0883-2897192 $5.00 + 0.00 Copyright 0 1992 Pergamon Press Ltd
Nucl. Med. Biol. Vol. 19, No. 8, pp. 881-888, 1992 Int. J. Radiat. Appl. Instrum. Part B Printed in Great Britain. All rights reserved
Pharmacokinetics of the SPECT Benzodiazepine Receptor Radioligand [ 123 I]Iomazenil in Human and Non-human Primates SAM1 S. ZOGHBI’*, RONALD M. BALDWIN’, JOHN P. SEIBYL2, MOHAMMED S. AL-TIKRITI’, YOLANDA ZEA-PONCE’, MARC LARUELLE2, ELZBIETA H. SYBIRSKA2, SCOTT W. WOODS2, ANDREW W. GODDARD’, ROBERT T. MALISON’, RALF ZIMMERMAN’, DENNIS S. CHARNEY2, EILEEN 0. SMITH’, PAUL B. HOFFER’ and ROBERT B. INNIS’ Departments of ‘Diagnostic Radiology and ZPsychiatry, Yale University School of Medicine, New Haven, CT 06510 and 3VA Medical Center, West Haven, CT 06516 U.S.A. (Received 19 March 1992)
The pharmacokinetics of [i231]iomazenil (Ro 16-0154) in 5 healthy human volunteers were compared to those in 2 hypothermic and 3 normothermic anesthetized monkeys. Following intravenous injection in humans and monkeys, [‘231]iomazenil rapidly diffused outside the vascular bed and was cleared from the arterial plasma triexponentially. The clearance half-times in hypothermic animals were protracted to values closer to those of the human. [1231]Iomazenilwas metabolized mainly to a polar radiometabolite (not extracted by ethyl acetate) in the human whereas an additional lipophilic radiometabolite was detected in the monkey. In vitro and in viuo studies showed that [iz31]iomazenil established equal concentrations in association with the cellular and plasma component of the blood, indicating that the plasma clearance of [‘231]iomazenil mirrors that of the blood. Analysis of organs from a monkey given [‘231]iomazenil showed that the parent compound was actively taken up by peripheral organs; the polar radiometabolite accumulated mainly in the bile and the kidneys whereas the non-polar radiometabolite accumulated in the urine and kidneys. Greater than 90% of the radioactivity in the different regions of
the brain was unchanged parent [‘231]iomazenil.
Introduction Iomazenil (Ro 16-0154) is an iodine-containing ligand for benzodiazepine (BZ) receptors and is a close analog of the fluorine-containing antagonist flumazenil, Ro 15-1788 (Beer et al., 1990). [‘251]Iomazeni1 binds reversibly and selectively to BZ receptors in brain tissue homogenates with high affinity (Ko = 0.5 nM at 37°C) and high ratio of specific to non-specific binding (40 : 1) (Johnson et al., 1990). In keeping with in vitro studies, SPECT imaging with [‘231]iomazeni1 has shown that the brain uptake of this radiotracer represents reversible and selective binding to the BZ receptor and that it has high potency (ED, = 15 nmol/kg i.v.) and high ratio of specific to non-specific binding (10: 1) (Beer et al., 1990; Innis er al., 1991b). The peak brain uptake of [‘231)iomazeni1 is unusually high (10-12% of injected *Address all correspondence to: Sami S. Zoghbi, Ph.D., Department of Diagnostic Radiology, 326 Brady Memo-
rial Laboratory, Yale University School of Medicine, 333 Cedar Street. New Haven, CT 06510, U.S.A.
dose) and shows a relatively slow washout of 3% per hour (Innis et al., 1991a). The high affinity of iomazenil for the BZ receptor may partially explain its high and slow washout. However, brain uptake its metabolism and clearance from the body also significantly determine its distribution over time in all body organs. The purpose of the present study was to develop methods for measurement of the free parent compound in plasma and to examine the kinetics and clearance of [‘231]iomazenil in human and non-human primates.
Materials and Methods Radiolabeling
The method of McBride et al. (1991) was modified to make [‘231]iomazenil suitable for intravenous injection. Absolute ethanol (77 pL) was added to a vial containing 50 pg (0.087 pmol) ethyl ‘I-(tributyl-stannyl)-5,6-dihydro-5-methyl-6-oxo-4H-imidazo-[l,5a][1,4]benzodiazepine-3-carboxylate and 100 Hg (0.22 pmol) 1,3,4,6- tetrachloro - 3a, 6a -diphenylglycoluril 881
882
S~hil
S. ZCIGHB~
(Iodogen, Pierce Chemical Co.) and the mixture was mixed by sonication for 3 min. No-carrier-added [ 123 Qodium iodide (20 mCi in 60 p L of 0.1 M NaOH; Nordion International Inc., Vancouver, Canada) was mixed with sufficient ethanol:2M NH,C1:0.5 M H,P04 (1: 1: 1 v/v/v) so that the final pH was 2.9-3.0. After 1.5 h at room temperature, the reaction was terminated by adding 0.1 mL (10 mg) aqueous NaHS03 (100 mg/mL). Volatile radioactivity, a byproduct of this reaction, was purged with a stream of argon into a charcoal column for 10-20 min, then 1 mL H,O and 0.2 mL saturated NaHCO, were added (final pH 7). The mixture was extracted with 3 x 1 mL toluene and the combined extracts were dried over anhydrous Na,SO,. The toluene was evaporated to dryness with a rotary evaporator under argon and the residue was dissolved in 110 PL methanol, followed by 90 PL water. Radiolabeled products were separated by high pressure liquid chromatography (HPLC), using a reverse-phase C- 18 column (Novapak 3.9 x 300 mm, Waters Division of Millipore, Milford, Mass.) and a mobile phase of methanol/water (55:45 v/v) at a flow rate of 0.7 mL/min, using an on-line NaI(T1) scintillation detector and a multichannel scaler. Under these conditions, the retention time of iomazenil was 8.4min, and those of possible by-products with chlorine and hydrogen in the 7-position were 6.9 and 5.8 min, respectively. The tributyltin precursor was retained on the column and was subsequently eluted with absolute methanol. The [‘231]iomazenil peak was collected in a flask containing 100~8 ascorbic acid and evaporated to dryness with argon. The residue was dissolved in 4OOpL absolute ethanol, diluted with 6mL normal saline, and sterilized by 0.2 pm membrane filtration (Acrodisc 13, Gelman Sciences, Ann Arbor, Mich.). Radiochemical purity was determined by HPLC in the same system. The preparation was stable for at least 24 h at room temperature. Sterility was confirmed by lack of bacterial growth in two media: trypticase and fluid thioglycollate. Apyrogenicity was confirmed using the limulus amebocyte lysate (LAL) test (Endosafe, Charleston, S.C.). Subjects Metabolic data were obtained in conjunction with an imaging study in normal volunteers (Woods et al., 1992). After giving written informed consent, 5 healthy human volunteers (2 female and 3 male) with ages of 28 f 8 y (with these and subsequent data expressed as mean + SD) and weights of 76 -+ 12 kg were administered 5.0 f 0.1 mCi [‘231]iomazenil intravenously. Five ovariectomized female baboons (9-l 1 kg Papio anubis) received 8.8 -+ 4.2 mCi [‘231Jiomazenil. The animals were divided into two groups: three normothermic baboons with body temperatures maintained at 3.5-37°C with a fluidcirculating blanket, and two hypothermic baboons with body temperatures maintained at 32-34°C. Monkeys were initially immobilized with ketamine
ef
al.
(15 mg/kg); anesthesia was induced with intravenous (i.v.) sodium pentobarbital (9.8 mg/kg) and maintained by sodium pentobarbital at approximately one third of the initial dose every 30min. The doses of radiopharmaceutical, measured in a gas inonization chamber dose calibrator (Capintec CRC-7, Montvale, N.J.), were injected over 20-30 s through an indwelling venous catheter. To prevent contamination from the site of radioligand injection, venous blood samples were collected from a line in the contralateral limb. Arterial blood samples were collected from an indwelling catheter in either the radial artery (human) or femoral artery (monkey). Two 10 mL blood samples were drawn in heparintreated syringes just prior to the adminis’tration of radioactivity. As controls, 10 ptci [‘231]iomazenil was added to the 10 mL blood samples at the time of injection (“standard A”) and at the time of metabolite analysis (“standard B”), approx. 15 h after injection. Beginning immediately after administration of radioactivity, 1.5 mL arterial blood samples were drawn continuously at IO s intervals for 120 s using a peristaltic pump (Harvard Apparatus Model 2501-001, Southwick, Mass.) and collected in test tubes containing 5OpL sodium heparin (5000 units/ml). Individual blood samples were then drawn in heparin-treated syringes at 2 min intervals for 10 min and subsequently at 15, 20, 30, 60 and 120 min. In the human, blood samples drawn at 1, 4, 30, 60 and 120 min were 10 mL. Samples were placed on ice within 30 min after drawing and were stored at 4°C until analysis. To verify recovery of radioactivity flowing through the peristaltic pump tubing, 20 PCi [‘231]iomazenil was added to 20mL of heparinized whole blood and drawn through the pump at the standard human flow rate of 9 mL/min with 1.5 mL samples collected every 10 s. Immediately thereafter, a 20 mL sample of non-radioactive blood was drawn through the pump and similar fractions were collected. All fractions were counted as below. Metabolite analysis Radioactive blood samples and standards A and B were centrifuged at 1000 g for 10 min. The concentration of radioactivity in plasma and in cellular blood elements was counted in equal volume aliquots (50 ,QL) in an automatic well-type y-counter (Model 8000, Beckman Instruments, Fullerton, Calif.). The counting efficiency of the system was determined using sources of 123Icalibrated in the gas ionization chamber with geometry similar to that of the blood samples. All radioactivity measurements were decaycorrected to the time of radioligand injection. Plasma and cellular fractions were extracted three times with equal volumes of ethyl acetate; extraction was calculated from the activity in the aqueous phase counted before and after extraction at a constant geometry and corrected for decay. For cellular blood elements in samples at 4,60 and 120 min, and standards A and
883
Pharmacokinetics of [I23Iliomazenil B, extraction was done after the addition of distilled water to produce hemolysis. The precision of the method was measured in one experiment by repeatedly extracting aliquots of a single sample of blood obtained at 60min post-injection; the coefficient of variation was 5.5% for 10 mL samples (n = 3), and 15.0% for 2mL samples (n = IO). Beginning with samples of lowest activity, selected organic extracts (usually samples collected at 1,4, 30,60 and 120 min) were evaporated to dryness on a rotary evaporator under argon in a 37°C water bath. The residue was dissolved in 82.5 PL methanol, diluted with 67.5 PL H,O and injected into the HPLC apparatus using the same conditions as above but with a column reserved for low-level radioactivity. Effluent fractions of 0.5 mL were collected in test tubes using a fraction collector (Microfractionator FC-80K, Gilson, Middleton, Wis.) and counted with the automatic y-counter to reconstruct a histogram of the HPLC profile. The parent fraction in the intermediate samples not chromatographed was estimated by interpolation. The composition of the samples was determined as the product of the fraction extracted and the parent composition (HPLC) divided by a recovery coefficient calculated as the product of the extraction and fraction of [123Iliomazenil (measured by HPLC) in standards A and B. The in vitro distribution of [‘231]iomazenil between plasma and cellular blood elements was determined following addition of radioligand to whole blood at approx. the same concentration as in the in uiuo studies. The blood was then centrifuged and plasma was separated from cells. Each fraction was counted for radioactivity and the relative distribution of radioactivity between cells and plasma was calculated. The cells were then washed three more times with equal volumes of non-radioactive plasma. Protein binding
The binding of [1231]iomazenil to plasma proteins by ultrafiltration through Centriconmembrane filters (Amicon Division, W.R. Grace & Co., Danvers, Mass.). In addition to plasma from standards A and B, 10 PL of a 1: 100 dilution of [‘231]iomazenil was added to 1 mL each of plasma and normal saline as control. Duplicate samples of 200 PL were placed in the unit and centrifuged 20 min at 10,OOOg; the separated compartments were counted on the y-counter. The protein bound fraction was calculated as the ratio of the radioactivity on the filter to the sum of the filter and filtrate. The same process was applied in the study of the effect of increasing amounts of added radioligand (0.02-0.8 PCi) to a constant 500 PL plasma. in vitro was determined
Data analysis
The plasma clearance profile of the free parent compound (non-protein bound) was obtained by expressing the amounts of [123IJiomazenil relative to the peak level of radioactivity in the respective
samples. The data were fitted to exponential curves with a graphics program (Kaleidagraph@ for the Macintosh, version 2.1, Synergy Software, Redding, Pa). Using a stripping technique and three component model, the relative clearance, C(t), of the free parent compound was described by the formula: C(t) = fie-'O.WW +f2e-W'3/W +f,e-W93/Td~ (1) where T, , T2 and T, are the half-times for clearance of the three components, and fi , f2and f3represent the fractional involvement of the three components relative to the initial concentration, defined as the ratio of the y-intercept for each component divided by the sum of the three intercepts. The volume of distribution was determined by dividing the injected dose of radioactivity (PCi) by the concentration of radioactivity (pCi/mL) present in the plasma at the peak level. Average values were expressed as mean +_ 1 SD. Statistical significance testing of the means was done by first comparing variances using the F test followed by the appropriate upper t test (Burr, 1972). Biodistribution in non-human primate 2 h after injection of 15.9 mCi [‘231]iomazenil in a separate hypothermic baboon (10.0 kg), the animal was euthanized with an overdose of pentobarbital. Tissue samples (c. 0.5 g) of organs of interest were weighed; brain was further dissected into occipital gray matter, frontal gray matter, white matter and cerebellum. Total radioactivity was counted in the y-counter and the uptake was calculated as percent injected dose per gram (X ID/g). Tissue samples were homogenized in a Polytron tissue disrupter (Brinkmann Instruments, Lucerne, Switzerland) while cooling in ice. Aliquots of the homogenates were extracted three times with an equal volume of ethyl acetate; the extraction fraction (E,) for each tissue was calculated and the parent compound quantified by HPLC as described above. The extraction fraction (E]) was further corrected by dividing it by its respective extraction efficiency (~g) which was determined by adding excess amounts of [123Iliomazenil (internal standard) to identical aliquots of the tissue homogenates and subjecting them to the same extraction procedure. Radioactivity was quantitated by counting each sample before (n,) and after (n2) the addition of internal standard and after extraction with ethyl acetate (n,). The extraction efficiency (so) was then calculated by the formula so = (n2 - n3 - el nl)/(n2 - nl ).
Results Radiolabeling
The final preparation was obtained in 54 + 22% (range 32-82%) overall yield and in greater than 99% radiochemical purity. At the termination of the iodination reaction, an average of 24 & 17% (range O-46%) of the total radioactivity was volatile. The
Sahn
884
S. ZoGHsr et al.
volatile fraction could be trapped in organic solvents such as toluene or absolute ethanol, but not in a basic solution of sodium bisulfite. The specific activity of these preparations was not measured directly; however, the tributyltin precursor and selected prep arations were analyzed by HPLC with U.V. detection and no evidence of iomazenil carrier or its analogs with H- or Cl-substitution were observed. The limit of detection was on the order of 5 x 10e3 pmol, which would imply a specific activity of greater than 4000 mCi/pmol; however, based on the results from McBride et al. (1991) in which the entire batch of a much larger scale reaction was analyzed, the specific activity probably is closer to 180,000 mCi/pmol. Protein binding
By the ultrafiltration method, 81.0 f 6.3% [‘231]iomazenil was protein bound in human plasma compared to 76.6 + 2.1% in the non-human primate. The two were not statistically different (P > 0.05). Relative protein binding was not affected by increasing concentrations of the radioligand in plasma over a 40-fold range encompassing concentrations typically found in the human and the non-human primate plasma samples. Recovery of radioactivity
Radioactivity recovered from the peristaltic pump used in early arterial sampling represented 99.4% of the total added activity in control experiments.
Metabolites
At the time of peak level of radioactivity in arterial plasma (40-100 s in the human and 20-80 s postinjection in monkey) 97.0 + 4.2% of activity in human plasma and 91.5 + 2.3% of activity in monkey plasma represented parent compound. Extraction with ethyl acetate allowed an operational definition of two fractions of plasma radioactivity: (1) a polar fraction which was not extracted and which increased over time; and (2) a lipophilic fraction which W(IS extracted and which decreased over time in both human and monkey. The lipophilic fraction in human
Hypothermic Monkey
Human loo
Only 0.4% of the radioactivity contaminated the first non-radioactive blood sample, and 0.3% was associated with the catheter. Plasma extraction was performed on the day after blood samples were obtained and typically 15-l 8 h post-injection of radioligand. Stability of [1231]iomazenil in blood was demonstrated by analysis of blood samples with added radiotracer (standard A) stored under conditions identical to those for the blood samples. The recovery coefficient obtained from extraction and HPLC analysis of standards A and B was 93.2 & 5.6% (range 79.2-97.6%). The extraction efficiency (~0) of [lz3Iliomazenil from tissue homogenates was 94.8 + 4.1% (range 83.7-97.6%). Moreover, incubation of the radioligand in vitro with samples of human or non-human primate blood for 2 h at 37°C showed no detectable metabolism.
Normothermic
Monkey
98.4%
80 60
689%
40
Ed’d
20 0
120 29.9%
I 0
5
Rt (ho)
IO
15
0
qia Ext’d
I"~'l""1
’
Rt (min) to
Fig. 1. Representative HPLC profiles of the lipophilic fractions extracted with ethyl acetate from plasma samples at early and late time points in a human and in a hypothermic and a normothermic monkey. The concentrations of the radiometabohte and the parent compound appear next to each peak and are expressed as a percentage of total extracted activity. The radiometabolite eluted prior to the parent compound at a retention time (R,) of 4.5 min. Each panel indicates the time of sample collection (expressed in min post-injection of radioligand) and the % of total plasma activity which was extracted into ethyl acetate (% Ext’d).
15
885
Pharmacokinetics of [I” I)iomazenil 1. Relative distribution (meanf SD) of radioactivity betweenpolar, non-polar metabolitcsand the parent compound
Table
([‘“I]iomazenil) in arterial plasma samples (n = number of observations) at 60 min after injection in humans, hypo- and normothermic monkeys Relative distribution of radioactivity in plasma Experimental unit
Polar metabolite
Non-polar metabolite
Parent compound
Human (n = 5) Hypothermic (n = 2) Nonnothermic (n = 3)
92.3 & 2.3 70.2 + 16.7 56.7 + 4.1
1.1 +0.9 8.9 + 3.7 10.0 f 6.4
5.8 f 2.6 20.9 _+13.0 30.1 k 2.8
Tie plasma consisted almost exclusively of the parent compound whereas that in the non-human primate plasma represented both parent compound and a radiometabolite which increased over time. Reversedphase HPLC showed this metabolite to be less lipophilic than the parent compound (Fig. 1). By 60 min post-injection, extensive metabolism of the [‘231]iomazenil occurred in both human and monkey (Table 1). The relative distribution of radiometabolites and parent compound in the plasma samples of hypothermic and normothermic animals was not significantly different (P > 0.25), although the hypothermic animals showed greater variability. The coefficient of variance ranged from 24% for the polar metabolite to 62% for the parent compound. Plasma kinetics
The venous plasma concentration of free (nonprotein bound) [‘231]iomazenil approached that of arterial plasma within 5 min after injection of normal human volunteers (Fig. 2). The arterial clearance profile was characterized with a triexponential mathematical model [equation (l)], and the individual results of the five human volunteers are presented in Table 2. More than 97.8% of the plasma concentration of [‘231]iomazenil cleared with an average half time of 0.6 _+0.3 min. The volume of distribution of the parent compound at peak plasma levels was 23 + 10% (range 11-35%) higher than the blood volume estimated in each subject as 8% body weight (Dittmer, 1961). These data suggest that even by the
(mitt)
Fig. 2. The relative arterial (a) and venous (0) plasma clearance curves of the free parent compound [1231]iomazenil in a healthy human volunteer. Arterial peak levels of [lz3Iliomazenil occurred at 40 s post-injection, and are better displayed in the inset panel for early time points.
time of peak arterial plasma activity, the radioligand had diffused outside the vascular bed. In the normothermic monkeys, between 78.3 and 98.8% of the plasma (‘231]iomazenil had clearance times (T,) of 0.1-1.1 min (Table 3). The hypothermic monkeys showed lower plasma clearance levels of 53.4 and 71.4% (P c 0.05) and a trend towards slower clearance half times (Tt) of 2.3 and 1.0 min (P < 0.1) than those of the normothermic monkeys. Also, T, was significantly longer in the hypothermic monkeys (59.3-61.2min) than in the normothermic monkeys (28646.0 min; P < 0.025). In comparison to values from the normothermic monkeys, T,, T2, and T, values of the hypothermic animals more closely resembled those of the human subjects (P > 0.10). The volume of distribution of the parent compound at peak plasma levels was 73 +_13% (range 51-89%) larger than the total blood volume estimated in each animal as 7.5% body weight (Dittmer, 1961). Distribution in blood elements In vitro, [‘231]iomazenil established equal concentrations in plasma and cellular blood elements within 1 min at room temperature, and this ratio was main-
Table 2. Arterial plasma clearance half-times of [‘z’I]iomazenil in humans, showing T, , T2and T, in min and respective involvement (%) by the three compartments. The volume of distribution was determined from the concentration of radioactivity in plasma at peak level (40-100 s post-injection)
Subject
I 2 3 4 5 Mean + 1 SD
l
T2 12.3 (0.6%) 24.8 (0.8%) 10.7 (0.8%) 4.4 (2.1%) 5.1
73 40.6 (0.7%) 70.6 (0.3%) 61.9 (0.2%) 72.0 (0.1%) 158.4
0.6 f 0.3 (98.6 f 0.5%)
11.5+8.2 (1.1 f 0.7%)
80.6 + 45.0 (0.3 * 0.3%)
T, 1.1 (98.7%) 0.6 (98.9%) (9z%) 0.4 (97.8%)
*Blood samples not available at the early time points.
Volume of distribution (liters) 9.1 5.8 4.4 3.7 l
5.8 +_2.4
886
SAM
S. ~%GHBI el a/.
Table 3. Arterial plasma clearance half-times of [‘231Jiomazenil in hypothermic and normothermic monkeys, showing T, , T2and r, in min and the respective involvement (%) by the three compartments. The volume of distribution was determined from the concentration of radioactivity in plasma at peak levels (20-80 s post-injection) Experimental animal
No.
T,
Hvuothermic
Nucl. Med. Biol. Vol. 19, No. 8, pp. 881-888, 1992 Int. J. Radiat. Appl. Instrum. Part B Printed in Great Britain. All rights reserved
Pharmacokinetics of the SPECT Benzodiazepine Receptor Radioligand [ 123 I]Iomazenil in Human and Non-human Primates SAM1 S. ZOGHBI’*, RONALD M. BALDWIN’, JOHN P. SEIBYL2, MOHAMMED S. AL-TIKRITI’, YOLANDA ZEA-PONCE’, MARC LARUELLE2, ELZBIETA H. SYBIRSKA2, SCOTT W. WOODS2, ANDREW W. GODDARD’, ROBERT T. MALISON’, RALF ZIMMERMAN’, DENNIS S. CHARNEY2, EILEEN 0. SMITH’, PAUL B. HOFFER’ and ROBERT B. INNIS’ Departments of ‘Diagnostic Radiology and ZPsychiatry, Yale University School of Medicine, New Haven, CT 06510 and 3VA Medical Center, West Haven, CT 06516 U.S.A. (Received 19 March 1992)
The pharmacokinetics of [i231]iomazenil (Ro 16-0154) in 5 healthy human volunteers were compared to those in 2 hypothermic and 3 normothermic anesthetized monkeys. Following intravenous injection in humans and monkeys, [‘231]iomazenil rapidly diffused outside the vascular bed and was cleared from the arterial plasma triexponentially. The clearance half-times in hypothermic animals were protracted to values closer to those of the human. [1231]Iomazenilwas metabolized mainly to a polar radiometabolite (not extracted by ethyl acetate) in the human whereas an additional lipophilic radiometabolite was detected in the monkey. In vitro and in viuo studies showed that [iz31]iomazenil established equal concentrations in association with the cellular and plasma component of the blood, indicating that the plasma clearance of [‘231]iomazenil mirrors that of the blood. Analysis of organs from a monkey given [‘231]iomazenil showed that the parent compound was actively taken up by peripheral organs; the polar radiometabolite accumulated mainly in the bile and the kidneys whereas the non-polar radiometabolite accumulated in the urine and kidneys. Greater than 90% of the radioactivity in the different regions of
the brain was unchanged parent [‘231]iomazenil.
Introduction Iomazenil (Ro 16-0154) is an iodine-containing ligand for benzodiazepine (BZ) receptors and is a close analog of the fluorine-containing antagonist flumazenil, Ro 15-1788 (Beer et al., 1990). [‘251]Iomazeni1 binds reversibly and selectively to BZ receptors in brain tissue homogenates with high affinity (Ko = 0.5 nM at 37°C) and high ratio of specific to non-specific binding (40 : 1) (Johnson et al., 1990). In keeping with in vitro studies, SPECT imaging with [‘231]iomazeni1 has shown that the brain uptake of this radiotracer represents reversible and selective binding to the BZ receptor and that it has high potency (ED, = 15 nmol/kg i.v.) and high ratio of specific to non-specific binding (10: 1) (Beer et al., 1990; Innis er al., 1991b). The peak brain uptake of [‘231)iomazeni1 is unusually high (10-12% of injected *Address all correspondence to: Sami S. Zoghbi, Ph.D., Department of Diagnostic Radiology, 326 Brady Memo-
rial Laboratory, Yale University School of Medicine, 333 Cedar Street. New Haven, CT 06510, U.S.A.
dose) and shows a relatively slow washout of 3% per hour (Innis et al., 1991a). The high affinity of iomazenil for the BZ receptor may partially explain its high and slow washout. However, brain uptake its metabolism and clearance from the body also significantly determine its distribution over time in all body organs. The purpose of the present study was to develop methods for measurement of the free parent compound in plasma and to examine the kinetics and clearance of [‘231]iomazenil in human and non-human primates.
Materials and Methods Radiolabeling
The method of McBride et al. (1991) was modified to make [‘231]iomazenil suitable for intravenous injection. Absolute ethanol (77 pL) was added to a vial containing 50 pg (0.087 pmol) ethyl ‘I-(tributyl-stannyl)-5,6-dihydro-5-methyl-6-oxo-4H-imidazo-[l,5a][1,4]benzodiazepine-3-carboxylate and 100 Hg (0.22 pmol) 1,3,4,6- tetrachloro - 3a, 6a -diphenylglycoluril 881
882
S~hil
S. ZCIGHB~
(Iodogen, Pierce Chemical Co.) and the mixture was mixed by sonication for 3 min. No-carrier-added [ 123 Qodium iodide (20 mCi in 60 p L of 0.1 M NaOH; Nordion International Inc., Vancouver, Canada) was mixed with sufficient ethanol:2M NH,C1:0.5 M H,P04 (1: 1: 1 v/v/v) so that the final pH was 2.9-3.0. After 1.5 h at room temperature, the reaction was terminated by adding 0.1 mL (10 mg) aqueous NaHS03 (100 mg/mL). Volatile radioactivity, a byproduct of this reaction, was purged with a stream of argon into a charcoal column for 10-20 min, then 1 mL H,O and 0.2 mL saturated NaHCO, were added (final pH 7). The mixture was extracted with 3 x 1 mL toluene and the combined extracts were dried over anhydrous Na,SO,. The toluene was evaporated to dryness with a rotary evaporator under argon and the residue was dissolved in 110 PL methanol, followed by 90 PL water. Radiolabeled products were separated by high pressure liquid chromatography (HPLC), using a reverse-phase C- 18 column (Novapak 3.9 x 300 mm, Waters Division of Millipore, Milford, Mass.) and a mobile phase of methanol/water (55:45 v/v) at a flow rate of 0.7 mL/min, using an on-line NaI(T1) scintillation detector and a multichannel scaler. Under these conditions, the retention time of iomazenil was 8.4min, and those of possible by-products with chlorine and hydrogen in the 7-position were 6.9 and 5.8 min, respectively. The tributyltin precursor was retained on the column and was subsequently eluted with absolute methanol. The [‘231]iomazenil peak was collected in a flask containing 100~8 ascorbic acid and evaporated to dryness with argon. The residue was dissolved in 4OOpL absolute ethanol, diluted with 6mL normal saline, and sterilized by 0.2 pm membrane filtration (Acrodisc 13, Gelman Sciences, Ann Arbor, Mich.). Radiochemical purity was determined by HPLC in the same system. The preparation was stable for at least 24 h at room temperature. Sterility was confirmed by lack of bacterial growth in two media: trypticase and fluid thioglycollate. Apyrogenicity was confirmed using the limulus amebocyte lysate (LAL) test (Endosafe, Charleston, S.C.). Subjects Metabolic data were obtained in conjunction with an imaging study in normal volunteers (Woods et al., 1992). After giving written informed consent, 5 healthy human volunteers (2 female and 3 male) with ages of 28 f 8 y (with these and subsequent data expressed as mean + SD) and weights of 76 -+ 12 kg were administered 5.0 f 0.1 mCi [‘231]iomazenil intravenously. Five ovariectomized female baboons (9-l 1 kg Papio anubis) received 8.8 -+ 4.2 mCi [‘231Jiomazenil. The animals were divided into two groups: three normothermic baboons with body temperatures maintained at 3.5-37°C with a fluidcirculating blanket, and two hypothermic baboons with body temperatures maintained at 32-34°C. Monkeys were initially immobilized with ketamine
ef
al.
(15 mg/kg); anesthesia was induced with intravenous (i.v.) sodium pentobarbital (9.8 mg/kg) and maintained by sodium pentobarbital at approximately one third of the initial dose every 30min. The doses of radiopharmaceutical, measured in a gas inonization chamber dose calibrator (Capintec CRC-7, Montvale, N.J.), were injected over 20-30 s through an indwelling venous catheter. To prevent contamination from the site of radioligand injection, venous blood samples were collected from a line in the contralateral limb. Arterial blood samples were collected from an indwelling catheter in either the radial artery (human) or femoral artery (monkey). Two 10 mL blood samples were drawn in heparintreated syringes just prior to the adminis’tration of radioactivity. As controls, 10 ptci [‘231]iomazenil was added to the 10 mL blood samples at the time of injection (“standard A”) and at the time of metabolite analysis (“standard B”), approx. 15 h after injection. Beginning immediately after administration of radioactivity, 1.5 mL arterial blood samples were drawn continuously at IO s intervals for 120 s using a peristaltic pump (Harvard Apparatus Model 2501-001, Southwick, Mass.) and collected in test tubes containing 5OpL sodium heparin (5000 units/ml). Individual blood samples were then drawn in heparin-treated syringes at 2 min intervals for 10 min and subsequently at 15, 20, 30, 60 and 120 min. In the human, blood samples drawn at 1, 4, 30, 60 and 120 min were 10 mL. Samples were placed on ice within 30 min after drawing and were stored at 4°C until analysis. To verify recovery of radioactivity flowing through the peristaltic pump tubing, 20 PCi [‘231]iomazenil was added to 20mL of heparinized whole blood and drawn through the pump at the standard human flow rate of 9 mL/min with 1.5 mL samples collected every 10 s. Immediately thereafter, a 20 mL sample of non-radioactive blood was drawn through the pump and similar fractions were collected. All fractions were counted as below. Metabolite analysis Radioactive blood samples and standards A and B were centrifuged at 1000 g for 10 min. The concentration of radioactivity in plasma and in cellular blood elements was counted in equal volume aliquots (50 ,QL) in an automatic well-type y-counter (Model 8000, Beckman Instruments, Fullerton, Calif.). The counting efficiency of the system was determined using sources of 123Icalibrated in the gas ionization chamber with geometry similar to that of the blood samples. All radioactivity measurements were decaycorrected to the time of radioligand injection. Plasma and cellular fractions were extracted three times with equal volumes of ethyl acetate; extraction was calculated from the activity in the aqueous phase counted before and after extraction at a constant geometry and corrected for decay. For cellular blood elements in samples at 4,60 and 120 min, and standards A and
883
Pharmacokinetics of [I23Iliomazenil B, extraction was done after the addition of distilled water to produce hemolysis. The precision of the method was measured in one experiment by repeatedly extracting aliquots of a single sample of blood obtained at 60min post-injection; the coefficient of variation was 5.5% for 10 mL samples (n = 3), and 15.0% for 2mL samples (n = IO). Beginning with samples of lowest activity, selected organic extracts (usually samples collected at 1,4, 30,60 and 120 min) were evaporated to dryness on a rotary evaporator under argon in a 37°C water bath. The residue was dissolved in 82.5 PL methanol, diluted with 67.5 PL H,O and injected into the HPLC apparatus using the same conditions as above but with a column reserved for low-level radioactivity. Effluent fractions of 0.5 mL were collected in test tubes using a fraction collector (Microfractionator FC-80K, Gilson, Middleton, Wis.) and counted with the automatic y-counter to reconstruct a histogram of the HPLC profile. The parent fraction in the intermediate samples not chromatographed was estimated by interpolation. The composition of the samples was determined as the product of the fraction extracted and the parent composition (HPLC) divided by a recovery coefficient calculated as the product of the extraction and fraction of [123Iliomazenil (measured by HPLC) in standards A and B. The in vitro distribution of [‘231]iomazenil between plasma and cellular blood elements was determined following addition of radioligand to whole blood at approx. the same concentration as in the in uiuo studies. The blood was then centrifuged and plasma was separated from cells. Each fraction was counted for radioactivity and the relative distribution of radioactivity between cells and plasma was calculated. The cells were then washed three more times with equal volumes of non-radioactive plasma. Protein binding
The binding of [1231]iomazenil to plasma proteins by ultrafiltration through Centriconmembrane filters (Amicon Division, W.R. Grace & Co., Danvers, Mass.). In addition to plasma from standards A and B, 10 PL of a 1: 100 dilution of [‘231]iomazenil was added to 1 mL each of plasma and normal saline as control. Duplicate samples of 200 PL were placed in the unit and centrifuged 20 min at 10,OOOg; the separated compartments were counted on the y-counter. The protein bound fraction was calculated as the ratio of the radioactivity on the filter to the sum of the filter and filtrate. The same process was applied in the study of the effect of increasing amounts of added radioligand (0.02-0.8 PCi) to a constant 500 PL plasma. in vitro was determined
Data analysis
The plasma clearance profile of the free parent compound (non-protein bound) was obtained by expressing the amounts of [123IJiomazenil relative to the peak level of radioactivity in the respective
samples. The data were fitted to exponential curves with a graphics program (Kaleidagraph@ for the Macintosh, version 2.1, Synergy Software, Redding, Pa). Using a stripping technique and three component model, the relative clearance, C(t), of the free parent compound was described by the formula: C(t) = fie-'O.WW +f2e-W'3/W +f,e-W93/Td~ (1) where T, , T2 and T, are the half-times for clearance of the three components, and fi , f2and f3represent the fractional involvement of the three components relative to the initial concentration, defined as the ratio of the y-intercept for each component divided by the sum of the three intercepts. The volume of distribution was determined by dividing the injected dose of radioactivity (PCi) by the concentration of radioactivity (pCi/mL) present in the plasma at the peak level. Average values were expressed as mean +_ 1 SD. Statistical significance testing of the means was done by first comparing variances using the F test followed by the appropriate upper t test (Burr, 1972). Biodistribution in non-human primate 2 h after injection of 15.9 mCi [‘231]iomazenil in a separate hypothermic baboon (10.0 kg), the animal was euthanized with an overdose of pentobarbital. Tissue samples (c. 0.5 g) of organs of interest were weighed; brain was further dissected into occipital gray matter, frontal gray matter, white matter and cerebellum. Total radioactivity was counted in the y-counter and the uptake was calculated as percent injected dose per gram (X ID/g). Tissue samples were homogenized in a Polytron tissue disrupter (Brinkmann Instruments, Lucerne, Switzerland) while cooling in ice. Aliquots of the homogenates were extracted three times with an equal volume of ethyl acetate; the extraction fraction (E,) for each tissue was calculated and the parent compound quantified by HPLC as described above. The extraction fraction (E]) was further corrected by dividing it by its respective extraction efficiency (~g) which was determined by adding excess amounts of [123Iliomazenil (internal standard) to identical aliquots of the tissue homogenates and subjecting them to the same extraction procedure. Radioactivity was quantitated by counting each sample before (n,) and after (n2) the addition of internal standard and after extraction with ethyl acetate (n,). The extraction efficiency (so) was then calculated by the formula so = (n2 - n3 - el nl)/(n2 - nl ).
Results Radiolabeling
The final preparation was obtained in 54 + 22% (range 32-82%) overall yield and in greater than 99% radiochemical purity. At the termination of the iodination reaction, an average of 24 & 17% (range O-46%) of the total radioactivity was volatile. The
Sahn
884
S. ZoGHsr et al.
volatile fraction could be trapped in organic solvents such as toluene or absolute ethanol, but not in a basic solution of sodium bisulfite. The specific activity of these preparations was not measured directly; however, the tributyltin precursor and selected prep arations were analyzed by HPLC with U.V. detection and no evidence of iomazenil carrier or its analogs with H- or Cl-substitution were observed. The limit of detection was on the order of 5 x 10e3 pmol, which would imply a specific activity of greater than 4000 mCi/pmol; however, based on the results from McBride et al. (1991) in which the entire batch of a much larger scale reaction was analyzed, the specific activity probably is closer to 180,000 mCi/pmol. Protein binding
By the ultrafiltration method, 81.0 f 6.3% [‘231]iomazenil was protein bound in human plasma compared to 76.6 + 2.1% in the non-human primate. The two were not statistically different (P > 0.05). Relative protein binding was not affected by increasing concentrations of the radioligand in plasma over a 40-fold range encompassing concentrations typically found in the human and the non-human primate plasma samples. Recovery of radioactivity
Radioactivity recovered from the peristaltic pump used in early arterial sampling represented 99.4% of the total added activity in control experiments.
Metabolites
At the time of peak level of radioactivity in arterial plasma (40-100 s in the human and 20-80 s postinjection in monkey) 97.0 + 4.2% of activity in human plasma and 91.5 + 2.3% of activity in monkey plasma represented parent compound. Extraction with ethyl acetate allowed an operational definition of two fractions of plasma radioactivity: (1) a polar fraction which was not extracted and which increased over time; and (2) a lipophilic fraction which W(IS extracted and which decreased over time in both human and monkey. The lipophilic fraction in human
Hypothermic Monkey
Human loo
Only 0.4% of the radioactivity contaminated the first non-radioactive blood sample, and 0.3% was associated with the catheter. Plasma extraction was performed on the day after blood samples were obtained and typically 15-l 8 h post-injection of radioligand. Stability of [1231]iomazenil in blood was demonstrated by analysis of blood samples with added radiotracer (standard A) stored under conditions identical to those for the blood samples. The recovery coefficient obtained from extraction and HPLC analysis of standards A and B was 93.2 & 5.6% (range 79.2-97.6%). The extraction efficiency (~0) of [lz3Iliomazenil from tissue homogenates was 94.8 + 4.1% (range 83.7-97.6%). Moreover, incubation of the radioligand in vitro with samples of human or non-human primate blood for 2 h at 37°C showed no detectable metabolism.
Normothermic
Monkey
98.4%
80 60
689%
40
Ed’d
20 0
120 29.9%
I 0
5
Rt (ho)
IO
15
0
qia Ext’d
I"~'l""1
’
Rt (min) to
Fig. 1. Representative HPLC profiles of the lipophilic fractions extracted with ethyl acetate from plasma samples at early and late time points in a human and in a hypothermic and a normothermic monkey. The concentrations of the radiometabohte and the parent compound appear next to each peak and are expressed as a percentage of total extracted activity. The radiometabolite eluted prior to the parent compound at a retention time (R,) of 4.5 min. Each panel indicates the time of sample collection (expressed in min post-injection of radioligand) and the % of total plasma activity which was extracted into ethyl acetate (% Ext’d).
15
885
Pharmacokinetics of [I” I)iomazenil 1. Relative distribution (meanf SD) of radioactivity betweenpolar, non-polar metabolitcsand the parent compound
Table
([‘“I]iomazenil) in arterial plasma samples (n = number of observations) at 60 min after injection in humans, hypo- and normothermic monkeys Relative distribution of radioactivity in plasma Experimental unit
Polar metabolite
Non-polar metabolite
Parent compound
Human (n = 5) Hypothermic (n = 2) Nonnothermic (n = 3)
92.3 & 2.3 70.2 + 16.7 56.7 + 4.1
1.1 +0.9 8.9 + 3.7 10.0 f 6.4
5.8 f 2.6 20.9 _+13.0 30.1 k 2.8
Tie plasma consisted almost exclusively of the parent compound whereas that in the non-human primate plasma represented both parent compound and a radiometabolite which increased over time. Reversedphase HPLC showed this metabolite to be less lipophilic than the parent compound (Fig. 1). By 60 min post-injection, extensive metabolism of the [‘231]iomazenil occurred in both human and monkey (Table 1). The relative distribution of radiometabolites and parent compound in the plasma samples of hypothermic and normothermic animals was not significantly different (P > 0.25), although the hypothermic animals showed greater variability. The coefficient of variance ranged from 24% for the polar metabolite to 62% for the parent compound. Plasma kinetics
The venous plasma concentration of free (nonprotein bound) [‘231]iomazenil approached that of arterial plasma within 5 min after injection of normal human volunteers (Fig. 2). The arterial clearance profile was characterized with a triexponential mathematical model [equation (l)], and the individual results of the five human volunteers are presented in Table 2. More than 97.8% of the plasma concentration of [‘231]iomazenil cleared with an average half time of 0.6 _+0.3 min. The volume of distribution of the parent compound at peak plasma levels was 23 + 10% (range 11-35%) higher than the blood volume estimated in each subject as 8% body weight (Dittmer, 1961). These data suggest that even by the
(mitt)
Fig. 2. The relative arterial (a) and venous (0) plasma clearance curves of the free parent compound [1231]iomazenil in a healthy human volunteer. Arterial peak levels of [lz3Iliomazenil occurred at 40 s post-injection, and are better displayed in the inset panel for early time points.
time of peak arterial plasma activity, the radioligand had diffused outside the vascular bed. In the normothermic monkeys, between 78.3 and 98.8% of the plasma (‘231]iomazenil had clearance times (T,) of 0.1-1.1 min (Table 3). The hypothermic monkeys showed lower plasma clearance levels of 53.4 and 71.4% (P c 0.05) and a trend towards slower clearance half times (Tt) of 2.3 and 1.0 min (P < 0.1) than those of the normothermic monkeys. Also, T, was significantly longer in the hypothermic monkeys (59.3-61.2min) than in the normothermic monkeys (28646.0 min; P < 0.025). In comparison to values from the normothermic monkeys, T,, T2, and T, values of the hypothermic animals more closely resembled those of the human subjects (P > 0.10). The volume of distribution of the parent compound at peak plasma levels was 73 +_13% (range 51-89%) larger than the total blood volume estimated in each animal as 7.5% body weight (Dittmer, 1961). Distribution in blood elements In vitro, [‘231]iomazenil established equal concentrations in plasma and cellular blood elements within 1 min at room temperature, and this ratio was main-
Table 2. Arterial plasma clearance half-times of [‘z’I]iomazenil in humans, showing T, , T2and T, in min and respective involvement (%) by the three compartments. The volume of distribution was determined from the concentration of radioactivity in plasma at peak level (40-100 s post-injection)
Subject
I 2 3 4 5 Mean + 1 SD
l
T2 12.3 (0.6%) 24.8 (0.8%) 10.7 (0.8%) 4.4 (2.1%) 5.1
73 40.6 (0.7%) 70.6 (0.3%) 61.9 (0.2%) 72.0 (0.1%) 158.4
0.6 f 0.3 (98.6 f 0.5%)
11.5+8.2 (1.1 f 0.7%)
80.6 + 45.0 (0.3 * 0.3%)
T, 1.1 (98.7%) 0.6 (98.9%) (9z%) 0.4 (97.8%)
*Blood samples not available at the early time points.
Volume of distribution (liters) 9.1 5.8 4.4 3.7 l
5.8 +_2.4
886
SAM
S. ~%GHBI el a/.
Table 3. Arterial plasma clearance half-times of [‘231Jiomazenil in hypothermic and normothermic monkeys, showing T, , T2and r, in min and the respective involvement (%) by the three compartments. The volume of distribution was determined from the concentration of radioactivity in plasma at peak levels (20-80 s post-injection) Experimental animal
No.
T,
Hvuothermic
