388
Chem. Res. Toxicol. 2007, 20, 388-399
Physiologically Based Pharmacokinetic/Pharmacodynamic Model for Acrylamide and Its Metabolites in Mice, Rats, and Humans John F. Young,† Richard H. Luecke,‡ and Daniel R. Doerge*,§ DiVisions of Biometry & Risk Assessment and Biochemical Toxicology, National Center for Toxicological Research (NCTR)/Food and Drug Administration, Jefferson, Arkansas 72079, and Department of Chemical Engineering, UniVersity of MissourisColumbia, Columbia, Missouri 65211 ReceiVed October 25, 2006
A physiologically based pharmacokinetic model was developed for acrylamide (AA) and three of its metabolites: glycidamide (GA) and the glutathione conjugates of acrylamide (AA-GS) and glycidamide (GA-GS). Liver GA-DNA adducts and hemoglobin (Hb) adducts with AA and GA were included as pharmacodynamic components of the model. Serum AA and GA concentrations combined with urinary elimination levels for all four components from male and female mice and rats were simulated from iv and oral administration of 0.1 mg/kg AA or 0.12 mg/kg GA. Adduct formation and decay rates were determined from a 6 week exposure to approximately 1 mg/kg AA in the drinking water and subsequent 6 week nonexposure period. Human urinary excretion data and Hb adduct data were utilized to extrapolate to a human model. The steady-state human liver GA-DNA adduct level from exposure to background levels of AA in the diet was predicted to be between 0.06 and 0.26 adducts per 108 nucleotides. Introduction Acrylamide (AA; see structure in Figure 1) is an important industrial chemical with an annual worldwide production estimated at >200 million kg. The recent discovery that AA is formed during cooking of many common starchy foods (1, 2) has stimulated research on the potential for human toxicity. The concerns are based on observations of AA neurotoxicity in experimental animals (reviewed in ref 3) and humans (4), formation of multiple DNA adducts (5), mutagenicity in somatic (reviewed in ref 6) (7) and germ cells (8), mutagenicity in vivo (9), and carcinogenicity in several organs of rats following chronic lifetime exposure (reviewed in ref 6). The experimental evidence cited above strongly links genotoxic effects with the metabolism of AA to its epoxide, glycidamide (GA; see structure in Figure 1) by the action of hepatic CYP 2E1. The earliest reported pharmacokinetic data for AA were from Miller et al. (10) who administered single iv or oral 14C-AA doses to male Fisher 344 rats and reported total radioactivity for up to 7 days in blood, plasma, and 11 tissues; they also determined the AA concentration in blood and five tissues for the first 24 h after the 10 mg/kg iv dose. Excretion of total radioactivity was primarily (>92%) into the urine. NONLIN was used to evaluate the pharmacokinetic parameters. Raymer et al. (11), in a primarily analytical methods paper, reported AA serum data out to 8 h following a single 75 mg/kg AA intraperitoneal (IP) dose to male Long Evans hooded rats. RSTRIP was used to estimate the elimination half-life from the serum. Barber et al. (12) administered 14C-AA to male SpragueDawley rats as single and multiple IP (50 mg/kg) or oral (20 mg/kg) doses and then used HPLC to measure AA and the metabolite GA plasma levels out to 10 h. WIN NONLIN was * To whom correspondence should be addressed. Tel: 870-543-7943. Fax: 870-543-7720. E-mail:
[email protected]. † Division of Biometry & Risk Assessment, National Center for Toxicological Research. ‡ University of MissourisColumbia. § Division of Biochemical Toxicology, National Center for Toxicological Research.
Figure 1. Structures of AA, its metabolites, and the associated circulating and urinary biomarkers.
used to estimate the pharmacokinetic parameters for a onecompartment model. AA and GA hemoglobin (Hb) adducts to cysteine residues were reported from the multiple IP and oral AA dosing. A physiologically based pharmacokinetic (PBPK) model comprised of five compartments each for AA and GA was presented by Kirman et al. (13). The model utilized MichaelisMenten kinetics for the metabolism of AA to GA and urinary elimination of both through the conjugation with glutathione. The blood and tissue data from Miller et al. (10) and Raymer et al. (11), as well as the urinary elimination data of Sumner et al. (14), were used to test the model. Doerge et al. (15, 16) have reported the low dose (0.1 mg/kg) toxicokinetics of intravenous and orally administered AA and GA in male and female mice and rats. Serum and tissue
10.1021/tx600287w CCC: $37.00 © 2007 American Chemical Society Published on Web 02/27/2007
Acrylamide PBPK/PD Model
Chem. Res. Toxicol., Vol. 20, No. 3, 2007 389
Figure 2. Block diagram for the metabolism of AA to GA and further metabolism of both to their glutathione conjugates. All pharmacokinetic and pharmacodynamic (PD) processes are first order.
concentrations of AA and GA were reported and analyzed using a model-independent spreadsheet approach. Hb adducts with AA and GA at the N-terminal valine residue and liver GADNA adducts were measured in these studies. These NCTR studies are the basis of the PBPK/pharmacodynamic (PD) model reported in this paper. The goal of this work is to develop a PBPK/PD model for AA and GA disposition and adduct formation that simulates all of the mouse and rat data gathered in our laboratory and to utilize that same model to simulate the human data available in the literature in order to predict the potential DNA adduct levels in the human population resulting from dietary exposure to AA.
Materials and Methods Software. The general purpose PBPK/PD model called PostNatal developed at NCTR was used to simulate all of the data generated in the NCTR laboratories and those found in the literature. This Windows-based program controls four PBPK models under one shell with multiple input and output options. Each PBPK unit is comprised of 28 organ/tissue/fluid components and can be maintained independently or connected through metabolic pathways. AA, GA, and their glutathione conjugates (AA-GS and GA-GS) occupy
the four PBPK units. Even though each chemical and metabolite involves a full PBPK model, the model is effectively limited through the use of unity-based partition coefficients. Only those tissues specifically analyzed for AA or GA are partitioned differently from the blood compartment. The specific organ/tissue weights and blood flows are all set internally by the PostNatal program based on animal species, gender, and total body weight. Optimization is based on minimizing the weighted sum of squares of the difference between each data point and its simulated value. The block diagram depicting the metabolism of AA is illustrated in Figure 2. Dose administration can be either AA or GA. All metabolism is first order. Elimination is first order into urine for AA and all of the metabolites, including GA and the mercapturic acid derivatives of the AA-GS and GA-GS conjugates. The pharmacodynamic component of the model is adduct formation with DNA (N7-GAGua) and Hb from AA and GA (AA-valine and GA-valine) but not from either glutathione conjugate. AA forms only a Hb adduct while GA forms both a Hb adduct and DNA adducts measured in the liver and other tissues. Toxicokinetic Data. Toxicokinetic data were generated by the NCTR laboratories using B6C3F1 mice and Fischer 344 rats (15-17). From both male and female rodents, serum data for AA and GA were simulated from single dose iv or gavage administration to mice (0.1 and 50 mg/kg AA; 0.12 mg/kg GA) and rats (0.1 mg/kg AA; 0.12 mg/kg GA) and 30 min of access to feed containing AA with a target dose of 0.1 mg/kg for mice and rats. From additional groups of male and female rodents, 24 h urinary elimination profiles were determined from the various routes of AA administration. Hb adducts and liver GA-DNA adducts were obtained at the terminal time point from the gavage studies, and Hb adducts alone were obtained from the iv and feeding studies. Hb adducts and liver GA-DNA adducts were obtained from rodents across time from a 6 week drinking water exposure to approximately 1 mg/kg AA followed by a nonexposure recovery for an additional 6 weeks. Actual doses were determined based on weekly averages in individual body weights and total cage water consumption data. A summary of these studies is listed in Table 1. The mean weights for the B6C3F1 mice used in this study were 19.8 g for males and 17.3 g for females; for the Fischer 344 rats, the mean weights were 143 g for males and 113 g for females. Tissue samples were obtained at two sacrifice times following the low dose gavage administration (0.1 mg/kg AA or 0.12 mg/kg GA) to male and female mice and rats. Tissue samples of liver, muscle, brain, lung (mice), testes (male rats), and mammary (female rats) obtained 1 and 2 h postadministration in mice and 2 and 4 h in rats were analyzed for AA and GA. Simulations of these tissue
Table 1. NCTR’s AA Studies dose (mg/kg)
compd dosed
serum analysis AA GA
gender
mouse
M and F M and F
0.12 0.12
GA GA
iv gavage
single single
M and F M and F
0.1 0.1
AA AA
iv gavage
single single
yes yes
yes yes
M and F M and F M and F
∼0.1 50 ∼1
AA AA AA
diet gavage drinking water
single/30 min single 42 days
yes yes yes
yes yes yes
rat
route
single or multidose
species
yes yes
M and F M and F
0.12 0.12
GA GA
IV gavage
single single
yes yes
M and F
0.1
AA
IV
single
yes
yes
M and F
0.1
AA
gavage
single
yes
yes
M and F M and F
∼0.1 ∼1
AA AA
diet drinking water
single/30 min 42 days
yes yes
yes yes
adduct data GA Hb adducts at 6 h GA liver and GA Hb adducts at 8 h AA and GA Hb adducts at 6 h GA liver and AA and GA Hb adducts at 8 h AA and GA Hb adducts at 12 h GA liver adduct values out to 8 h GA liver and AA and GA Hb adduct values out to 42 days and then decay GA Hb adducts at 8 h GA liver and GA Hb adducts at 10 h AA and GA Hb adducts at 8 h GA liver and AA and GA Hb adducts at 10 h AA and GA Hb adducts at 12 h GA liver and AA and GA Hb adduct values out to 42 days and then decay
390 Chem. Res. Toxicol., Vol. 20, No. 3, 2007
Young et al.
samples were used to establish the partition coefficients for all doses and routes of administration. Pharmacokinetic literature data from the single and multiple IP administration of 50 mg/kg AA to male rats were combined and simulated using this same PBPK/PD model: Barber et al. (12) (AA and GA plasma concentration data), Sumner et al. (18) (urinary excretion profile and Hb AA and GA adducts), and Doerge et al. (17) (liver GA-DNA adducts). The 20 mg/kg AA single and multiple oral data from Barber et al. (12) and the 75 mg/kg AA single IP data of Raymer et al. (11) were each combined with the urinary elimination data of Sumner et al. (14, 18) and simulated in a similar manner. The partition coefficients determined from the above NCTR studies were used for all of the literature simulations. Data from graphical presentations in the literature were obtained by scanning the figures and saving them as pcx files. These files were in turn analyzed utilizing Un-Scan-It (Silk Scientific, Inc., Orem, UT), which digitized the data from graphs into x,ycoordinates. PostNatal Simulations. Pharmacokinetic analysis of the data occurred in four phases with the male and female mouse and rat each simulated independently. All partition coefficients were set to 1 except for those specifically measured; those tissue partition coefficients were initially set at the tissue/serum ratio. The GA administration was modeled first since it was a simpler simulation using only half of the model structure, i.e., PBPK-2 (GA) and PBPK-4 (GA-GS) in Figure 2. Initially, the GA iv dose was simulated since only three parameters were necessary to fit the GA serum data and GA and GA-GS elimination data: the first-order metabolism rate constant for the conversion of GA to its glutathione conjugate (GA-GS) and the GA and GA-GS to urine first-order elimination rate constants. Following administration of the 0.12 mg/ kg GA gavage dose, the GA serum level, GA tissue data, and GA elimination estimates were all fit by varying mainly the GA absorption parameter but also allowing the metabolism and elimination components to vary as well. Holding these GA parameters constant, the AA and GA serum data and all elimination estimates from the 0.1 mg/kg AA iv dose were simulated by varying the first-order metabolism rate constants for the conversion of AA to GA and AA to its glutathione conjugate (AA-GS) and the urinary elimination rate constants for AA and AA-GS. The AA gavage dose was then utilized to evaluate these same parameters plus the AA absorption rate constant and partition coefficients when simulating the AA tissue levels. Once these AA parameters were optimized while holding the GA parameters constant from the previous simulation, all metabolic and elimination parameters in the model were allowed to vary so that an optimal fit could be obtained for the complete AA and GA data set from the AA administration. The third phase of the simulations involved fitting the diet and high dose gavage data sets and utilizing the urinary excretion fractions reported by Sumner et al. (14) (high dose gavage in mice, AA-GS:GA-GS:GA proportions of 41:33:26) while holding the partition coefficients constant and varying only the absorption, metabolism, and elimination parameters. The fourth phase was simulating the pharmacodynamic adduct data. Holding the pharmacokinetic parameters constant, only the formation and decay rate constants of the adducts were allowed to vary. Initially, the drinking water adduct data from a target dose of 1 mg/kg/day were simulated since the study design included exposure and sampling that would account for both the accumulation and the decay of the adducts. Once obtained, the decay rate was held constant and only the formation rate was allowed to vary when fitting the single time point adduct data. The equation that governs this pharmacodynamic interaction is as follows: dX/dt ) Kf × Yi (organj, chemi) - Kd × X
(1)
where X is the adduct concentration, Y is the concentration of chemical i in the jth organ, Kf is the formation rate constant, and Kd is the degradation rate constant; both rate constants are first order. The adduct formation is considered to be at such a low level
Table 2. Partition Coefficients Derived from the Low Dose Gavage Studies GA administration mouse
rat
AA administration mouse
rat
male female male female male female male female GA brain GA liver GA lungs GA muscle GA adipose all other GA AA brain AA liver AA lungs AA muscle AA gonads AA adipose all other AA
0.73 0.39 0.68 0.85 1 1
0.60 0.45 0.69 0.70 1 1
1.003 0.44 1 1.115 1 1
0.89 0.44 1 0.845 0.45 1
0.85 0.53 0.92 1.23 1 1 0.90 0.90 0.63 1.35 1 1 1
1.01 0.75 1.42 1.47 1 1 0.94 1.09 0.78 1.47 1 1 1
1.16 0.44 1 1.1 1 1 0.75 0.32 1 0.74 0.45 1 1
1 0.44 1 0.95 0.55 1 0.75 0.42 1 0.63 1 0.39 1
as to have no effect on the kinetics of Yi; that is, dYi/dt does not contain a component for adduct formation. The validity of this assumption was previously tested experimentally. Tareke et al. (23) reported that for administration of single doses of AA at 0.1 mg/ kg, the percentage of serum AUC for either AA or GA that was estimated to be present as adducts with either Hb or DNA was less than 1.5%. Human Data Simulation. Because there were no serum concentration data available from human exposure studies, simulations were based on similar model considerations and available exposure and elimination data. The individual excretion curves for AA, AA-GS, and GA-GS from a low dose AA dietary administration to three men and three women reported by Fuhr et al. (19) were used as the foundation to estimate the absorption, metabolism, and excretion parameters for the human model (individual weights were provided by U. Fuhr, personal communication). As GA was not found above the detection limit in any urine sample and glyceramide was not measured in the Fuhr et al. (19) study, an estimate of the excretion of these metabolites was made based on the ratio of total GA:AA-GS excretion from a 3 mg/kg dose reported in Fennell et al. (20). That single GA elimination data point (15.8% of final AA-GS value) was added to the Fuhr et al. (19) elimination curves for each individual. These data were supplemented by the AA and GA Hb adduct data from background dietary exposures to AA in the general human population reported by Boettcher et al. (21), which were used to estimate a human exposure dose and in turn to estimate a human liver GA-DNA adduct level. Statistical Comparisons. All parameter comparisons of mean data from both genders of rats (n ) 5-7) and humans (n ) 3) were based on Students t test using two-tail statistics. Only values 2p < 0.05 were considered significant.
Results Pharmacokinetics in Rodents. The partition coefficients utilized for all of the simulations in this study are reported in Table 2. These values are based on tissue samples taken from two to four rats or mice from the low dose gavage studies. All partition coefficient values are fairly close to unity. Even though many of the partition coefficients were unity, the full 28 components were maintained for each of the four PBPK models. This was done since the PostNatal software maintained the full complement of components and cost nothing in simulation time or user effort while changing dose, route of administration, or species. There are slight variations between the GA partition coefficients obtained from dosing by either GA or AA. These differences contribute to the variation in the formation rate constant for the GA-DNA tissue adducts. Simulations of the low dose rat data were based on individual animal serial blood samples; means and standard deviations of
Acrylamide PBPK/PD Model
Figure 3. Pharmacokinetic (solid lines) and pharmacodynamic (broken lines) simulation of data (symbols) following gavage administration of 0.1 mg/kg AA to an individual male rat. Serum concentration data for AA (solid squares) and GA (solid circles) are in nmol/mL. The liver GA concentration curve (nmol/g) is included to illustrate its relationship to the GA serum curve. The number of adducts for Hb (Hb AA, open square; Hb GA, open circle) is in pmol/g globin, and the number of liver GA-DNA adducts (open diamonds) is in adducts per 108 nucleotides.
Figure 4. Pharmacokinetic and pharmacodynamic simulation of data following administration of 0.121 mg/kg AA in the diet (30 min access) to an individual female rat. Units and symbols are the same as in Figure 3.
the derived parameters were based on 5-7 rats per route of administration. Typical simulations of the oral dosing are given for individual rats in Figures 3 (male rat, gavage) and 4 (female rat, diet). Because the sensitivity of the analytical procedure would not allow serial samples from an individual mouse, a single serum curve based on mean data from three mice per sacrifice time was simulated for each route of administration. Rat iv data were collected in a similar manner for technical reasons (16). Figures 5 (female mouse, gavage) and 6 (male mouse, diet) illustrate the simulation of the mean mouse data. Figure 7 illustrates the simulation of the 50 mg/kg gavage in male mice. For this high dose simulation, the urinary excretion proportions from Sumner et al. (14) were used to calculate target values for AA-GS, GA, and GA-GS elimination. The urinary elimination curves for the low dose simulations (Figures 3-6) are not included in each figure to simplify the data presentation. The first-order absorption, metabolism, and elimination rate data are presented in the top portion of Tables 3 (rats) and 4 (mice). For the rat, the serum concentration data were adequately fit by specifying 100% of the administered dose absorbed directly from the stomach (i.e., fraction absorbed from the stomach ) 1, Table 3). Absorption, as expected, was the most variable pharmacokinetic parameter with the coefficient of variation (i.e., standard deviation/mean) ranging from a low of 13% for female
Chem. Res. Toxicol., Vol. 20, No. 3, 2007 391
Figure 5. Pharmacokinetic and pharmacodynamic simulation of data following gavage administration of 0.1 mg/kg AA to a set of mean data (n ) 3 per sacrifice time) from female mice. Variance as represented by (1 standard deviation (horizontal dash above and below data symbol) is contained within the size of the symbol except where specifically presented. Units and symbols are the same as in Figure 3.
Figure 6. Pharmacokinetic and pharmacodynamic simulation of data following administration of 0.16 mg/kg AA in the diet (30 min access) to a set of mean data (n ) 3 per sacrifice time) from male mice. Variance as represented by +1 standard deviation (horizontal dash above the symbol) is contained within the size of the symbol except where specifically presented. Units and symbols are the same as in Figure 3.
Figure 7. Pharmacokinetic and pharmacodynamic simulation of data following gavage administration of 50 mg/kg AA to a group of five male mice from a pilot study. Units and symbols are the same as in Figure 3 with the addition of urinary target value (solid triangles) amounts (nmol), which were based on data from Sumner et al. (14).
rat AA gavage to a high of 51% for male rat GA gavage. Absorption from the diet was slower (i.e., longer half-life) than from the gavage (2p < 0.001), again as expected. There was no difference in absorption half-life between genders for GA gavage, but there was for AA gavage (2p < 0.001) and AA
min
GA-GS to urine
129 ( 19 a,c (109-151) 5.5 ( 0.8 a (4.5-6.6)
pmol/g globin pmol/g globin in 108 nucleotides
143
0.0114
min-1
0.0023
37
48
0.0023
45.7
min-1
47.4 ( 12.5 a,d (38-69) 4.9 ( 0.4 a (4.4-5.4)
11.4
4.3 69.4 5.5 20.8
12.4
0.0024
0.0023
48.2
16.9 83.1
11.3 ( 0.8 a (10-13)
min-1
min-1
min-1
20.8 79.2
10.0
12.0
9.2
1.5
AA
AA
AA 0.96 drinking water 1c 1c
14.9
16.2
31
36.5
21.2
10.1
8.8
10.2
35.4
37.3
11.8
5.2
9232 651
105
12970
0.0114
0.0023
0.0024
8870
3820
11.0 ( 1.8 a (9-14) 95 ( 14 b (80-119) 2.9 ( 0.8 b (2-4)
0.0023
0.0024
0.0114 21.0 ( 4.9 b (15-29) 98.5 ( 19.8 b (76-123)
0.0023
0.0024
10.2
11.3
25.7
37
12.3
5.9
1
1
elimination pattern (% of excretion) same as 69.6 low dose 6.7 gavage 23.7
low dose gavage
same as
17g
1c
75 IP
literature AAf
40g
pharmacodynamic rate constantsa 13.9 ( 2.0 b 11.0 5.4 (11-17) 70.9 ( 12.6 b 50.1 44.7 (62-96) 6.0 8.7
4.5 34.0 13.4 48.1
27.1 ( 3.4 b (23-33) 16.2 ( 2.0 a (13-20) 19.2 ( 3.7 a (13-23) 1.3 ( 0.2 b (1.0-1.6) 4.4 ( 0.9 b (2-6) 3.6 ( 0.2 b (3-4) 3.9 ( 0.2 c (3-5)
1c
50 IP
literature AAe
parametersa
20 gavage
pharmacokinetic rate 344 ( 65 c infusion 108 (281-450) 1 1
7
0.094 diet
0.0114
0.0023
5.5 ( 1.1 a (4.2-7.1) 83.8 ( 9.6 b (67-95) 6.3 ( 1.2 b (4.6-7.7) 0.0024
2.4 46.0 8.9 42.6
19.1 ( 4.1 a (13-27) 17.2 ( 3.7 a (12-24) 22.4 ( 4.5 a (18-31) 1.6 ( 0.3 a (1.2-2.3) 11.4 ( 2.4 a (9-16) 13.3 ( 1.3 a (11-16) 14.0 ( 0.9 b (12-16)
189 ( 52 b (142-272) 1
7
0.1 gavage
literature AAd GA
177
0.0025
48.2
26.8 73.2
5.67
6.89
12.5
1c
0.12 iv
GA
165 ( 48 a (104-227) 6.9 ( 1.1 c (4.9-8.4)
0.0118
0.0025
41.3 ( 9.8 a (25-56) 4.21 ( 0.52 c (3.6-5.0)
20.6 79.4
6.51 ( 0.21 d (6.2-6.9)
4.83 ( 0.71 c (4.0-6.2)
11.2 ( 1.7 b (9-15)
68 ( 24 a (53-122) 1
7
0.12 gavage
AA
49.8
50.4
0.0025
0.0024
69.3
10.7
4.5 60.9 9.2 25.4
6.39
6.90
11.8
0.9
19.8
7.35
4.09
1c
0.1 iv
23.2 ( 3.9 b (19-29) 132 ( 17 a (118-153) 8.1 ( 1.8 c (6.5-11.4)
0.0118
0.0025
5.96 ( 1.03 a (4.5-7.4) 53.7 ( 6.6 c,d (47-63) 7.73 ( 1.50 b (6.0-10.0) 0.0024
2.6 50.9 9.9 36.6
6.16 ( 0.61 c (5.5-7.3) 6.77 ( 0.67 b (6.1-8.0) 7.35 ( 0.73 c (6.6-8.6) 0.59 ( 0.06 c (0.5-0.7) 3.45 ( 0.30 d (3.0-3.9) 7.23 ( 0.44 c (6.6-7.9) 9.15 ( 0.77 e (8.5-10.4)
56 ( 7 a (48-65) 1
6
0.1 gavage
AA
female rat AA
24.5 ( 2.8 b (21-29) 103 ( 18 b,c (71-116)
0.0025
0.0024
16.7 ( 2.8 b (14-22) 73.9 ( 17.0 b (45-89)
2.2 21.0 32.3 44.5
25.5 ( 2.2 b (22.5-27.5) 7.66 ( 0.66 b (6.7-8.3) 22.2 ( 3.1 a (20.3-27.8) 0.41 ( 0.06 d (0.3-0.5) 6.40 ( 1.11 e (5.2-7.8) 2.08 ( 0.35 d (1.7-2.6) 1.59 ( 0.29 f (1.3-2.0)
524 ( 160 d (382-775) 1
5
0.121 diet
AA
0.0118
0.0025
0.0024
3.0
30.1
5.73
low dose gavage
same as
infusion
1.07 drinking water 1c
a Statistical comparisons were made for each parameter; a different letter designates a statistical difference at the 2p < 0.05 level for each row. Comparisons that were made were male vs female, GA gavage vs AA gavage, GA gavage vs AA diet, and AA gavage vs AA diet. b Single value that was not allowed to vary within species. c Data from a group of rats were combined for a single PBPK/PD analysis. d Refs 12 (AA and GA plasma concentration data) and 14 (urinary excretion data). e Refs 12 (AA and GA plasma concentration data), 14 and 18 (urinary excretion and Hb adduct data), and 17 (liver GA-DNA adduct data). f Refs 11 (AA plasma concentration data) and 14 (urinary excretion data). g IP absorption half-life.
formation of Hb AA adducts formation of Hb GA adducts formation of liver GA adducts decay of Hb AA adductsb decay of Hb GA adductsb decay of liver GA adductsb hemoblobin AA adducts hemoblobin GA adducts liver GA adducts
min-1
min-1
AA-GS to urine
AA AA-GS GA GA-GS
min-1
GA to urine
8.8 ( 1.8 a (6-11)
min-1
AA to urine
10.0
min-1
GA to GA-GS 14.6
13.7
min-1
AA to AA-GS
18.0 ( 3.3 a (13-22)
5.3
min-1
1c
min-1
17.4
87 ( 44 a (39-152) 1
AA 0.1 iv
AA to GA
first-order rate constants
stomach absorption half-life fraction absorbed from stomach
6
1c
GA
n
GA
0.12 gavage
units
0.12 iv
chemical
dose (mg/kg) route
male rat
Table 3. Pharmacokinetic and Pharmacodynamic Parameters from AA and GA Administration to Rats [Mean ( Standard Deviation (Range)]
392 Chem. Res. Toxicol., Vol. 20, No. 3, 2007 Young et al.
Acrylamide PBPK/PD Model
Chem. Res. Toxicol., Vol. 20, No. 3, 2007 393
Table 4. Pharmacokinetic and Pharmacodynamic Parameters from AA and GA Administration to Micea male mouse chemical
units
GA
GA
AA
AA
AA
dose (mg/kg) route
0.12 iv
0.12 0.1 gavage iv
0.1 0.16 gavage diet
n
mean
mean
mean
stomach absorption min half-life fraction absorbed from stomach intestinal absorption min half-life fraction absorbed from intestines AA to GA
min-1
AA to AA-GS GA to GA-GS AA to urine GA to urine AA-GS to urine GA-GS to urine
min-1
min-1 min-1 min-1 min-1 min-1
AA AA-GS GA GA-GS
mean
AA
50 2.41 gavage drinking water mean mean
GA
mean
mean
mean
mean
10
1
82
28
1
0.73
0.62
1
0.74
0.33
1
275
321
333
224
0.27
0.38
0.26
0.67
12.3
23.3
13.7
10.6
4.66 6.90 0.45 4.34 0.87 0.57
4.80 7.04 0 4.47 1.51 0.66
7.42 6.25
0.30
3.69 19.1 0.22 7.19 1.26 0.65
70.2 29.8
2.4 23.8 41.9 32.0
2.4 13.8 59.6 24.2
0 25.7 56.0 18.3
14.6
20.3
10.4
11.0
88.6
84.3
55.4
45.0
9.0
20.1 4.02 10.0 0.2 2.2 1.0 0.7
1
first-order rate constants 30.0 11.5 same as low dose gavage 13.8 8.04 10.8 8.5 14.1 0 7.5 3.8 3.02 3.5 10 2.2 11 0.66
0.50
52.1 47.9
5.7 54.5 23.7 16.1
elimination pattern (% of excretion) 1.6 0 16.3 32.0 41.3 42.8 42.3 25.9 56.2 39.2 25.7 32.8 43.8
13.8
pharmacodynamic rate constants 13.3 18.8 3.2
61.7
66.2
pmol/g globin pmol/g 143 globin 8 in 10 nucleotides
mean
13
0.95
hemoblobin AA adducts hemoblobin GA adducts liver GA adducts
AA
pharmacokinetic rate parameters 10 80 24 infusion
2.35
min-1
AAb
AA
50 2.96 gavage drinking water mean mean
4.22
51.5
AA
0.1 0.28 gavage diet
12.6 7.0 0.8 3.1 1.6 1.2
min-1
AA
0.12 0.1 gavage iv
16.7
formation of Hb AA adducts formation of Hb GA adducts formation of liver GA adducts decay of Hb AA adducts decay of Hb GA adducts decay of liver GA adducts
GA
0.12 iv
13.6
60.3 39.7
min-1
mean
female mouse AAb
50.1 6.10
84.5
4.34
20.8 2.73
58.2
1.49
12.2 2.60
48.8 5.04
4.05
infusion
same as low dose gavage
3.41 6 12
41.4 25.7 33.0
1.81
2.02
min-1
0.0033 0.0033 0.0033
0.0033
0.0038 0.0038 0.0038
0.0038
min-1
0.0033 0.0033 0.0033 0.0033 0.0033
0.0033
0.0034 0.0034 0.0034 0.0034 0.0034
0.0034
min-1
0.0110
146 6.7
0.0110
19.5
11.9
83
136 4.6
0.0110 0.0110
0.0110
biomarkers: number of adducts 13.4 1400 124
19350 1271
259
158
149 6.7
0.0110
0.0110 0.0110
23.1
12.0
18.5
2570
75.3
131
135
24730
4.6
1153
238
a Each route yielded only one set of data with three mice sacrificed at each time point, and all mean data combined for a single simulation and set of pharmacokinetic and pharmacodynamic parameters. b This pilot study data set only had one mouse per gender for each time point and utilized the urinary excretion data from ref 14.
diet (2p < 0.025) routes of administration. However, the female rat absorbed the gavage dose nearly four times faster than the male but the absorption from diet was nearly two times slower. These long absorption half-life values assigned to the stomach absorption do not reflect maintenance of the AA in the stomach for extended periods of time but rather reflect that the model utilized a single component for absorption from the gastrointestinal tract. The male and female rats are significantly different (2p < 0.001) for all metabolic first-order rate parameter comparisons except for the AA to GA and GA to GA-GS conversion for the AA diet administration (Table 3). The male rat metabolism is faster than the female; that is, the rate constants are larger. For the male, the AA to GA metabolism step is statistically different when comparing across routes of administration (2p < 0.005). However, for the female, only the AA to AA-GS metabolism step is not statistically different when comparing across routes
of administration. Despite these apparent differences, the range of values across gender, dose, and route of administration only varies four- (AA to AA-GS) to eight (AA to GA)-fold. All of the urinary elimination parameters vary significantly across gender and routes of administration. As urinary elimination is usually thought to be independent of route of administration, these differences may be a skewed result of using a single urinary excretion pattern for each route and gender with individual animal serum data with a result of rather small standard deviations and biased statistical significance. For example, the rate constants for AA to urine vary from means of 0.41-1.6 with a range of individual values of about a factor of 8; yet, every mean value is statistically different from the other. The means of the other three urinary parameters vary from three- to nine-fold. The largest of these, GA-GS to urine, has six means that vary from 1.6 to 14 with each being statistically different from the other five.
394 Chem. Res. Toxicol., Vol. 20, No. 3, 2007
Figure 8. Pharmacokinetic and pharmacodynamic simulation of data following IP administration of 50 mg/kg AA to male rats. Plasma concentration data were taken from Barber et al. (12), urinary excretion curves and Hb adduct levels came from Sumner et al. (18), and the liver GA-DNA adduct level came from Doerge et al. (17). Units and symbols are the same as in Figures 3 and 7.
The high dose oral and IP administration data taken from the literature (11, 12, 14, 17, 18) yield a slightly different set of metabolic and elimination parameters when compared to the oral low dose male rat studies (Table 3). The percentage conversion of AA to GA (i.e., rate constants AA f GA divided by the sum of the rate constants for AA f GA and AA f AA-GS) is decreased from about 57 to 32% as the dose increases. Figure 8 depicts the simulation of the 50 mg/kg AA IP combined data set. The elimination pattern (see middle rows of Table 3) for the IP high dose reflects the low dose iv much more closely than the low dose oral values; this route effect was predicted by previous toxicokinetic work including Barber et al. (12) and Doerge et al. (15, 16). The mouse data only yield a single pharmacokinetic value for each parameter, route, and gender (Table 4) since each mouse provided only a single serum sample. Data were averaged at each sacrifice time, and then, all sacrifice times for a given route, dose, and gender were combined to form a single serum concentration-time curve (see Figures 5 and 6). There is little or no difference in absorption, metabolism, or elimination between the male and the female mouse or route of low dose AA administration. The AA absorption for the mouse is different from the rat in that a portion of the absorption occurs in the intestines or a second site for absorption. As the AA gavage dose increases from 0.1 to 50 mg/kg, the percentage metabolism to GA decreases from 83 to 59%. That is, at lower doses for the mouse and the rat, the conversion to GA is more efficient. This dose effect was predicted by previous toxicokinetic evaluations including Bergmark et al. (22) and Doerge et al. (15, 16). GA absorption is faster from mice than rats. Low dose AA absorption is probably similar in time between the mouse and the rat, but an accurate comparison is difficult because a fraction of the dose was absorbed from the intestines in mice. All of the mouse metabolic parameters fall within the range of the rat values. The percentage metabolism to GA for the low dose oral routes in the rat is about 57% (as compared to the mouse at 83%), which indicates that the rat is less efficient at forming GA than the mouse; this species effect was noted by previous toxicokinetic studies including Doerge et al. (15-17). Michaelis-Menten kinetics for the metabolism of AA to GA were investigated for the high and low oral AA doses in the mouse. The pharmacokinetic and pharmacodynamic values are
Young et al.
listed in Table 5. Except for the AA to GA metabolism, all other parameters are essentially the same as reported in Table 4 for the mouse. The approximately three-fold range of Vmax is essentially the same as for the first-order rate constant. Thus, the use of Michaelis-Menten kinetics for the formation of GA offers no advantage. The Km values derived from the microsomal incubations (23) and used for these mouse simulations (9.74 mM) are very much higher than the peak serum concentrations of AA produced by either low dose (0.1 mg/kg and
