Tracer kinetic analysis of the SV2A ligand ¹¹C-UCBA as a PET marker for synaptic density in humans

Objectives: Quantitative imaging of the synaptic vesicle glycoprotein 2A (SV2A) with PET can be used as a measure of synaptic density in the human brain (Finnema et al, Science Tr Med 2016), changes of which occur in many neurodegenerative diseases. ¹¹C-UCBA has previously been validated as an SV2A tracer in pigs (Estrada et al, Nucl Med Biol 2016), showing dose-dependent blocking and reversible binding. The aim of the present work was to evaluate tracer kinetic models and simplified methods for quantification of synaptic density using ¹¹C-UCBA in humans.

Methods: Eight subjects (6 epilepsy patients, 2 controls) underwent 90 min PET scans starting with injection of 5 MBq/kg ¹¹C-UCBA on a time-of-flight integrated PET-MR scanner (Signa PET-MR, GE Healthcare). Arterial blood was withdrawn for measurements of whole blood and plasma concentrations and metabolite analysis. Images were reconstructed using zero-echo-time MR-based attenuation correction, accounting for bone attenuation. A probabilistic VOI template was defined on a T1-MRI image, acquired during the PET scan, and transferred to the dynamic PET images. A centrum semiovale VOI was drawn as potential reference tissue. Data were analysed using single-tissue (1T2k), two-tissue irreversible (2T3k) and reversible (2T4k) models, as well as the simplified reference tissue model (SRTM) and plasma- and reference-Logan methods, resulting in total distribution volume (V_T) and binding potential (BP_ND) values, with binding potential both estimated directly and as distribution volume ratio to centrum semiovale (DVR). The optimal compartment model was determined using the Akaike information criterion (AIC). Standardized uptake value ratios (SUVR) at various time points were compared to modelling outcomes using regression analysis.

Results: Plasma and brain kinetics of ¹¹C-UCBA were slow, with peak activity in brain at 70-80 min. Parent fraction was approximately 50% at 90 min. Plasma-input data were best described using the 2T4k model, but this could often not provide robust V_T or BP_ND values. Mean plasma-Logan V_T was 24±17. Plasma-Logan DVR using centrum semiovale as reference tissue correlated well with 2T4k DVR (R² 0.94) for those regions where robust DVR values could be determined. Reference-Logan DVR showed good correlation with plasma-Logan DVR (R² 0.72). Plasma-Logan DVR-1 and SUVR-1 images are shown in Figure 1. SUVR for the 40-60 and 70-90 min intervals correlated well with reference-Logan DVR (R² 0.92 and 0.98).

Conclusion: Slow kinetics of ¹¹C-UCBA resulted in poor robustness of outcome parameters of reversible compartment models. However, reference-Logan DVR correlated well with plasma-Logan DVR. SUVR at 70-90 min p.i. correlated well with DVR and may be used as a simplified measure of synaptic density using ¹¹C-UCBA. Research Support: Uppsala County Council

• Download figure
• Open in new tab
• Download powerpoint