Measurement of translocator protein in white matter using [11C]PBR28

402 Objectives White matter inflammation is involved in several brain disorders, including multiple sclerosis, traumatic brain injury, and HIV encephalitis. Changes in translocator protein (TSPO), the marker for inflammation, have been reported in white matter by in vitro binding (Gulyas, et al. 2009), immunostaining (Cosenza-Nashat, et al. 2009), and PET measurement of total distribution volume (VT) (Suridjan, et al. 2015). However, the percentage of VT that is specifically bound to TSPO in white matter remains unknown. This study performed a Lassen plot using data from healthy subjects with two affinity types (high- and mixed-affinity binders) to measure nondisplaceable distribution volume (VND), and then calculated specific distribution volume (VS). Because some white matter areas are adjacent to gray matter and CSF, partial volume correction was performed to correct for spill in and out. Methods Fourteen healthy subjects (eight high-affinity (HABs) and six mixed-affinity binders (MABs), 48±17 y.o.) had a 90-minute 11C-PBR28 PET scan using Advance scanner (GE) with measurement of metabolite-corrected arterial input function. All subjects had a structural MRI scan using Biograph mMR (Siemens). VT was calculated using Logan plot and unconstrained two-compartment model. Volumes of interest (VOI) were defined based on MRI by PNEURO/PMOD, which uses segmentation, spatial normalization, and N30R83 Hammers atlas. These VOIs included three white matter VOIs: the entire cerebral white matter, the entire cerebellar white matter, and the corpus callosum. The entire cerebral white matter was studied because white matter lesions are often widespread. The corpus callosum was studied because it is often damaged by traumatic brain injury. For these VOIs and CSF, geometric partial volume correction (PVC) (Rousset et al. 1998) was applied. In addition, for both cerebral and cerebellar white matter, a shrunk version was prepared by thresholding at 99% probability of white matter and applying erosion. Using all these VOI data, VND was estimated by Lassen plot by plotting the data of HABs vs. (HABs – MABs). VS was calculated by subtracting VND from VT. Results: The Lassen plot gave a VND of 1.72 and 1.65 for data with and without PVC, respectively, with narrow 95% confidence intervals of 1.45-1.94 and 1.44-1.83, respectively. The average VS in white matter areas was ~90% of that in gray matter (Table). VS in the shrunk versions of cerebral white matter and corpus callosum were ~70% of that in gray matter. PVC made little difference in VS. Shorter durations of data yielded slightly smaller VT in white than in gray matter. In white matter areas, VT from 60-minute data was 14% smaller than that from 90-minute data. In gray matter areas, VT from 60-minute data was 12% smaller than that from 90-minute data. Unconstrained two-compartment model did not identify VT well, with %SE > 10% in 21% of VOIs, mostly in small gray matter areas. Conclusions:11C-PBR28 had measurable specific binding in white matter. The ratios of VS between gray matter and the shrunk version of cerebral white and corpus callosum were in line with the ratios of VT for 18F-FEPPA (Suridjan, et al. 2015) and of in vitro binding (Gulyas, et al. 2009). Slightly worse time stability of VT in white than in gray matter—that is, differences in VT obtained from full and shortened data—may be due to the greater influence of radiometabolites or noise in white matter. The time stability needs to be investigated in order to compare TSPO in the cerebral white and corpus callosum within subjects or between groups. Abbreviations: PVC: partial volume correction; HAB: high-affinity binder; MAB: mixed-affinity binder; WM: white matter