Assessment of a white matter reference region for 11C-UCB-J PET quantification

Abstract

11C-UCB-J is a positron emission tomography (PET) radioligand that has been used in humans for synaptic vesicle glycoprotein 2A (SV2A) imaging and as a potential synaptic density marker. The centrum semiovale (CS) is a proposed reference region for noninvasive quantification of 11C-UCB-J, due to negligible concentrations of SV2A in this region in baboon brain assessed by in vitro methods. However, in displacement scans with SV2A-specific drug levetiracetam in humans, a decrease in 11C-UCB-J concentration was observed in the CS, consistent with some degree of specific binding. The current study aims to validate the CS as a reference region by (1) optimizing CS region of interest (ROI) to minimize spill-in from gray matter with high radioactivity concentrations; (2) investigating convergence of CS ROI values using ordered subset expectation maximization (OS-EM) reconstruction, and (3) comparing baseline CS volume of distribution (VT) to nondisplaceable uptake in gray matter, VND. Improving ROI definition and increasing OS-EM iterations during reconstruction decreased the difference between CS VT and VND. However, even with these corrections, CS VT overestimated VND by ∼35-40%. These measures showed significant correlation, suggesting that, though biased, the CS may be a useful estimate of nondisplaceable uptake, allowing for noninvasive quantification for SV2A PET.

Keywords: Ordered subset expectation maximization reconstruction; positron emission tomography; reference region; synaptic density; synaptic vesicle glycoprotein 2A.

Figures

Figure 1.

Optimization of the centrum semiovale reference region. (a) VT values from 47 subjects using centrum semiovale (CS) regions averaged across multiple subjects in AAL space (CS_AAL), regions defined from each subjects’ individual probability maps with 2-, 4-, and 6-mL volumes (CS_2, CS_4, and CS_6), and the original CS region that was manually drawn on the MNI template, eroded by two voxels in each direction (CS_orig). Error bars indicate 1 standard deviation from the mean. CS_orig (red diamonds) and CS_AAL (green circles) ROIs overlaid on a representative subject’s MRI (b) and 11C-UCB-J scan (c, SUV, units: g/mL).

Figure 2.

(a) Baseline and (b) blocking parametric VT images from a representative 11C-UCB-J scan. (c) Lassen Occupancy plots using VT values from two (blue circles), three (red squares), and four (green triangles) iterations of the representative subject.

Figure 3.

(a) Individual (N = 13) centrum semiovale (CS) volume of distribution, VT, at baseline from images reconstructed with 2, 3, and 4 OS-EM iterations of 30 subsets. Results from post-hoc testing with Dunn’s multiple comparison test are shown as p values. (b) Individual (N = 28) CS VT under blocking conditions from images reconstructed with 2, 3, and 4 OS-EM iterations. (c) Average baseline (filled circles) and blocking (open circles) CS VT at 2 OS-EM iterations. (d) Average baseline (filled circles) and blocking (open circles) CS VT at 4 OS-EM iterations. For (c) and (d), results from Mann-Whitney U test are shown as p values.

Figure 4.

(a) Average (N = 25) nondisplaceable volume of distribution, VND, estimates at 2, 3, and 4 OS-EM iterations. Results from post-hoc testing using Tukey’s multiple comparisons test resulted in significant differences between 2i and 4i, and between 3i and 4i. (b) Baseline CS VT and VND estimates at 2i30s (p < 0.0001, Mann-Whitney U test). (c) Baseline CS VT and VND estimates at 4i30s (p < 0.0001, Mann-Whitney U test). Error bars represent standard deviation from the mean.

Figure 5.

(a) Representative subject’s anatomical MRI, (b) average (n = 9) baseline parametric VT, (c) parametric images of VS, and (d) VND overlaid on the MRI.

Figure 6.

(a) Linear regression analyses of baseline CS VT and gray matter VND at 2i (R2 = 0.3915, p = 0.04). (b) Linear regression analyses of baseline CS VT and gray matter VND at 4i (R2 = 0.4142, p = 0.03). Each dot in (a) and (b) represents data from one subject. (c) Average BPND values from a subset of seven subjects in 23 ROIs, calculated using VND from blocking studies (gold diamonds, BPNDTrue), baseline CS VT at 2i (blue circles, BPND2i) and 4i (green triangles, BPND4i) as estimates of nondisplaceable 11C-UCB-J uptake.