Practical Methods for Estimating Metabolic Flux (Ki) to Assess Response to Therapy via Static PET Scans.

Abstract

320 Introduction: We have recently developed [18F]MNI-1126, as an 18F analog of the established [11C]UCB-J PET radioligand, for imaging synaptic vesicle glycoprotein 2A (SV2A). The tracer was previously tested in non-human primates, which demonstrated its specificity for SV2A as well as comparable imaging characteristics with [11C]UCB-J. We herein report a pilot study with [18F]MNI-1126 performed to further assess its in vivo characteristics in human brain, and to examine its capacity to measure differences in SV2A density between healthy subjects and those diagnosed with Parkinson disease (PD), and Alzheimer disease (AD). Methods: Brain PET imaging was performed in 11 human subjects, 4 HV (males, 53 ± 3 y), 4 PD (3 males, 1 female, 70 ± 7 y) and 3 AD (1 male, 2 females, 67 ± 10 y, MMSE 3,14 and 22). Dynamic image data were acquired for up to 180 min on a Siemens ECAT HR+ scanner following a 3 min bolus injection of [18F]MNI-1126 (297.7 ± 58.3 MBq). Arterial blood sampling with associated metabolite assays was acquired and enabled the generation of an arterial input function for kinetic modeling. Dynamic images were motion corrected and normalized into MNI space. Regions of interest were defined by Hammers, N30R83 atlas and intersected with gray and white matter segments. Regional brain time-activity curves (TACs) were generated and analyzed by one-tissue (1T), two-tissue (2T) compartment models, and Logan graphical analysis (LGA) to estimate regional volumes of distribution (VT). A binding potential (BPND) was estimated by assuming that the VT of centrum semiovale (CS) was reflective of the non-displaceable signal. Non-invasive LGA (NI-LGA) and simplified reference tissue model (SRTM), with CS as reference region were also explored. Time stability of both VT and BPND outcome measures was evaluated. Cross-sectional analysis of VT and BPND were performed across the healthy, PD and AD subject groups. Results: [18F]MNI-1126 readily entered the brain and its distribution was consistent with SV2A similar to [11C]UCB-J - high brain uptake (max SUV 9-12) reached at 15-30 min after injection, moderate washout, and high specific binding (max BPND ~5.0). Unmetabolized parent fraction was 52.4 ± 15.5 %, 24.9 ± 8.5 %, and 22.6 ± 7.3 %, at 15 min, 60 min, and 120 min post injection. Fraction of unbound tracer to plasma proteins (i.e., free fraction) was 32% ± 2%.VT and BPND estimates were relatively stable for scan durations of 40 min and 60 min, respectively (~5% different from 120 min). The 1T model produced better (lower AIC) and more reliable (low SE) fits than 2T. Across methods regional BPND computed with 1T model was highly correlated (R2~ 0.99) with LGA (no bias) and NI-LGA. NI-LGA estimates presented an increasing negative bias with BPND (slope ~0.85). Lower global VT (6%, 14%) and BPND (20%, 29%) were found in PD and AD subjects, respectively, compared to HV. In AD, the largest reductions in BPND were found in hippocampus (51%), and superior lateral temporal cortex (39%). These results indicate that reductions in SV2A density in AD can be measured in brain regions besides hippocampus, which have been previously reported with [11C]UCB-J. In PD, the largest reductions in BPND were found in caudate nucleus (38%), hippocampus (33%), and thalamus (30%). Conclusions: These first-in-human studies with [18F]MNI-1126 have confirmed its favorable in vivo imaging properties. Disease specific decreases in regional SV2A density were detected and quantified in both PD and AD subjects. These encouraging proof-of-concept data provide a rationale for future multicenter clinical trials to validate the utility of SV2A PET imaging in AD, PD and other neurodegenerative disorders.