Stefano Zamuner and Roberto Gomeni
GlaxoSmithKline - Clinical PK/Modelling and Simulation, Verona, Italy
Background: 11C is the most used labelling ligand despite its short half life (20 min). As a consequence no more than 90 minutes of PET scans can be followed because of tracer degradation. This limitation may constitute a relevant issue in the assessment of receptor occupancy in the presence of slow ligand-receptor kinetics.
Objectives: Assess the impact of possible bias in the estimation of 90% receptor occupancy after drug challenge when 90 minutes vs 240 scan time is considered for a compound showing slow ligand-receptor kinetics. The simulation study conducted also aimed to evaluate the impact of increasing measurements error, variability (inter and intra subject) and change in cerebral blood flow.
Methods: Reference time activity curves were generated using a standard compartmental model for ligand-receptor binding [3]. The model accounted for arterial input function, cerebral blood flow influence [4,5] and ligand-binding parameters (kon, koff, Bmax). A slow ligand-receptor dissociation was considered (kon = 0.04 min–1, koff = 0.008 min–1, Bmax = 1 nM) compared with other well characterised data from 90 minutes PET scans ([11C]FLB 457: kon = 0.05 and koff = 0.05, ([11C]raclopride kon = 0.01 and koff = 0.1). In the simulation study, the equilibrium condition of the tracer was not properly reached in 90 minutes. Data were generated up to 240 min following tracer and cold treatment in 40 subjects, considering a measurement error of 2-5%, an intra-subject variability of 5-15% and inter-subjects variability of 10-30%. The impact of a 50% change on cerebral blood flow on the binding parameters was evaluated. Binding parameter estimates were obtained using both arterial input model [1] and simplified reference tissue model [2] based on the presence of a region without specific receptors. The criterion used to evaluate the results was based on the quantification of bias. All the analyses were made using NONMEM V.
Results: No bias was observed in both models using the entire data set (measurement up to 240 minutes). A statistical significant bias was observed using data up to 90 min irrespective of the model used. The bias increased with both the measurements error and the inter-intra subject variability. In any case, the range of bias in RO estimation (90%) was negligible (0.17- 1.73%). No change in the binding parameters and RO estimation with an increase of cerebral blood flow (50%) was observed.
Conclusions: The simulation study indicated that, despite the presence of bias, the RO estimation, in a full occupancy scenario, was not significantly affected (< 2%) by a reduction of PET time scan. The use of simulation in PET data modelling constitutes a powerful tool to better understand and interpret the possible impact of experimental design (PET scan duration) or change in physiological parameters (cerebral blood flow) in the evaluation of the binding parameters.
References:
[1] Perlmutter et al. J Cereb Blood Flow Metab. 1986; 6:154-69.
[2] Lammertsma et al. Neuroimage. 1996; 4: 153-8.
[3] Mintum et al. Ann. Neurol. 1984; 15:217-227.
[4] Crone. Acta physiol scand. 1964; 58: 295-305.
[5] Renkin. Am J Physiol. 1959; 197:1205-1210.
Reference: PAGE 12 (2003) Abstr 439 [www.page-meeting.org/?abstract=439]
Poster: poster