204 S
11
Table 1
subjects. [ C]FMZ is a BZ antagonist widely used in
Characteristics of the early-blind subjects
human brain studies using positron emission tomography
Subjects Age at test
Age of blindness Etiology
PET. BZ receptors BZR are functionally associated to
years onset
GABA receptors which are made up of pentameric
A
assemblies of subunits a, b, g, d, r that form a chloride
EB1 69
Birth Unknown
2
EB2 45
18 months Retinoblastoma enucleated
channel. The Cl conductance increases when GABA
EB3 23
Birth Retrolental fibroplasia
binds to the b-subunit and the effect is facilitated when BZ
a
EB4 54
Birth–10 years Congenital glaucoma
binds to the ag subunit complex in the same channel
EB5 57
,1 year Unknown enucleated
[11,53]. FMZ binds to most of the GABA receptors
A
a
EB4 had very poor vision from birth. Loss of the right eye at the age of
resulting from different subunit combinations in mam-
6 months, and definitive loss of the left eye at 10 years.
malian brain, except to those containing d subunit [53]. The FMZ affinity for receptors containing a
subunit is
6
lower than the affinity for receptors with a subunit
proved by the Medical Ethics Committee of the School of
1
´ [27,53]. GABA receptors containing a subunit are only
Medicine at the Universite Catholique de Louvain.
A 6
expressed in cerebellar granule cells and represent 45 of
11
all GABA receptors in the rat cerebellum [26]. 2.2. Synthesis of [ C]FMZ
A
The kinetics of in vivo ligand binding of FMZ to central
11 11
BZR was described using a nonlinear three compartment [ C]FMZ [Methyl- C] Ro 15-1788 was labelled by
model [15]. The model parameters were accurately esti- N-alkylation of the desmethyl compound Ro 15-5528
11
mated by using a three-injection protocol which guarantees with anhydrous [ C]methyl iodide in acetone by use of
11
a unique solution and small parameter estimation un- sodium hydroxide as base [20]. [ C]FMZ was purified by
certainties [15,16]. The input curve to the compartmental semi-preparative reverse phase using high-performance
model was obtained by applying a metabolite correction liquid chromatography HPLC. The HPLC columns used
11
method which estimates the relative fraction of [ C]FMZ were Alltech Econosil with 0.01M H PO acetonitrile and
3 4
in the total radioactivity in plasma curve from the tissue 70 30 as eluent. The radiopharmaceutical formulation has
kinetic data without the need for actual FMZ metabolite been performed following a method that uses a C
Sep-
18
measurements [52]. Pak Plus cartridge from Waters [7].
11
Labeling and quantification of BZR using [ C]FMZ may provide information on the effects of early blindness
2.3. Experimental protocol on the distribution of GABA receptors in the cerebral and
A
cerebellar cortex of human subjects. The effect of visual The experimental protocol lasted 90 min and consisted
11
deprivation on the distribution of GABA neurons and its of three injections of [ C]FMZ and or unlabeled FMZ
11
receptors in visual brain areas has been mainly studied in [15]. Firstly, an injection of about 370 MBq of [ C]FMZ
cases of monocular deprivation in adult monkeys [8,21– with high specific activity was performed and 30 min later
23]. However, the effect of early binocular deprivation on a displacement injection of 0.7 mg of unlabeled FMZ was
GABA distribution, to our knowledge, has not been performed. At 60 min, a coinjection of approximately 160
11
reported, neither in animals nor in humans. This study may MBq of [ C]FMZ and 1.2 mg of unlabeled FMZ was
provide insight on the contribution of such inhibitory made. The specific activity of the first injection ranged
synapses to the measured abnormal metabolism in the EB from 5300 to 16200 MBq mmol. Each time, the FMZ was
visual cortex and cerebellum [13,58]. injected as a bolus of 30 s through a 22-gauge catheter
Abbocath E in a forearm vein. A 24-gauge catheter
Abbocath E was inserted in the radial artery of the other
arm under local anesthesia with bipuvacaine for blood
2. Materials and methods sampling. After the labeled tracer injections at 0 and 60
min, blood samples |0.2 ml were withdrawn manually 2.1. Subjects
as quickly as possible every 3 or 4 s for the first 2 min. Thereafter, the sampling interval increased progressively to
Five male volunteers with peripheral blindness of early about 10 min. The selected timing allowed a good sam-
onset Table 1 and five sighted control subjects partici- pling of the plasma curve for the 30-s bolus injection. The
11
pated in this study. EB subjects had no residual light plasma curves were corrected for
C decay and expressed
11
perception, but were otherwise neurological normal. EB in pmol ml by using initial specific activity of [ C]FMZ.
and SC subject mean ages were 49617 and 42613 years, respectively P50.46. None of the subjects was under any
2.4. PET imaging medication. All subjects gave their informed consent
before undergoing the PET acquisition. These experiments PET studies were performed using the ECAT EXACT
comply with the Declaration of Helsinki and were ap- HR from CTI Siemens, a whole body high spatial res-
S .M. Sanabria-Bohorquez et al. Brain Research 888 2001 203 –211
205
olution scanner. Transaxial resolution is approximately 4 describe simple diffusive transport of FMZ between the
mm full width at half maximum FWHM at the center of plasma and free tracer compartments. The association and
the field of view FOV, and decreases to 6.75 mm FWHM the dissociation rate constants k
and k , respectively
on off
at 20 cm from the center. The axial resolution is 4 mm describe the FMZ exchange between the free and spe-
FWHM at FOV center [61]. Acquisition was performed in cifically bound tracer compartments. The equilibrium
three-dimensional 3D mode and images were recon- dissociation constant K was obtained from the ratio
d
structed using the 3D reprojection 3DRP algorithm [32] between k
and k . When describing in vivo reversible
off on
including scatter correction [60]. A Hanning filter with a binding, K
corresponds to the ligand concentration at
d
70 relative frequency cutoff was used in both transaxial which half of the receptors are bounded to the ligand. If a
and axial directions to achieve a quasi isotropic spatial ligand act at low concentrations on a receptor, the K value
d
resolution of about 8 mm in the whole brain. For each is low and the ligand is said to have high affinity for those
subject, a 15-min transmission scan was performed prior to receptors. The complete description of the compartmental
tracer administration to estimate attenuation correction. model can be found in the work of Delforge et al. [15].
68
The transmission scan used three rotating Ge rod sources
The vascular compartment in the model only accounts
11
with electronic windowing, so attenuation is scatter free. A for the concentration of [ C]FMZ in plasma because none
11
sequence of 16 frames was obtained after each injection of the [ C]FMZ metabolites crosses the blood–brain
8315, 3360, and 53300 s, and 47 contiguous transaxial barrier BBB [14]. For each subject, the contribution of
slices were reconstructed with a voxel size of 2 mm in radioactive metabolites to the total plasma curve was
tomographic direction and 3.125 mm in axial direction. estimated by applying a mathematical correction method
11
The head of each subject was positioned in the scanner not requiring additional [ C]FMZ metabolite measure-
FOV by aligning two sets of low power laser beams with ment [52]. The model equations were solved numerically
the canthomeatal and the mid-saggital lines. Adhesive by applying the Levenberg–Marquardt method [48] and
bands were used to minimize head movements during the parameter coefficients of variation were obtained from the
study. covariance matrix resulting from the sensitivity function
Three-dimensional magnetic resonance images MRI matrix of the weighted least-squares minimization [9,16].
were obtained on a 0.5 Tesla Philips Gyroscan unit using Programs were implemented in
MATLAB
The MathWorks, the Fast Field Echo technique. T-1 weighted images TR5
Inc., Natick, MA. 30 ms, TE513 ms, flip angle5308, slice thickness52 mm
were obtained in the bicommissural AC–PC orientation. 2.5.2. Regions of interest
Before delineating regions of interest ROI, PET and 2.5. Data analysis
MRI data were processed as follows. First, a summed PET image was obtained from the first 16 frames after the first
11
2.5.1. Kinetic model [ C]FMZ injection. Then, for each subject, the summed
11
The kinetics of in vivo ligand binding of [ C]FMZ to PET image was realigned to the MRI using the AIR
central BZ receptors was based on compartmental analysis package [62,63]. SPM96 Wellcome Department of Cogni-
[15]. The model considered here includes the plasma space tive Neurology, Institute of Neurology, London, UK was
together with two extra-vascular compartments represent- used to normalize the matching MRI and summed PET
ing the ligand in tissue free and non-specifically bound images in the Talairach and Tournoux coordinate system
FMZ and the ligand specifically bound to BZR Fig. 1. [56] with a 2-mm cubic voxel. The accuracy of realign-
This model is non-linear with four rate constants k , k , ment and normalization procedures were assessed with an
1 2
k , k plus the concentration of the receptor sites
interactive home made image display software [44] im-
on off
available for binding B . The rate constants k and k
plemented in IDL language IDL Research System, Inc..
max 1
2
11 11
Fig. 1. Compartmental model for [ C]FMZ–BZR interactions. The intravascular compartment represents the concentration of non-metabolized [ C]FMZ
11
in plasma, C t. Extravascular compartments account for free ligand, C t, and [ C]FMZ bound to BZR, C t. The rate of binding of the free ligand
p f
b
depends on the association rate constant k and on the receptor density available for binding [B
2 C t 2 C t]. B denotes the concentration of
on max
b b
max
BZR. C t is non-zero when unlabeled FMZ is introduced in the system. The rate of dissociation of bound FMZ from BZR is k . The variables with an
b off
asterisk denote labeled concentrations.
206 S
Subsequently, the spatial transformer obtained for the TACs corresponded to the mean radioactivity concen-
11
summed PET image normalization was applied to each tration on the defined VOI corrected for
C decay. For frame of the kinetic PET study. Since MRI was not
each VOI, the BZR density B and the four rate constants
max
available for one of the control subjects, a FMZ template were obtained by fitting the kinetic model to the ex-
was obtained by averaging the normalized PET images of perimental TACs.
the four other control subjects. This template was then used to normalize the summed and kinetic PET images in
the Talairach and Tournoux coordinate system. Using the image analysis program Mediman [10], ROIs
3. Results