Kinetic modeling of N-[11C]methylpiperidin-4-yl propionate: alternatives for analysis of an irreversible positron emission tomography trace for measurement of acetylcholinesterase activity in human brain.

N-[11C]Methylpiperidin-4-yl propionate ([11C]PMP) is a substrate for hydrolysis by acetylcholinesterase (AChE). This work evaluates kinetic analysis alternatives for estimation of relative AChE activity using dynamic positron emission tomography (PET) studies of [11C]PMP. The PET studies were performed on three groups of subjects: (1) 12 normal volunteer subjects, aged 20 to 45 years, who received a single intravenous injection of 16 to 32 mCi of [11C]PMP; (2) six subjects, aged 21 to 44 years, who received two 16-mCi injections of [11C]PMP (baseline and visual stimulation, respectively); and (3) five subjects, aged 24 to 40 years, who received two 16-mCi injections separated by 200 minutes (baseline and after a 1-hour constant infusion of 1.5 mg of physostigmine, respectively). Dynamic acquisition consisted of a 17-frame sequence over 80 minutes. All analysis methods were based on a first-order kinetic model consisting of two tissue compartments with the parameter k3, representing PMP hydrolysis, being the index of AChE activity. Four different schemes were used to estimate k3: (1) an unconstrained non-linear least-squares fit estimating blood-brain barrier transport parameters, K1 and k2, in addition to the hydrolysis rate constant k3; (2) and (3), two methods of constraining the fit by fixing the volume of distribution of free tracer (DVfree); and (4), a direct estimation of k3 without use of an arterial input function based on the shape of the tissue time-activity curve alone. Results showed that k3 values from the unconstrained fitting and no input methods were estimated with similar accuracy, whereas the two methods using DVfree constraints yielded similar results. The authors conclude that the optimal analysis method for [11C]PMP differs as a function of AChE activity. All four methods gave precise measures of k3 in regions with low AChE activity (approximately 10% coefficient of variation in cortex), but surprisingly, with unconstrained methods yielding estimates with lower variability than constrained methods. In regions with moderate to high AChE activity, constrained methods were required to yield meaningful estimates and were superior to the unconstrained methods.