Background
We have proposed reference tissue-based linear analysis without arterial blood sampling for measurement of brain AChE activity using N-[11C]methylpiperidin-4-yl acetate ([11C]MP4A), where cerebellum or striatum that have high AChE activity was used as a reference 1 . The striatal reference analysis was also applicable for N-[11C]methylpiperidin-4-yl propionate ([11C]MP4P) 2 . In this study, characterization and optimization of this method, including assessment of regions where precise k3 (an index of AChE activity) would be measurable, were performed by computer simulation.
Methods
Monte Carlo simulations were performed to assess precision and bias of k3 estimates obtained from cerebellar or striatal reference analysis of [11C]MP4A and striatal reference analysis of [11C]MP4P. The time-radioactivity curves were derived based on full compartment model with fixed input function and fixed kinetic parameter except k3 for each tracer. The target k3 values were varied so as to include regions with low (cerebral cortex) and moderate (thalamus) AChE activity. The reference k3 was estimated from reported in vitro AChE activity ratio across temporal cortex/cerebellum/striatum and normal temporal k3 value. PET data error was estimated as described previously 3 , where a region-of-interest volume of 1 to 5 mL was used, and 100 data sets were generated for each target k3 condition.
Results
The COV (coefficient of variation) of k3 in each linear analysis was less than 20% across the k3 range tested for each tracer (corresponded to regions with low-to-moderate AChE activity), although gradually increased with increasing k3. The k3 bias in striatal reference analysis of [11C]MP4A remained within ±5% across the k3 range tested. The cerebellar reference analysis of [11C]MP4A gave a minus-bias that gradually increased (absolute value) with increasing k3, reaching about −15% in the thalamus. In the case of [11C]MP4P, the striatal reference analysis gave small minus-biases (within 5%) to k3 in the cerebral cortex and thalamus. When the PET data error was added to reference time-radioactivity data as well as target data, the COV of k3 was slightly increased. The error-added reference curve was also fitted by biexponential function; the COV was improved without significant change in bias. The nonlinear calculation in the reference tissue-based analysis, using another operational equation derived without linear transformation, was also available; the k3 estimates were also subjected.
Conclusions
The cerebellar or striatal reference analysis of [11C]MP4A and straital reference analysis of [11C]MP4P gave acceptable bias and COV to k3 in the regions with low-to-moderate AChE activity. Since the k3 value of [11C]MP4P is relatively small compared to the value of [11C]MP4A, more precise k3 estimation was expected in the thalamus by using [11C]MP4P. The use of fitted reference data did not affect the bias but improved COV. The fitting procedure would be useful, though the raw reference data was used in the original protocol 1 .