Abstract
Species differences occur in the brain concentrations of drugs, but the reasons for these differences are not yet apparent. This study was designed to compare brain uptake of three radiolabeled P-glycoprotein (P-gp) substrates across species using positron emission tomography. Brain concentrations and brain-to-plasma ratios were compared; [11C]verapamil in rats, guinea pigs, and monkeys; [11C](S)-(2-methoxy-5-(5-trifluoromethyltetrazol-1-yl)-phenylmethylamino)-2(S)-phenylpiperidine (GR205171) in rats, guinea pigs, monkeys, and humans; and [18F]altanserin in rats, minipigs, and humans. The fraction of the unbound radioligand in plasma was studied along with its metabolism. The effect of P-gp inhibition was investigated by administering cyclosporin A (CsA). Pronounced species differences were found in the brain and brain-to-plasma concentrations of [11C]verapamil, [11C]GR205171, and [18F]altanserin with higher brain distribution in humans, monkeys, and minipigs than in rats and guinea pigs. For example, the brain-to-plasma ratio of [11C]GR205171 was almost 9-fold higher in humans compared with rats. The species differences were still present after P-gp inhibition, although the increase in brain concentrations after P-gp inhibition was somewhat greater in rats than in the other species. Differences in plasma protein binding and metabolism did not explain the species-related differences. The findings are important for interpretation of brain drug delivery when extrapolating preclinical data to humans. Compounds found to be P-gp substrates in rodents are likely to also be substrates in higher species, but sufficient blood-brain barrier permeability may be retained in humans to allow the compound to act at intracerebral targets.
Species differences in receptor and enzyme systems are well known, but very little is known about species differences in the transport of drug molecules across the blood-brain barrier. Transport across this barrier occurs by passive diffusion either alone or in combination with active influx or efflux. P-glycoprotein (P-gp; ABCB1; MDR1) is an ATP-dependent efflux transporter that carries a wide range of drug molecules at the blood-brain barrier as well as in other organs, such as the liver, kidney, and gastrointestinal tract. It has a profound influence on brain concentrations of its substrates. Because of its noninvasiveness, positron emission tomography (PET) is one of the few methods that allows for direct in vivo comparisons between preclinical species and humans regarding drug interactions with P-gp.
[11C]Verapamil is the most frequently used PET radioligand for P-gp studies (Hendrikse et al., 1999; Bart et al., 2003; Sasongko et al., 2005; Lee et al., 2006; Syvänen et al., 2006). Other examples include [11C]carazolol, [18F]fluorocarazolol, [11C]carvedilol, [11C]GR218231, [11C]loperamide, and [11C]desmethyl loperamide (Doze et al., 2000; Bart et al., 2005; de Vries et al., 2005; Lazarova et al., 2008; Zoghbi et al., 2008). Some radioligands shown to be P-gp substrates in rodents are nonetheless taken up by the brain in primates, indicating that there may be species differences in P-gp function. For example, Liow et al. (2007) have shown that the 5HT1A receptor antagonist [11C]RWAY is a P-gp substrate in the rat and mouse and displays a low brain uptake in these species. However, uptake of [11C]RWAY in human and monkey brains is relatively high (Yasuno et al., 2006; Zhang et al., 2007). Likewise, although uptake of the chemically related 5HT1A receptor antagonist [18F]MPPF by the human brain is relatively high, the molecule has been shown to be a P-gp substrate in rats (Passchier et al., 2000a,b). Furthermore, GR205171, an NK1 receptor antagonist, is a P-gp substrate in rodents (Rupniak et al., 2003), but uptake of radioactively labeled [11C]GR205171 is high in human and monkey brains (Bergström et al., 2000; Furmark et al., 2005; Michelgård et al., 2007). Uptake of [18F]altanserin, a 5HT2A receptor antagonist, in human, baboon, and pig brains is also relatively high (Biver et al., 1994; Staley et al., 2001). However, initial ex vivo studies of [18F]altanserin in rats have shown relatively low brain concentrations (Neurobiology Research Unit Copenhagen, Copenhagen, Denmark; M. Palner and G. M. Knudsen, unpublished data).
One explanation for the species differences in brain concentrations observed with these radioligands could be the existence of a species difference in P-gp function, despite the high degree of homology in the amino acid content of P-gp in different species (Table 1). Several in vitro studies have shown that the transport capacity of P-gp substrates differs among cell lines transfected with P-gp from different species, indicating a species difference in substrate recognition (Yamazaki et al., 2001; Ohe et al., 2003; Katoh et al., 2006; Xia et al., 2006; Baltes et al., 2007). Besides differences in substrate affinity, differences in P-gp expression could also result in differences in P-gp transport capacity. In vivo evidence for species differences in P-gp function has so far only been reported by one group, who compared rats and guinea pigs and found that higher concentrations of the P-gp inhibitor GF-120918 were needed in guinea pigs than in rats to achieve the same increase in brain concentrations of an undisclosed P-gp substrate (Cutler et al., 2006). In contrast, when Hsiao et al. (2006) compared the increase of [11C]verapamil in the human and rat brain after treatment with low doses of the potent in vivo P-gp inhibitor cyclosporin A (CsA), they did not observe any species differences (Hsiao et al., 2006). Other explanations could be species differences in blood-brain barrier transport that are not related to P-gp function, but rather to other causes, e.g., tightness of the endothelial cell tight junctions or expression of other active efflux transporters. Differences in intrabrain distribution because of differences in lipophilic content of brain tissue, and in the degree of plasma protein binding, metabolism, and elimination from plasma could also lead to different levels of brain uptake among species.
The in vivo brain uptake of [11C]verapamil, [11C]GR205171, and [18F]altanserin in several species was investigated in this study. Interspecies comparisons of brain concentrations and brain-to-plasma concentration ratios, fraction of unbound radioligand in plasma, and radioligand metabolism were carried out. The effect of P-gp was studied by administering CsA to a subgroup of animals.
Materials and Methods
General. The radioligands [11C]verapamil, [11C]GR205171, and [18F]altanserin (Fig. 1) were synthesized as reported previously (Lemaire et al., 1991; Bergström et al., 2000; Syvänen et al., 2006). All chemicals were obtained commercially and were of reagent grade or better. The P-gp inhibitor CsA (Sandimmun; Novartis, Basel, Switzerland) was diluted to the desired concentration in saline. All animal experiments were approved by the Uppsala Animal Ethics Committee (approval nos. C117/4 and C153/4) and the Danish Animal Research Inspectorate (journal 2003/561-745). The clinical data used in this study were acquired from clinical studies in healthy volunteers (Syvänen, 2003; Liptrot et al., 2004) and were performed in accordance with the Helsinki declaration. A summary of the number of subjects for each radioligand and species is given in Table 2.
Anesthesia and Surgical Procedures. Before the experiments, rats and guinea pigs were housed in ventilated cabinets at a room temperature of 20 ± 2°C and maintained on a 12-h light/12-h dark cycle. They were given unlimited access to food and water. Gaseous anesthesia with isoflurane 2.6% (Baxter Medical AB, Kista, Sweden) in a mixture of oxygen and medical air (40%/60%; AGA Gas AB, Sundbyberg, Sweden) was used for the rat and guinea pig studies with [11C]GR205171 and [11C]verapamil, whereas hypnorm/midazolam (VetaPharma Ltd., Leeds, UK/Hameln Pharmaceuticals, Hameln, Germany) was used for the studies with[18F]altanserin. After anesthesia, a femoral artery cannula was inserted for blood sampling and an intravenous catheter was inserted into the left femoral vein in guinea pigs and the tail vein in rats for radioligand and CsA administration. The [11C]GR205171 and [11C]verapamil uptake in rats and guinea pigs was studied dynamically with PET, whereas the brain uptake of [18F]altanserin was measured after euthanization of the animals (ex vivo studies).
The minipigs used for the [18F]altanserin studies were initially anesthetized with 0.15 ml/kg i.m. midazolam (Hameln Pharmaceuticals) plus Zoletil 50 vet. (Virbac Animal Health, Carros, France). One ampoule of Zoletil 50 vet (containing 125 mg each of zolazepam and tiletamine) was dissolved in 8 ml of midazolam solution (1 mg/ml). The anesthesia was maintained with intravenous propofol (1 ml/kg/h) (Fresenius Kabi AB, Uppsala, Sweden). The minipigs were intubated endotracheally and ventilated during the entire experiment. During the scan, they were placed on a heating carpet and covered to maintain body temperature. Catheters were surgically inserted in the femoral vein for radioligand and CsA administration, and into the femoral artery for blood sampling. Body temperature and physiological functions (blood pressure, blood oxygenation, and heart rate) were monitored continuously, and hematocrit, blood glucose, and acid-base parameters (pH, pCO2, pO2, H2CO3, and O2 saturation) were measured in full blood samples on an ABL system (625; Radiometer, Copenhagen, Denmark) at regular intervals. Deviations from normal values were corrected appropriately.
The cynomolgus monkeys were transported to the investigation site at Uppsala Imanet on the morning of the experiment. Venous catheters were inserted in both hind legs, and propofol (50 mg of lipuro; B. Brown Medical AB, Bromma, Sweden) was administered to induce anesthesia. The catheters were also used for administration of the radioligands and Ringer-acetate (2 ml/kg/h; Fresenius Kabi AB). The animals were kept anesthetized with 1.3 to 2.5% sevoflurane (Abbott Scandinavia AB, Solna, Sweden) via tracheal intubation during the PET scan. A venous catheter in the tail was used for blood sampling. One monkey had two PET scans and another had three. The PET scan was started when the radioligand was administered. Data from an earlier study with [11C]GR205171 in rhesus monkeys were also included in this study (Syvänen et al., 2007).
Blood (0.2 ml) was sampled 0.5, 1, 3, 5, 10, 15, 20, 30, 45, and 60 min after administration of all compounds for all species, and the radioactivity concentration in blood and plasma was measured in a well counter that was cross-calibrated with the PET system. Additional blood samples (0.3 ml) were obtained at 0, 15, 30, and 60 min for measurement of CsA concentrations in whole blood using an immunoenzymatic assay (cloned enzyme donor immunoassay; Microgenics, Ekerö, Sweden), with a limit of detection of 25 ng/ml. Three to five blood samples (1–2 ml) were obtained for metabolite analysis in minipigs, monkeys, and humans during the scan.
PET Scanners. In rodents, a microPET R4 scanner (Concorde Microsystems, Knoxville, TN) was used for [11C]GR205171 and [11C]verapamil studies. In minipigs, a combined PET/CT Discovery LS Scanner (General Electric, Milwaukee, WI) was used for [18F]altanserin studies and, in monkeys, a Hamamatsu SHR-7700 tomograph (Hamamatsu Photonics, Hamamatsu, Japan) was used for [11C]GR205171 and [11C]verapamil studies. The human [11C]GR205171 studies were performed using an ECAT EXACT HR+ scanner (Siemens/CTI, Knoxville, TN), and the [18F]altanserin studies used a GE Advance scanner (General Electric). Transmission scans, used for correction of attenuation, were run before the emission scans.
Studies with [11C]Verapamil. Six rats (mean weight, 412 ± 45 g) received [11C]verapamil as a bolus injection of 26 ± 11 MBq. Three of them received a bolus dose of 22.5 mg/kg CsA followed by a constant infusion of 7.5 mg/kg/h, starting 20 to 25 min before administration of the radioligand. The same protocol was used for six guinea pigs (mean weight, 437 ± 48 g) at a [11C]verapamil dose of 15 ± 5 MBq. One female cynomolgus monkey (weight, 3.8 kg) received a [11C]verapamil bolus injection of 68 MBq. Four and a half hours later, a 15-mg/kg CsA bolus was followed by a constant infusion of 5 mg/kg/h, started 20 to 25 min before the second [11C]verapamil bolus of 74 MBq. Regions of interest (ROIs) were drawn over the transaxial brain images in the region of the cerebellum.
Studies with [11C]GR205171. Nine rats (mean weight 400 ± 26 g) received a bolus injection of 26 ± 13 MBq [11C]GR205171. Three of them received a bolus dose of 22.5 mg/kg CsA followed by a constant infusion of 7.5 mg/kg/h, whereas three others were administered a bolus dose of 7.5 mg/kg CsA followed by a constant infusion of 2.5 mg/kg/h, starting 20 to 25 min before administration of the radioligand. Six guinea pigs (mean weight, 488 ± 58 g) received a bolus injection of 27 ± 8 MBq [11C]GR205171. Three of them received a bolus dose of 22.5 mg/kg CsA followed by a constant infusion of 7.5 mg/kg/h, starting 20 to 25 min before the radioligand administration. The first of two female cynomolgus monkeys (weights, 6.4 and 3.8 kg) received [11C]GR205171 as a single bolus of 95 MBq, and 4.5 h later they received a bolus dose of 15 mg/kg CsA and a constant infusion of 5 mg/kg, 20 to 25 min before a second bolus dose of 112 MBq [11C]GR205171. The second animal received one single bolus dose of 78 MBq [11C]GR205171. ROIs were drawn over the transaxial brain in the region of the cerebellum where the distribution of NK1 receptors is negligible (Bergström et al., 2000; Haneda et al., 2007). In the monkeys, ROIs were also drawn over the striatum in which NK1 receptors are abundant.
Studies with [18F]Altanserin. Nine rats (mean weight, 371 ± 70 g) received 8 ± 4 MBq [18F]altanserin as a bolus injection. Four of them received a 22.5-mg/kg bolus of CsA followed by a constant infusion of 7.5 mg/kg/h, starting 20 to 25 min before the administration of radioligand. Six Göttingen minipigs (mean weight, 31 ± 8 kg) received a bolus injection of 197 ± 154 MBq [18F]altanserin. One of them received a bolus of 22.5 mg/kg CsA followed by a constant infusion of 7.5 mg/kg/h starting 20 to 25 min before the radioligand administration. In rat studies, the brain was dissected at 60 min, and the radioactivity was measured in the cerebellum, in which the distribution of 5HT2A receptors is negligible, and in the frontal cortex, in which the density of 5HT2A receptors is high (Biver et al., 1994). In minipig studies, ROIs were automatically applied in the region of the cerebellum and the frontal cortex using a minipig brain atlas (Watanabe et al., 2001).
Plasma Protein Binding. The extent of plasma protein binding was determined using equilibrium dialysis. Plasma samples were obtained, 1 to 2 MBq of the radioligand was added to each sample, and 500 μl of plasma was then transferred to a dialysis chamber with a semipermeable membrane separating the other chamber containing 500 μl of phosphate-buffered saline buffer. After the chambers had been shaken in a 37°C water bath for 2 h, a sample from each chamber was removed and the radioactivity concentrations were measured. The plasma protein unbound fraction, fp, was calculated as follows: Rbuffer and Rplasma are the radioactivity concentrations in the buffer and plasma chambers, respectively. All values are based on n ≥ 3 unless otherwise stated and are reported as means and S.D. A two-tailed t test was used to test whether the difference in plasma protein binding between plasma containing and not containing CsA was statistically significant.
Metabolism. The metabolism of [11C]verapamil and [11C]GR205171 was evaluated in rat and guinea pig plasma. The metabolism of [18F]altanserin in rats has been studied previously (Tan et al., 1999). The metabolism of [11C]verapamil and [11C]GR205171 in monkey plasma and the metabolism of [18F]altanserin in minipigs was performed in parallel with the imaging studies. The methods for separating and analyzing parent and metabolite fractions of [11C]verapamil, [11C]GR205171, and [18F]altanserin have been published previously (Bergström et al., 2000; Pinborg et al., 2003; Syvänen et al., 2006).
Data Analysis. Radioactivity measurements in plasma were corrected for metabolism before analysis. The time-activity profiles in the brain and plasma were expressed using the standardized uptake values (SUVs; normalized concentrations) to enable comparison between scans. To account for individual and species differences in the plasma concentrations, which are the driving force for the brain concentrations, the brain-to-plasma ratio Kp was calculated using the SUV in plasma and brain. SUV in brain and plasma and Kp were calculated over the scanning time, one value for each PET time frame. Average SUV and Kp values for each animal were also determined from calculations of SUV and Kp every 5 min between 0 and 60 min. All values are reported as means and S.D. The effect of P-gp inhibition was calculated by comparing SUV or Kp in CsA-untreated and -treated animals. SUVincrease and Kp,increase were calculated from the average values for SUV and Kp. Kp,increase takes into account CsA-induced differences in the plasma pharmacokinetics and may therefore be a more reliable parameter than SUVincrease for assessing the change in CsA-induced P-gp inhibition. A one-tailed t test was used to see whether the increase was statistically significant.
Results
Species Differences in Radioligand Brain and Plasma Concentrations. Distribution to the brain of the three studied radioligands [11C]verapamil, [11C]GR205171, and [18F]altanserin were clearly higher in humans, monkeys, and minipigs than in rats and guinea pigs, as described with brain concentrations; SUV; and the brain-to-plasma concentration ratios, Kp (Figs. 2, 3, 4; Table 3). For [11C]verapamil, the brain distribution was highest in monkeys, followed by rats and guinea pigs, with Kp values of 4.6 to 1.1 and 0.65, respectively (Table 3). Monkeys had a 4-fold higher brain distribution ratio (Kp) than rats. The same trend was observed with [11C]GR205171 and [18F]altanserin, where monkeys and minipigs showed higher brain uptake than rodents (Table 3). PET images with [11C]GR205171 in monkey, guinea pig, and rat are shown in Fig. 5. Humans had a [11C]GR205171 Kp ratio of 30, whereas rats had a ratio of 3.4, thus an 8.6-fold higher brain distribution in humans than in rats and 3-fold higher in humans than in monkeys. The Kp value of [18F]altanserin was 4.5-fold higher in humans than in rats.
The metabolite corrected plasma SUV values after [11C]verapamil administration, describing the drug concentrations in plasma in relation to dose administered (eq. 2), were highest in guinea pigs, followed by rats and monkeys. For [11C]GR205171, plasma SUVs were highest in monkeys, followed by guinea pigs and then humans and rats. Plasma SUV values of [18F]altanserin were highest in rats, followed by humans and minipigs. Thus, in contrast to the brain concentrations of the three radioligands, no tendency toward lower concentrations in the rodents than in the higher species was observed.
Effect of CsA Induced P-gp Inhibition. The potent P-gp inhibitor CsA was administered to a subgroup of the animals (not to humans). Because plasma concentrations are the driving force for brain concentrations, it is more relevant to compare the Kp (eq. 2) than the SUV values with and without CsA treatment to obtain information on how much of the increase is because of P-gp inhibition. Kp was increased in all species after CsA administration, apart from the Kp of [18F]altanserin in the minipig (Table 4). The increase was highest in rats for all three radioligands. From a clinical point of view, where the actual brain concentrations are of interest for a given dose, the increase in SUV may also be of interest. Both Kp,increase (eq. 5) and SUVincrease (eq. 4) are therefore shown in Table 4. PET images with and without CsA treatment for [11C]GR205171 in monkey, guinea pig, and rat are shown in Fig. 5; for [11C]verapamil in the monkey are shown in Fig. 6; and for [18F]altanserin in the minipig are shown in Fig. 7.
The CsA concentrations influence the degree of P-gp inhibition. The mean CsA concentrations were 19.1 ± 9.3 μg/ml in guinea pigs and 11.7 ± 1.7 μg/ml in rats. Three of the [11C]GR205171-administered rats received a lower dose of CsA, which resulted in a mean CsA concentration of 2.2 ± 0.5 μg/ml. The two monkeys treated with CsA had somewhat different CsA concentrations, despite being dosed according to weight. The monkey receiving [11C]verapamil showed CsA concentrations ranging from 10 μg/ml at the beginning of the scan to 6.6 μg/ml at the end, whereas the monkey receiving [11C]GR205171 showed CsA concentrations ranging from 20 μg/ml at the beginning of the scan to 16 μg/ml at the end of the scan. In the [18F]altanserin-administered minipig, the CsA concentration continued to increase throughout the scan, ranging from 62 μg/ml at the start to 73 μg/ml at 60 min, i.e., it was approximately 3 times higher than in rats. In spite of this, the minipig studied did not show an increase in Kp, although the brain SUV increased almost 2-fold (Table 4).
Metabolism of the Radioligands. The metabolism of [11C]-verapamil in plasma was affected by CsA, which resulted in a lower fraction of intact [11C]verapamil in CsA-treated animals. In the monkey, half of the radioactivity originated from metabolites at 24 min in CsA-untreated animal, compared with 17 min in the CsA-treated animal. CsA did not affect the metabolism of [11C]GR205171 and [18F]altanserin. The metabolism of [11C]GR205171 was very similar in rats and guinea pigs. After approximately 40 min, half of the radioactivity in plasma from these species originated from metabolites, whereas in humans and monkeys, this point had been reached by 20 min; thus, a slower metabolism in the rodents. Plasma samples from rats 4 h after [18F]altanserin administration consisted of approximately 80% intact radioligand (Tan et al., 1999). In [18F]altanserin-administered minipigs and humans, half of the radioactivity in plasma originated from metabolites at approximately 80 and 90 min, respectively.
Plasma Protein Unbound Fraction. The fp values (eq. 1) for radioligands in the different species are shown in Table 5. The fp for [11C]verapamil was highest in monkeys, followed by rats and guinea pigs. No species difference in protein binding was observed for [11C]GR205171 and [18F]altanserin, although the fp value was very low for [18F]altanserin, making a reliable measurement difficult. A general trend was that the fp was somewhat increased after CsA administration, but the increase was only significant in [11C]GR205171-administered rats.
Discussion
This article reports pronounced species differences in the brain pharmacokinetics of three PET radioligands that are P-gp substrates, both with and without CsA-induced P-gp inhibition. Rats and guinea pigs showed lower Kp ratios for the radioligands studied than did minipigs, monkeys, and humans. This could be because of more P-gp present in the blood-brain barrier of rodents or because of more efficient P-gp in these species, leading to a higher P-gp-mediated transport capacity. Alternatively, there could be species-related differences in plasma protein binding, leading to different concentrations of molecules available for transport across the blood-brain barrier. Finally, differences in the intrabrain distribution because of differences in brain tissue properties also could lead to the observed species differences.
If transport capacity would be the only explanation to the differences observed when P-gp is functioning normally, complete P-gp inhibition should result in similar brain-to-blood concentration ratios (Kp) across all species. However, the large species differences in brain concentrations were not abolished after P-gp inhibition in the present study. Hence, our conclusion is that the differences in P-gp transport capacity alone do not explain the observed species differences.
In vitro studies have demonstrated differences in the efficacy of P-gp-mediated transport in cell lines transfected with the MDR1 gene (gene coding for P-gp) from different species. In general, most of the evaluated substrates are transported to a higher extent by mouse or rat P-gp than by human P-gp (Yamazaki et al., 2001; Ohe et al., 2003; Katoh et al., 2006; Baltes et al., 2007). However, there are exceptions. CsA, for example, was more effectively transported in cells transfected with human P-gp than in those transfected with rat or mouse P-gp (Katoh et al., 2006; Baltes et al., 2007). Another in vitro study showed that the IC50 value of typical P-gp inhibitors differed with species and substrate (Suzuyama et al., 2007). Cutler et al. (2006) demonstrated in vivo differences between rats and guinea pigs in the IC90 value of the P-gp inhibitor GF-120918. They showed that a higher plasma concentration of GF-120918 was needed in guinea pigs than in rats to achieve the same level of P-gp inhibition. Taken together, the in vitro and in vivo studies indicate that both transport capacity of P-gp and the IC50 values of P-gp inhibitors may differ among species. This would support the hypothesis that there are species differences in both P-gp transport and the IC50 values of CsA, and it may explain the observed in vivo species differences at both baseline and during P-gp inhibition in the present study.
Species differences in intrabrain distribution could also be a relevant explanation for our results. The total substrate brain concentrations observed with PET will depend on the binding potential of the substrate to brain tissue. In this instance, the term binding is taken to include both specific binding to receptors and nonspecific binding to brain tissue components. Brain concentrations of the radioligand will be higher in species with a higher binding potential to brain tissue than in those with a lower binding potential. Species differences in passive transport over the blood-brain barrier would lead to the same outcome as differences in the intrabrain distribution because both active and passive transport contribute to the Kp. However, it is unlikely that the physicochemical properties of, and tightness of tight junctions in, the blood-brain barrier itself would differ to a large extent among species (Cserr and Bundgaard, 1984). Furthermore, the expression of other active transporters such as multidrug resistance protein, organic anion transporter, and organic anion-transporting polypeptide could also contribute to the species differences noted.
Other possible explanations could be differences in the plasma protein binding of the substrate. A lower level of plasma protein binding in one species would, given the same total plasma concentration, contribute to higher brain concentrations, because more substrate molecules would be available for transport across the blood-brain barrier. Indeed, in our study, plasma protein binding was somewhat lower in the higher species than in rats and guinea pigs, but the differences were not large enough to explain the species differences in brain concentrations.
To account for species-related differences in the metabolism of the radioligands, which could have contributed to differences in brain concentrations, the plasma concentrations were corrected for metabolites carrying the radioactive label in the calculation of Kp in our study. Nonetheless, the entry of radioactive metabolites into the brain could also conceivably have affected our results, because the brain kinetics could vary among species depending on the relative fractions of intact radioligand and radiolabeled metabolites. However, this is not a very likely explanation for the species differences seen in this study because the differences were already observed at the beginning of the PET scans, when most of the radioactivity originated from the intact radioligand. Previous studies focusing in more detail on the metabolism of [11C]verapamil and [18F]altanserin in the rat have showed that after [11C]verapamil intravenous administration, most activity in the brain originates from intact [11C]verapamil and that no lipophilic metabolites are formed during the metabolism of [18F]altanserin (Tan et al., 1999; Luurtsema et al., 2005; Syvänen et al., 2008). Thus, metabolism and plasma protein binding were ruled out as explanations for the species differences.
There were some differences in the anesthesia protocols of the different experiments. Hypnorm, which was used in the [18F]altanserin study in rat, contains fentanyl. Fentanyl has been shown to be a P-gp substrate (Henthorn et al., 1999) and could therefore, in theory, increase the uptake of [18F]altanserin by competing for P-gp substrate binding sites. Isoflurane, which was used in the rodent [11C]verapamil and [11C]GR205171 studies, has been shown to moderately increase radioligand binding (Tsukada et al., 1999; Elfving et al., 2003; Harada et al., 2004). Both Hypnorm (VetaPharma Ltd.) and isoflurane anesthesia may therefore reduce the measured difference in uptake between rodents and humans rather than being the cause of the difference. Isoflurane has been shown to decrease the cerebral blood flow (Tsukada et al., 1999). This could indirectly change the radioligand uptake in brain and may be a confounding factor for the results. However, based on the results obtain by Tsukada et al. (1999), the change in cerebral blood flow was too small to explain the differences between the rats and the humans.
Although our study of [11C]verapamil included only two PET scans in monkeys, one untreated monkey and one monkey receiving CsA, our results are in line with those in other publications. For example, Lee et al. (2006) studied uptake of [11C]verapamil after P-gp inhibition and reported that a 2.3-fold increase in Kp values after a 20-mg/kg infusion over 2 h of the P-gp inhibitor PSC833 (P-gp inhibition potency is similar to that of CsA), starting 1 h before the PET scan. This is to be compared with our 3.6-fold increase.
The CsA concentrations in our study were rather similar between species, apart from in the minipig. The guinea pigs had higher CsA concentrations than the rats. Still, the increase in Kp was smaller in the guinea pigs than in the rats. This may underestimate the Kp difference between guinea pigs and rats, rather than being an explanation for the difference. The CsA-treated [11C]verapamil monkey had lower CsA plasma concentrations than the corresponding rodents, which may have resulted in relative underestimation of Kp,increase in the monkey.
Liow et al. (2007) speculated that P-gp inhibition caused a smaller increase in [11C]RWAY concentrations in the cerebellum than in 5HT1A receptor-rich regions. Our results showed a similar trend; SUVincrease and Kp,increase values were somewhat smaller in the cerebellum than in the high binding regions of striatum and frontal cortex for [11C]GR205171 and [18F]altanserin, respectively. In line with these results, Lacán et al. (2008) showed in a recent article a somewhat larger increase in brain uptake of [18F]MPPF in the 5HT1A receptor-rich area of hippocampus compared with cerebellum after P-gp inhibition (Lacán et al., 2008). Regional differences could be attributed to heterogeneous distribution of P-gp within the brain. In the study by Lacán et al. (2008), P-gp protein was expressed more in cerebellum than in other brain regions, a somewhat contradictory finding to the results reported above.
In the development of new drugs, in vitro methods of screening for P-gp-mediated transport and in vivo animal models are usually used before clinical experiments in humans. If a drug that is intended for a target inside the brain is found to be a P-gp substrate or is found in a rat model to have low brain concentrations because of it being a P-gp substrate, it might simply be discarded from further development. However, this study showed that such compounds, even though they are likely to be P-gp substrates in humans as well, could reach relatively high total brain concentrations in humans. In fact, [18F]altanserin and [11C]GR205171 were successfully used in humans before they were found to be P-gp substrates. The findings are important for the interpretation of brain drug delivery when extrapolating preclinical data to humans.
In conclusion, pronounced species differences were found in the brain-to-plasma concentrations of three radiolabeled molecules used in PET. It is suggested that the differences observed are a combination of species-related differences in P-gp transport, potency of CsA to inhibit P-gp, and possibly differences in the intrabrain distribution of the molecules. Species differences should be considered when extrapolating data obtained in animals to humans; compounds found to be P-gp substrates in rodents are likely to also be substrates in higher species, but sufficient blood-brain barrier permeability may be retained in humans to allow the compound to act at intracerebral targets.
Acknowledgments
This work was a collaboration between Uppsala University, Uppsala Imanet (GE Healthcare) and Neurobiology Research Unit Copenhagen. We thank the staff at Uppsala Imanet, particularly Helena Wilking and Margareta Sprycha for metabolite analysis; Lieuwe Appel for the [11C]GR205171 human data; My Quach for assistance with rodent microPET studies; and the staff at the Department of Clinical Chemistry and Pharmacology (Uppsala University Hospital) and the Department of Clinical Biochemistry (Copenhagen University Hospital), especially Søren Senniksen, for the CsA analyses.
Footnotes
-
This work was supported by GE Healthcare and EC-FP6-project DiMI [Grant LSHB-CT-2005-512146].
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.108.024745.
-
ABBREVIATIONS: P-gp, P-glycoprotein; MDR, multidrug resistance; PET, positron emission tomography; 5HT, 5-hydroxytryptamine (serotonin); RWAY, 2,3,4,5, 6,7-hexahydro-1{4-[1[4-(2-methoxyphenyl)-piperazinyl]]-2-phenylbutyry}-1H-azepine; [18F]MPPF, 4-(2′-methoxyphenyl)-1-[2′-(N-2″-pyridinyl)-p-[18F]fluorobenzamido]ethylpiperazine; GR205171, 2 (S)-(2-methoxy-5-(5-trifluoromethyltetrazol-1-yl)-phenylmethylamino)-2(S)-phenylpiperidine; NK, neurokinin; GF-120918, N-(4-[2-(1,2,3,4-tetrahydro-6,7-dimethoxy-2-isoquinolinyl)ethyl]-phenyl)-9,10-dihydro-5-methoxy-9-oxo-4-acridine carboxamide; CsA, cyclosporin A; ROI, region of interest; fp, plasma protein unbound fraction; Kp, brain-to-plasma concentration ratio; Kp,increase, increase in Kp after cyclosporin A administration; Rbuffer, radioactivity concentration in buffer; Rplasma, radioactivity concentration in plasma; SUV, standardized uptake value; SUVincrease, increase in standardized uptake value after cyclosporin A administration; PSC833, valspodar.
- Received October 7, 2008.
- Accepted November 24, 2008.
- The American Society for Pharmacology and Experimental Therapeutics