Autoradiography studies have shown cerebellum

Autoradiography studies have shown Safingol to be devoid of D1-R [40], confirming that this region may serve as reference for estimation of free and non-specifically bound radioligand concentration in tissue (= non-displaceable uptake). In initial human studies using [11C]SCH 23390 and [11C]NNC 112, quantification of binding was performed by either subtracting the total measured radioactivity in the reference region from the measured activity in the target region [41], or by calculating a ratio of specific (target − reference region) and reference region radioligand uptake at transient equilibrium [42,43]. For [11C]SCH 23390, Chan et al. then used arterial samples as input function to obtain total distribution volume (VT) and distribution volume ratios (DVR, calculated as VT divided by non-displaceable distribution volume (VND), with VT in cerebellum as an estimate of VND). High reliability was shown for DVR whereas reliability was low for VT [44]. In the same paper, non-displaceable binding potential (BPND, corresponding to ratio of specific over non-specific binding at equilibrium [45]) was estimated using a linear approach (Logan non-invasive) with cerebellar activity as input function, showing equal or higher reliability than plasma-input DVR [44]. Subsequently, kinetic modelling using the simplified tissue reference model (SRTM) with cerebellum as reference input was validated for [11C]SCH 23390 [31,46]. In test-retest studies, absolute variability for BPND obtained using this approach has shown to be 4.2-8.4% for striatal regions and 7.0-11.1% for frontal cortical regions, with intraclass coefficients (ICCs) of 0.81-0.94 and 0.81-0.97 respectively [31,47]. For [11C]NNC 112, Abi-Dargham et al. performed a test-retest study using arterial input function and kinetic modelling using the 2-tissue compartment model (2TCM) [30], evaluating a range of different outcomes measures. The test-retest variability for most parameters were similar for striatal and cortical regions, which may reflect the higher specific − to nonspecific binding ratio for this radioligand compared to [11C]SCH 23390. For VT and DVR (calculated as VT/VND, using 1TCM-derived VT values for cerebellum as estimate for VND), mean absolute variability for all regions was 8 ± 2% (standard deviation) and 6 ± 2% respectively, with ICC values of 0.91 ± 0.06 and 0.84 ± 0.10. For BPND calculated using rate constants k3/k4, absolute variability values and ICC were 13 ± 5% and 0.84 ± 0.11. The outcome variable V3 (which corresponds to specific binding over free radioligand in plasma, according to current nomenclature BPF [45]), showed similar reliability as BPND, with a mean absolute variability of 14 ± 5% and a mean ICC of 0.9 ± 0.06. To date, no formal validation of kinetic models using reference tissue as input function has been performed for [11C]NNC 112. Using both [11C]SCH 23390 and [11C]NNC 112, D1-R binding has in cross-sectional studies shown to decline by around 7–8% per decade in striatum and 8–14% in cortex [[48], [49], [50]]. In adolescents a similar rate of decrease was found for dorsal striatum (8%), whereas frontal cortical receptor levels showed a more rapid decline corresponding to around 26% per decade [51]. Hence, age is an important confounder in clinical studies, in particular when examining cortical regions in younger samples. Moreover, for both these radioligands part of the binding in cortical regions is explained by 5HT2a receptor binding, despite showing more than 100-fold higher affinity for D1-R in vitro [23]. In a study on non-human primates, where blocking was performed using the selective 5HT2a antagonist MDL 100907, both radioligands showed a 25% reduction in binding [23]. For [11C]NNC 112, this was subsequently confirmed in a human blocking study, showing that specific binding in cortical regions was reduced by 20–30%, whereas striatum showed no change [52].