Identifying improved TSPO PET imaging probes through biomathematics: The impact of multiple TSPO binding sites in vivo

To date, 11C-(R)-PK11195 has been the most widely used TSPO PET imaging probe, although it suffers from high non-specific binding and low signal to noise. A significant number of 2nd generation TSPO radioligands have been developed with higher affinity and/or lower non-specific binding, however there is substantial inter-subject variation in their affinity for the TSPO. TSPO from human tissue samples binds 2nd generation TSPO radioligands with either high affinity (high affinity binders, HABs), or low affinity (LABs) or expresses both HAB and LAB binding sites (mixed affinity binders, MABs). The expression of these different TSPO binding sites in human is encoded by the rs6971 polymorphism in the TSPO gene. Here, we use a predictive biomathematical model to estimate the in vivo performances of three of these 2nd generation radioligands (18F-PBR111, 11C-PBR28, 11C-DPA713) and 11C-(R)-PK11195 in humans. The biomathematical model only relies on in silico, in vitro and genetic data (polymorphism frequencies in different ethnic groups) to predict the radioactivity time course in vivo. In particular, we provide estimates of the performances of these ligands in within-subject (e.g. longitudinal studies) and between-subject (e.g. disease characterisation) PET studies, with and without knowledge of the TSPO binding class. This enables an assessment of the different radioligands prior to radiolabelling or acquisition of any in vivo data. The within-subject performance was characterised in terms of the reproducibility of the in vivo binding potential (%COV[BPND]) for each separate TSPO binding class in normal and diseased states (50% to 400% increase in TSPO density), whilst the between-subject performance was characterised in terms of the number of subjects required to distinguish between different populations. The results indicated that the within-subject variability for 18F-PBR111, 11C-PBR28 and 11C-DPA713 (0.9% to 2.2%) was significantly lower than 11C-(R)-PK11195 (16% to 36%) for HABs and MABs in both normal and diseased states. For between-subject studies, sample sizes required to detect 50% differences in TSPO density with the 2nd generation tracers are approximately half that required with 11C-(R)-PK11195 when binding class information is known a priori. As binding class can be identified using a simple genetic test or from peripheral blood assays, the combination of binding class information with 2nd generation TSPO imaging data should provide superior tools to investigate inflammatory processes in humans in vivo.