PET studies of 18F-memantine in healthy volunteers

Nuclear Medicine and Biology 29 (2002) 227–231

PET studies of

18

F-memantine in healthy volunteers

S.M. Ametameya,*, M. Bruehlmeiera, S. Kneifelb, M. Kokica, M. Honera, M. Arigonib, A. Buckb, C. Burgerb, S. Samnickd, G. Quackc, P.A. Schubigera a

Center for Radiopharmaceutical Science of ETH, PSI and USZ, CH-5232 Villigen-PSI, Switzerland b Division of Nuclear Medicine, University Hospital Zurich, Switzerland c Merz⫹ Co. GmbH Co., D-60318 Frankfurt/Main, Germany d Division of Nuclear Medicine, University of Saarland, D-66421 Homburg/Saar, Germany Received 4 March 2001; received in revised form 11 October 2001; accepted 15 October 2001

Abstract Previous studies in mice and PET investigations in a Rhesus monkey showed that the regional uptake of 18F–memantine could be blocked by pharmacological doses of memantine and (⫹)-MK-801. In the present study, the binding characteristics of 18F–memantine was examined in five healthy volunteers. In humans, 18F-memantine was homogeneously distributed in gray matter i.e. cortex and basal ganglia regions, as well as the cerebellum. No radioactive metabolites were detected in plasma during the time-frame of the PET studies. The uptake of 18F-memantine in receptor-rich regions such as striatum and frontal cortex could be well described by a 1-tissue compartment model. The DV⬙ values of all gray matter regions were similar and ranged from 15 to 20 ml/ml. The white matter showed lower DV⬙ values of 15 ⫾ 1.4 ml/ml. These results suggest that 18F-memantine distribution in human brain does not reflect the regional NMDA receptor concentration, and therefore, this radioligand is not suitable for the PET imaging of the NMDA receptors. © 2002 Elsevier Science Inc. All rights reserved. Keywords:

18

F-memantine; NMDA receptor; PET

1. Introduction The N-methyl-D-aspartate (NMDA) receptor belongs to the family of ionotropic glutamate receptors that mediates fast excitatory synaptic transmission in the central nervous system. It is a ligand-gated ion channel that is permeable to monovalent and divalent ions. At resting membrane potential, NMDA receptors are subject to extracellular magnesium blockade which is relieved only after neuron depolarisation. NMDA receptors unlike other ionotropic glutamate receptors require for their activation the simultaneous binding of glutamate and the co-agonist glycine [8]. Overactivation of the NMDA receptors leads to a large intracellular accumulation of calcium ions, which triggers a cascade of events leading to cell death [5]. The receptor has been implicated in the etiology and pathophysiology of many neurological diseases such as ischemia [17], Alzheimer’s disease [6], Huntington’s disease [22] and epilepsy [16]. * Corresponding author. E-mail address: [email protected] (S. Ametamey). Tel.: ⫹41563104260

Physiologically, the NMDA receptor has been shown to be involved in learning and memory [13]. A number of derivatives of uncompetitive NMDA receptor channel blockers such as (⫹)-MK-801, and TCP (thienylcyclohexyl piperidine) have been radiolabelled with either carbon-11, fluorine-18 or iodine-123 and evaluated in vivo as potential agents for imaging the NMDA receptor [3,14,20]. The results of these studies have so far not been successful because the radioligands have been plagued by a lack of specificity. The development of PET or SPECT radioligands for the phencyclidine (PCP)-binding site of the NMDA receptor remains to be a challenge to scientists working in this field of research. The reasons for this are two-fold: a) under physiological conditions it is unknown how many of the NMDA receptor ion channels are activated or are in the open state and b) in contrast to e.g. dopamine receptors and transporter proteins, the PCP-binding sites are located deep within the NMDA receptor ion channel so that access to the binding sites is severely restricted. We previously reported the synthesis and radiolabelling of 1-amino-3-[18F]fluoro-5-methyladamatane (18F-meman-

0969-8051/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved. PII: S 0 9 6 9 - 8 0 5 1 ( 0 1 ) 0 0 2 9 3 - 1

228

S.M. Ametamey et al. / Nuclear Medicine and Biology 29 (2002) 227–231

Fig. 1. Structures of memantine and fluorine-18 labeled derivative.

tine) (Figure 1) [22]. In vitro receptor binding assays indicated that this new derivative binds with similar affinity to the PCP-binding site of the NMDA receptor complex when compared to the parent compound, memantine [18]. Biodistribution studies in mice and PET investigations in a Rhesus monkey showed that the regional brain uptake of 18F-memantine could be blocked by memantine and (⫹)-MK-801 suggesting competition for the same binding sites [2]. To further examine the binding characteristics of 18F-memantine, PET investigations in humans were undertaken. We report here on the results of 18F-memantine binding in five healthy volunteers using PET.

2. Materials and methods RADIOCHEMISTRY: 18F-Memantine was prepared by nucleophilic substitution reaction using a two-step reaction sequence as reported previously [18]. STUDY SUBJECTS: Five healthy male volunteers with a mean age of 23.5 years (range: 23–24) were recruited. The volunteers were screened by psychiatric interview to ensure that they had neither personal nor family histories of major psychiatric disorders. Subjects using drugs that are known to interfere with NMDA receptors were excluded from the study. All subjects gave written informed consent. The study was approved by the local and the National Ethics Committees and by the Swiss Federal Health Office (BAG). PET IMAGING: PET imaging was performed on a GE Advance PET scanner (General Electrics, Waukesha, Wisconsin, U.S.A.) with an axial field of 14.45 cm, divided into 35 slices each with a slice thickness of 4.25 mm. After the placement of a radial artery and a cubital vein catheter, the subjects were positioned supine in the scanner. Prior to the injection of the radioligand a 10 min transmission scan with a 68Ge pin source of 400 MBq activity was performed to correct for attenuation. After i.v. injection (slow bolus over

5 min with an infusion pump) of 74 to 265 MBq 18Fmemantine, a dynamic PET scan was initiated according to the following protocol: 33 frames in 120 min (9x20 sec, 4x30 sec, 2x60 sec, 4x120 sec, 9x300 sec, 3x600 sec, 2x900 sec ). BLOOD SAMPLING AND DETERMINATION OF LABELLED METABOLITES IN PLASMA: Arterial blood samples were collected every 30 sec for the first 6 min, and then at the following time points: 6, 7, 8, 9, 10, 12.5, 15, 17.5, 20, 25, 30, 45, 60, 75, 90 and 120 min p.i.. An aliquot of each sample was measured in a ␥-counter and the plasma was analyzed to correct the input function for metabolized radioligand activity. Arterial blood samples (500 ␮l) were deproteinised with acetonitrile (600 ␮l). The fractions were counted in a ␥-counter and the samples obtained at 15, 30, 45, 60 and 90 min were analyzed with radio-TLC (silicagel; dichloromethane : methanol ⫽ 7:3) for radioactive metabolites and parent radioligand. DATA ANALYSIS: Reconstruction was performed using filtered backprojection (Hanning filter, cut-off: 4 mm transaxial; Ramp filter, cut-off: 8.5 mm axial) and a 128x128 pixel output matrix. The plasma data (well counter units, cpm) were expressed in kBq/ml, by performing calibration measurements of both the well counter and the PET scanner using a 20 cm cylindrical phantom. The PET images were co-registered with T1-weighted magnetic resonance images (MRI) in 3 subjects. Regions of interest (ROIs) were defined over the occipital cortex, frontal cortex, striatum (caudate and putamen), thalamus, parahippocampus, plexus choroideus, white matter and cerebellum. To obtain the regional time-activity curves, regional radioactivity was calculated for each time frame, corrected for decay and plotted against time. PARAMETRIC IMAGES: A Logan plot [10] was applied to the PET images pixelwise, resulting in images of parametric distribution volumes (DV) of 18F-memantine in the brain.

S.M. Ametamey et al. / Nuclear Medicine and Biology 29 (2002) 227–231

Fig. 2. Time activity curves from different brain regions of subject 4 (for brain region abbreviations see Table 1). In addition, plasma radioactivity is plotted (filled diamonds).

KINETIC ANALYSIS: The data were analyzed using both one- and two-tissue compartment models. The parameters of radioligand transport K1, k2⬙ and DV⬙s in the 1-tissue compartment model and K1, k2⬘, k3⬘ k4 and DV⬘ in the 2-tissue compartment model according to the terminology of Koeppe et al. [9] were determined using least squares fitting method based on an unweighted Marquard-Levenberg algorithm. In the one-tissue model, K1, k2⬙ are measured and reflect tracer uptake from plasma into brain tissue (K1) and respective washout (k2⬙) to plasma, assuming that non-specific and specific binding exchange fast and can be lumped together into a one kinetic compartment. In the two-tissue-model, the rate constants K1 and k2⬘ represent influx and efflux rates of radioligand transport across the blood brain barrier, respectively, and k3⬘ and k4 correspond to the rates of radioligand transfer between non-displaceable and specific radioligand binding to the receptor. Two statistical methods were used to compare the two models: Akaike Information Criterion (AIC) [1] and F statistics [7]. To account for the spillover of blood radioactivity, a cerebral blood volume of 5% was assumed and ROI data were accordingly corrected. All required data processing steps such as ROI delineation, time-activity curve generation and kinetic model fitting were performed using a software package dedicated for PET data quantification (PMOD, 11).

3. Results After intravenous injection of 18F-memantine, an initial rapid uptake of radioactivity was observed. Time-activity curves of different brain regions and of plasma are shown in Figure 2. The clearance of 18F-memantine from plasma was fast, while radioactivity peaked at the end of tracer administration at 5 min after the start of the PET scan. From 15

229

Fig. 3. Simple, time dependent ratios between tissue radioactivity and plasma radioactivity for different brain regions in subject 4. For brain region abbreviations see Table 1.

min onwards, the level of radioactivity in plasma remained almost constant, whereas in brain tissues a continuous increase of radioactivity was observed. Tissue and plasma radioactivity ratios kept increasing during the 120 min scanning period (Figure 3). Neither a peak nor a plateau was attained during the 2 hour PET measurements. Metabolites of 18F-memantine were determined in plasma by radio-thin layer chromatography (RTLC). The RTLC revealed only a single radioactive peak that corresponded to parent radioligand by co-migration with reference compound. The results obtained from the kinetic analysis were not consistent in all the subjects. In some subjects the white matter, which is actually devoid of NMDA receptors, showed a kinetic behavior typical of a 2-tissue compartment model. In contrast, the uptake curves of receptor-rich regions such as striatum and frontal cortex were better described by a 1-tissue compartment model, that is, the Akaike information criterion scores were lower for the 2 compartment model than for the 3-compartment model (data not shown). The regional K1, k2⬙ and DV⬙s values determined by iterative curve fitting are shown in Table 1. The DV⬙ values of all gray matter obtained from the 1-tissue compartment model ranged from 15 to 20 ml/ml. K1, k2⬙ and DV⬙ values of all gray matter i.e. cortex and basal ganglia regions, as well as the cerebellum were similar. Co-registration of MR and PET images in 3 subjects confirmed that the ROIs were correctly placed and partial volume effects by neighboring brain tissues were minimal. Figure 4 shows a parametric image of subject 4. The image giving parameter is the distribution volume calculated according to Logan et al. [10] which refers to DV⬙ in the 1-tissue compartment model. The DV⬙ values were homogenous within all gray matter regions and were consistent with the values calculated by kinetic data analysis.

230

S.M. Ametamey et al. / Nuclear Medicine and Biology 29 (2002) 227–231

Table 1 subject cereb_ctx parahip_l parahip_r choroid_l choroid_r front_ctx white_l white_r thal K1

k2⬙

caud_l

caud_r put_l

put_r

occ_ctx

#1 #2 #3 #4 #5

0.177 0.198 0.129 0.225 0.227

0.135 0.136 0.100 0.140 0.181

0.128 0.131 0.102 0.145 0.172

0.297 0.247 0.212 0.444 0.371

0.252 0.261 0.200 0.424 0.335

0.151 0.172 0.122 0.190 0.207

0.092 0.103 0.068 0.121 0.100

0.099 0.109 0.081 0.122 0.096

0.189 0.236 0.133 0.238 0.249

0.158 0.180 0.115 0.222 0.219

0.168 0.187 0.120 0.177 0.207

0.178 0.182 0.124 0.213 0.206

0.190 0.183 0.130 0.226 0.211

0.176 0.196 0.142 0.230 0.263

mean SEM

0.191 0.020

0.138 0.014

0.135 0.013

0.314 0.047

0.294 0.043

0.168 0.017

0.097 0.010

0.101 0.008

0.209 0.024

0.179 0.022

0.172 0.016

0.181 0.017

0.188 0.018

0.201 0.023

#1 #2 #3 #4 #5

0.0111 0.0118 0.0104 0.0136 0.0139

0.0104 0.0092 0.0084 0.0085 0.0138

0.0089 0.0076 0.0076 0.0083 0.0111

0.0255 0.0219 0.0250 0.0440 0.0314

0.0201 0.0249 0.0221 0.0415 0.0292

0.0085 0.0112 0.0093 0.0139 0.0120

0.0067 0.0057 0.0049 0.0086 0.0074

0.0068 0.0049 0.0058 0.0086 0.0058

0.0093 0.0114 0.0081 0.0116 0.0112

0.0073 0.0116 0.0079 0.0146 0.0120

0.0084 0.0105 0.0085 0.0085 0.0110

0.0084 0.0089 0.0073 0.0116 0.0104

0.0096 0.0087 0.0085 0.0117 0.0103

0.0094 0.0110 0.0100 0.0145 0.0156

mean SEM

0.0122 0.0008

0.0101 0.0011

0.0087 0.0007

0.0296 0.0044

0.0275 0.0042

0.0110 0.0011

0.0067 0.0064 0.0103 0.0107 0.0094 0.0093 0.0098 0.0121 0.0007 0.0007 0.0008 0.0015 0.0006 0.0008 0.0007 0.0014

K1/k2⬙(DV⬙) #1 #2 #3 #4 #5 mean SEM

15.9 16.7 12.4 16.6 16.4

13.0 14.8 11.8 16.4 13.1

14.4 17.2 13.4 17.5 15.4

11.7 11.3 8.5 10.1 11.8

12.5 10.5 9.1 10.2 11.5

17.8 15.4 13.1 13.6 17.3

13.7 18.0 14.1 14.0 13.5

14.6 22.4 13.9 14.1 16.5

20.4 20.6 16.6 20.4 22.1

21.5 15.5 14.4 15.2 18.3

19.9 17.9 14.2 20.8 18.7

21.3 20.4 17.0 18.4 19.8

19.7 21.0 15.3 19.2 20.6

18.6 17.8 14.2 15.9 16.9

15.6 0.9

13.8 0.9

15.6 0.9

10.7 0.7

10.8 0.7

15.5 1.1

14.7 0.9

16.3 1.8

20.0 1.0

17.0 1.5

18.3 1.3

19.4 0.8

19.2 1.1

16.7 0.9

4. Discussion Previous studies in mice and PET investigations in a Rhesus monkey showed that the regional uptake of 18F– memantine could be blocked by pharmacological doses of memantine and (⫹)-MK-801. In the present study, the binding characteristics of 18F–memantine was examined in the brain of five healthy volunteers. After intravenous administration, 18F radioactivity was homogeneously distributed

Fig. 4. Transversal, parametric PET image through the basal ganglia in subject 4. A pixelwise analysis, applying a so-called Logan plot to PET and blood data (see text), was used to calculate the tracer distrbution volume (DV) in brain tissue.

in gray matter areas (cortex, basal ganglia) with lower uptake in the white matter. The observed pattern of 18F-memantine uptake was not consistent with autoradiographic studies performed on postmortem human brain using tritiated TCP [21]. These results are surprising, considering the regional differences observed in mice and one Rhesus monkey [2]. To our knowledge no species differences in the distribution of NMDA receptors have been reported. However, regional variations in the binding characteristics of uncompetitive NMDA channel blockers have been published [4]. For example, whereas (⫹)-MK-801, TCP or PCP preferentially bind to distinct subunit compositions of the NMDA receptor complex in forebrain regions, memantine and its structural analogue amantadine display similar affinities to cortical and cerebellar binding sites and therefore 18F–memantine can be expected to show similar radioactivity levels in the cortical and cerebellar regions. In mice, the uptake of radioligand was consistent with the known distribution of NMDA receptors, but only a few mouse brain regions could be included in such a correlation. In the monkey brain, however, the uptake curves obtained with 18F–memantine were similar in all the gray matter regions examined [19]. Considering the observed distribution pattern of 18F–memantine in humans, the regional differences observed in mice have to be questioned, as these in part may have been due to unspecific effects. The meaning of the blockade effect observed in both mice and monkey with 18F–memantine is thus unclear. There exists the possibility that the reduced brain uptake observed in the previous studies with blocking

S.M. Ametamey et al. / Nuclear Medicine and Biology 29 (2002) 227–231

agents such as memantine and (⫹)-MK-801 could be due to a global blood flow effect. In the present study, no blocking studies were undertaken since this would require the administration of unlabelled NMDA antagonists, and blockade of the NMDA receptors is expected to produce alterations in normal brain physiology and illicit severe side-effects. The unspecific pattern of radioligand binding of 18Fmemantine in humans cannot be attributed to radiolabelled metabolites that might have been formed in the human brain or penetrate the blood-brain barrier. Memantine has been shown to be poorly metabolized in man [15]. In the present work, 18F-memantine was found to be metabolically stable within the time-frame of the PET studies. No radiolabelled compound other than the parent radioligand could be detected. Similarly, more than 97% of radioligand was found to be intact in monkey plasma 60 min after i.v. administration of 18F-memantine [19]. Mice brain extracts also showed that more than 93% of extracted radioactivity was parent compound. The effect of protein binding would be negligible since drug treatment did not significantly change plasma radioactivity level. The kinetic analysis of 18F-memantine uptake was performed using 1- and 2-tissue compartment models to determine the rate constants of radioligand binding. The results obtained from the 2-tissue compartment model were, however, inconsistent with a large inter-individual variation. For example, in some subjects, the white matter, a region that lacks NMDA receptors, showed a kinetic behavior typical of a 2-tissue compartment model. In those brain regions, where the existence of the second tissue compartment could kinetically be verified, the compartment was poorly defined and a considerable inter-subject variability was observed. The physiological meaning of the two kinetic tissue compartments is unclear, however, none of them reflects a specific receptor compartment. We conclude, that 18F-memantine shows a high perfusion dependent uptake in the human brain and is therefore not a suitable radioligand for the PET imaging of the NMDA receptors. References [1] H.A. Akaike, A new look at the statistical model identification IEEE, Trans. Automat. Contr. 19 (1974) 716 –723. [2] S.M. Ametamey, S. Samnick, K.L. Leenders, P. Vontobel, G. Quack, C.G. Parsons, P.A. Schubiger, Fluorine-18 radiolabelling, biodistribution studies and preliminary PET evaluation of a new memantine derivative for imaging the NMDA receptor J, Receptor Signal Transduction Res. 19 (1– 4) (1999) 129 –141. [3] J. Blin, A. Denis, T. Yamaguchi, C. Crouzel, E.T. MacKenzie, J.C. Baron, PET studies of [18F]methyl-MK-801, a potential NMDA receptor complex radioligand, Neurosci Lett. 121 (1991) 183–186. [4] I. Brensink, W. Danysz, C.G. Parsons, E. Mutschler, Different binding affinities of NMDA receptor channel blockers in various brain regions-Indication of NMDA receptor heterogeneity Neuropharmacology, 34 (5) (1995) 533–540. [5] D.W. Choi, Glutamate neurotoxicity and diseases of the nervous system, Neuron 1 (1988) 623– 634.

231

[6] D. Dewart, D.T. Chalmers, A. Shand, D.I. Graham, J. McCulloch, Selective reduction of quisqualate (AMPA) receptors in Alzheimer cerebellum, Ann Neurol. 28 (1990) 805– 810. [7] R. Hawkins, M. Phelps, S.-C. Huang, Effects of temporal sampling, glucose metabolic rates, and disruption of the blood brain barrier on the FDG model with and without a vascular compartment, Studies in human brain tumours with PET, J. Cer. Blood Flow Metab. 6 (1986) 170 –183. [8] J.W. Johnson, P. Ascher, Glycine potentiates NMDA response in cultured mouse brain neurones, Nature 325 (1987) 529 –531 (1987). [9] R.A. Koeppe, V.A. Holthoff, K.A. Frey, M.R. Kilbourn, D.E. Kuhl, Compartmental analysis of [11C]flumazenil kinetics for the estimation of ligand transport rate and receptor distribution using positron emission tomography, J. Cereb. Blood Flow Metab. 11 (1991) 735–744. [10] J. Logan, J.S. Fowler, N.D. Volkow, A.P. Wolf, S.L. Dewey, D.J. Schlyer, R.R. MacGregor, R. Hitzemann, B. Bendriem, S.J. Gatley, D.R. Christman, Graphical analysis of reversible radioligand binding from time-activity measurements applied to N-11C-methyl-(-)cocaine: PET studies in human subjects, J. Cer. Blood Flow Metab. 10 (1990) 740 –747. [11] K. Mikolajczyk, M. Szabatin, P. Rudnicki, M. Grodyki, C. Burger, A JAVA environment for medical image data analysis, Initial application for brain PET quantification, Med Inf. 23 (1998) 207–214. [12] M.A. Mintum, M.E. Raichle, M.R. Kilbourn, G.F. Wooten, M.J. Welch, A quantitative model for the in vivo assessment of drug binding sites with positron emission tomography, Ann Neurol. 15 (1984) 217–227. [13] R.G.M. Morris, A. Anderson, G.S. Lynch, M. Baudry, Selective impairment of learning and blockade of long-term potentiation by an N-methyl-D-aspartate receptor antagonist, Nature, 319 (1986) 774–776. [14] J. Owens, D.J. Wyper, J. Patterson, D.R.P. Brown, A.T. Elliott, G.M. Teasdale, J. McCoulloch, First SPET images of glutamate (NMDA) receptor activation in vivo in cerebral ischaemia, Nucl. Med. Comm. 18 (1997) 149 –158. [15] C.G. Parsons, R. Gruner, J. Rozental, J. Millar, D. Lodge, Patch clamp studies on the kinetics and selectivity of NMDA receptor antagonism by memantine, Neuropharmacology, 32 (12) (1993) 1337–1350. [16] S. Patel, A.G. Chapman, J.L. Graham, B.S. Meldrum, P. Frey, Anticonvulsant activity of the NMDA antagonists, D(-)4-(3-phosphonopropyl)piperazine-2-carboxylic acid (D-CPP) and D(-)(E)-4-(3-phosphonoprop-2-enyl)piperazine-2-carboxylic acid (D-CPP-ene) in a rodent and primate model of reflex epilepsy, Epilepsy Res. 7 (1990) 3–10. [17] S.M. Rothman, J.W. Olney, Excitotoxicity and the NMDA receptor, Trends in Neurosci. 10 (1987) 299 –302. [18] S. Samnick, S. Ametamey, M.R. Gold, P.A. Schubiger, Synthesis and preliminary in vitro evaluation of a new memantine derivative 1-amino-3-[18F]-fluoromethyl-5-methyl adamantane: A potential ligand for mapping the NMDA receptor-complex, J. Lab. Compd. Radiopharm, 39 (1997) 241–250. [19] S. Samnick, S. Ametamey, K.L. Leenders, P. Vontobel, G. Quack, C.G. Parsons, H. Neu, P.A. Schubiger, Electrophysiological study biodistribution in mice and preliminary PET evaluation in a Rhesus monkey of 1-amino-3-[18F]-fluoromethyl-5-methyl adamantane (18FMEM): A potential radioligand for mapping the NMDA-receptor complex, Nucl. Med. Biol. 25 (1998) 323–330. [20] Y. Shibayama, S. Sasaki, U. Tomita, T. Nishikawa, Synthesis and evaluation of new 18F-labelled thienylcyclohexylpiperidine (TCP) analogues as radioligands for the NMDA receptor-channel complex, J. Label Compd. Radiopharm. 38 (1995) 77– 86. [21] A.D. Weissmann, M.F. Casanova, J.E. Kleinmann, E.B. De Souza, PCP and sigma receptors in brain are not altered after repeated exposure to PCP in humans, Neuropsychopharm, 4 (2) (1991) 95– 102. [22] A.B. Young, J.T. Greenamyre, Z. Hollingsworth, R. Albin, C. D’Amato, I. Shoulson, J.B. Penny, NMDA receptor losses in putamen from patients with Huntington’s disease, Science (1988) 981–983.