TY - JOUR T1 - Quantification of PET tracer specific binding without reference tissue JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 361 LP - 361 VL - 58 IS - supplement 1 AU - Yun Zhou AU - Qian Zhao AU - James Brasic AU - Robert Dannals AU - Susanne Ostrowitzki AU - Edilio Borroni AU - Meret Martin-Facklam AU - Weiguo Ye AU - Dean Wong Y1 - 2017/05/01 UR - http://jnm.snmjournals.org/content/58/supplement_1/361.abstract N2 - 361Objectives: It is still challenging to quantify PET tracer specific binding reliably without reference tissue. The objective of the study is to evaluate a physiological constraint kinetic modeling approach to estimate [11C]RO5013853 binding potential (BPND) in human brain glycine transporter type 1 (GlyT1) PET imaging study without identified reference tissue.Methods: A standard sequential two-tissue compartment model, free plus nonspecific binding and specific binding compartments, with dynamic PET and plasma input was used to quantify [11C]RO5013853 nonspecific and specific cerebral binding. Two dynamic PET scans with plasma input function measurements were collected for all 14 subjects: one is baseline and one post glycine reuptake inhibitor bitopertin dosing of 5 groups (5, 15, 30, 60, 175 mg/per day for 10-12 days). Twelve regions of interest (ROIs) including high and low GlyT1 density brain regions were manually drawn on co-registered structural MRIs. A constraint of all ROIs in two scans of same free-plus-nonspecific distribution volume (VND) and rate constant from specific binding to free-plus-nonspecific binding compartment (k4) (2-scan constraint) was implemented in model fitting with a nonlinear regression algorithm. For comparison, model fitting without constraint and with constraint for each single scan separately (1-scan) was also applied to the same data set. Model-independent Logan plot to estimate total distribution volume VT and Lassen plot to estimate occupancy and VND from all ROI VTs in two scans were also performed for reference.Results: There were highest coefficients of variation with a number of outliers in VND, VT, and BPND estimates from the compartmental model fitting without constraint. There were no significant difference between baseline scan and blocking in the VND and k4 estimates from 1-scan constraint model fitting (p>0.05). The BPNDsat baseline estimated from the 2-scan constraint were (5.83 ± 1.31, 4.58 ± 1.40, 3.45 ± 1.86, 1.81 ± 0.59, 1.57 ± 0.62, 1.27 ± 0.49, 1.24 ± 0.44, 1.16 ± 0.45, 1.13 ± 0.44, 1.11 ± 0.39, 1.05 ± 0.41, 1.02 ± 0.39) for (pons, thalamus, cerebellum, putamen, caudate, occipital, cingulate, parietal, orbital frontal, lateral temporal, superior frontal, and prefrontal), respectively, and the corresponding nonspecific binding VND estimates was 0.40 ± 0.05 ml/cc (mean±SD, n=14). The 2-scan constraint model fitting method provided the best correlations between the percent change of VT or BPND and blocking dose as compared to the other 3 referred estimates. The VT estimates from Logan plot and the whole brain percent change of BPND in blocking scan were comparable to the ones for the 2-scan constraint method.Conclusion: The simultaneous 2-scan based compartmental model fitting with physiological constraint provided reliable estimates specific and nonspecific binding in [11C]RO5013853 human dynamic PET study. The application of the method to other radioligand-receptor imaging such as (R)-11C-rolipram PET is under investigation. Research Support: ER -