A₁ adenosine receptor PET using [¹⁸F]CPFPX: Displacement studies in humans

Abstract

Background.

Imaging of cerebral A₁ adenosine receptors (A₁AR) with positron emission tomography (PET) has recently become available for neurological research. To date, it has still not been unraveled if there is a valid reference region without specific radioligand binding that may be used to improve image quantification. We conducted in vivo displacement studies in humans to elucidate this important question using the A₁AR ligand [¹⁸F]CPFPX.

Methods.

Five healthy male volunteers underwent [¹⁸F]CPFPX bolus/infusion PET with short infusion of unlabelled CPFPX as competitor (n = 4; 0.9 to 4.0 mg) or vehicle (n = 1; control condition) after equilibrium of [¹⁸F]CPFPX distribution was attained.

Results.

Infusion of CPFPX induced a rapid displacement of [¹⁸F]CPFPX binding in all regions, including the cerebellum (region with lowest binding). Even at the highest competitor dose, no full displacement was reached. Displacement was dose-dependent in all regions except the cerebellum where floor effects and/or noise might have obscured dose dependency. Specific binding was estimated to account for about one third and two thirds of total equilibrium uptake in cerebellum and cortex, respectively.

Conclusions.

Although the cerebellum is the region with lowest in vivo [¹⁸F]CPFPX binding, it is not an ideal reference region devoid of specific binding. Nevertheless, as will be discussed, the use of a reference region analysis may be a useful, non-invasive alternative analysis method in carefully selected applications.

Introduction

In the recent years, several radioligands have been proposed for in vivo positron emission tomography (PET) imaging of cerebral A₁ adenosine receptors (A₁AR) (for review see Holschbach and Olsson, 2002), which constitute an attractive target for investigating the neuromodulatory actions of adenosine (for review see Fredholm et al., 2005, Cunha, 2001, Haas and Selbach, 2000). The most successfully employed ligands are [¹⁸F]CPFPX (Holschbach et al., 2002) and [¹¹C]MPDX (Noguchi et al., 1997). The equilibrium total distribution volume (DV_t) of the ligand in cerebral tissue has been proposed as a PET outcome measure that is directly related to the total density of A₁AR available for binding (B_max′) (Meyer et al., 2004; Fukumitsu et al., 2005). Albeit DV_t offers the advantage that it can be reliably estimated using various methods (Meyer et al., 2004, Meyer et al., 2005a, Meyer et al., 2005b, Fukumitsu et al., 2005, Naganawa et al., 2005), even on a voxel-level (Meyer et al., 2005b, Meyer et al., 2006, Naganawa et al., 2005), it suffers two major drawbacks: first, the DV_t summarizes both the specific distribution volume (DV_s, also termed binding potential (BP) (Mintun et al., 1984), equal to B_max′/K_D with K_D being the equilibrium dissociation rate) and the distribution volume of free and non-specifically bound ligand (DV_f + ns). The relative contribution of DV_s to DV_t needs to be known for accurate interpretation of DV_t as a measure of A₁AR density. Second, the determination of DV_t relies on a cumbersome and error-prone measurement of the individual plasma input function, either arterially (Meyer et al., 2004, Fukumitsu et al., 2005) or venously (Meyer et al., 2005a, Meyer et al., 2005b). These drawbacks can be avoided if a valid reference region without specific binding is available and if DV_f + ns is the same in all regions: the DV_t estimate of the reference region could be employed to calculate DV_s in all other regions (either by direct calculation or by parameter constraints in non-linear compartment model fitting). Alternatively, a reference region analysis without blood sampling (for instance Logan's non-invasive graphical analysis, NIGA; Logan et al., 1996) may be used for non-invasive estimation of the DV_s/DV_f + ns ratio which is another convenient outcome measure related to B_max′ (also termed BP₂, equal to f₂·B_max′/K_D with f₂ being the fraction of free ligand in tissue (Abi-Dargham et al., 1994)).

It is widely known that there are considerable species differences in regional A₁AR expression among various rodents, cats, primates and humans. These differences are particularly striking in the cerebellum, which in agreement with relatively low receptor binding in in vitro autoradiographic studies (Fastbom et al., 1987a; Svenningsson et al., 1997; Bauer et al., 2003) is the region with the lowest DV_t in human A₁AR PET studies (Bauer et al., 2003, Meyer et al., 2004, Fukumitsu et al., 2005) and therefore has been proposed as a possible reference region (Kimura et al., 2004, Naganawa et al., 2005). In contrast to humans, compared to the A₁AR density in cerebral cortex, the A₁AR density of the cerebellum is higher in rats, comparably high in mice and (slightly) lower in guinea pigs and cats (Fastbom et al., 1987b). Furthermore, A₁AR PET studies in monkeys demonstrated comparable cerebellar and cortical ligand uptake (Wakabayashi et al., 2000, Herzog et al., 2003). This underlines the necessity to conduct in vivo displacement studies in humans to answer the important question of whether there is a valid reference region for A₁AR PET studies in humans. This aim was pursued in the present study using bolus/infusion (B/I) [¹⁸F]CPFPX PET scanning (Meyer et al., 2005a).