic density in the spinal cord (SC) can provide benefits in the study of conditions such as SC injury1 and amyotrophic lateral sclerosis2. Positron emission tomography (PET) imaging with 11C-UCB-J or other SV2A PET radioligands can potentially provide information about synaptic integrity in multiple CNS conditions. However, due to the small diameter of the SC and the limited spatial resolution of most PET systems, SC imaging with PET is challenging. Here, we propose a method for imaging SV2A, a presynaptic vesicle protein, in the human SC with 11C-UCB-J PET/CT, by using the CT image to define the location of the SC. We also assess the specificity of 11C-UCB-J binding in the cervical SC through an occupancy analysis conducted on the HRRT. Methods: Four subjects underwent dynamic PET scanning with 11C-UCB-J on the Siemens Biograph mCT. A CT image was acquired prior to PET scanning for attenuation correction and anatomical definition. The CT image was used to define the SC region-of-interest (ROI) by segmenting the spine, binarizing the image to create a mask of the spine, and closing (dilate + erode) the mask. On this mask, the center of mass in x- and y- directions was calculated on each axial slice, and a circle ROI with radius of 5 pixels was generated. This ROI was used to define the SC time activity curves. In the second study, 3 subjects underwent 11C-UCB-J baseline and blocking scans on the HRRT. As the superior cervical SC was within the field of view, a cervical SC ROI was manually drawn on an early summed PET image (0-10min), and the VT within the cervical SC was estimated using the one-tissue compartment model and arterial input function for both baseline and blocking scans to measure the blockade by levetiracetam (LEV) or brivaracetam (BRV). Using the baseline and blocking VT to calculate the observed SC occupancy, VND, VS, and BPND in the cervical SC were estimated. Results: The developed algorithm successfully defined the full SC in whole-body PET images. A continuous SC is most clearly visible in the cervical and thoracic SC, where 11C-UCB-J uptake is higher than in the lumbar and sacral regions. In the automatically defined ROI, some extraneous tissue surrounding the SC was also included. In the whole-body full SC image, 11C-UCB-J uptake was low compared to the brain; the max SUV was ~ 1.5-2.0 g/cm3 compared to 7-8 g/cm3 in the brain. In the HRRT images, a continuous cervical SC was present. The average (n=3) baseline VT for cervical SC was 1.81 (± 0.07) and the average blocking VT = 1.51 (± 0.27). Across the three subjects, average gray matter occupancy was ~77%, and average cervical SC occupancy was ~17%. Using these values, we calculated a SC VND of 1.41 mL/cm3 (typical GM VND is 3.03), a VS of 0.4 mL/cm3, and a BPND of 0.28. Conclusions: Although total uptake relative to the whole brain is low, there is evidence that 11C-UCB-J specifically binds to SV2A in the human SC. The automatically defined SC region is subject to the partial volume effect, as the ROI diameter was larger than the full SC diameter (~10-12mm). Also, this ROI includes both gray and white matter in the SC. These factors can contribute to an underestimation of SV2A binding in the SC. Future directions of this work include optimizing full SC ROI definition, and expanding this analysis to nonhuman primate images acquired on the mCT (full SC) and Focus 220 microPET (cervical SC). Also, the acquisition of whole body dynamic scans with SV2A blockade in humans will be helpful in evaluating specific SV2A binding in the full SC.