In vivo Role of Cytochrome P450 2E1 and Glutathione-S-Transferase Activity for Acrylamide Toxicokinetics in Humans

Acrylamide, an industrial and scientific chemical, is present in high concentrations in some fried and baked starch-enriched food (1, 2). The substance is carcinogenic in rodents (2-4) and considered as a probable carcinogen in humans (5). Dietary exposure to acrylamide affects all nutritional patterns with different exposure levels across age categories, countries, and composition of diet. Based on food contents, the estimated acrylamide intake is around 22 to 100 μg/d (1, 6, 7).

Toxicokinetics of acrylamide were investigated mainly in rodents, whereas recently, some data on acrylamide disposition in healthy volunteers after acrylamide administration in a test-meal (8, 9), in drinking water (10, 11), or dermally (12) have been published. As shown in rats and mice (6, 10, 13) and in humans (8-10), acrylamide is readily absorbed and widely distributed to tissues, crosses the human placenta, and is transferred into the breast milk (8, 14). Acrylamide is in part biotransformed to the epoxide glycidamide. The extent of glycidamide formation is dose and route dependent and it is highest for low-dose dietary acrylamide exposure (15, 16). Glycidamide is considered to be more genotoxic and carcinogenic than acrylamide itself (3, 16-18). In mice, the conversion of acrylamide to glycidamide is mediated by cytochrome P450 2E1 (CYP2E1; refs. 19, 20), whereas respective human data in vivo are lacking. Whereas the low molecular weight and the high hydrophilicity and reactivity of glycidamide make in vitro studies on this metabolic step difficult, a recent report showed that in expressed human cytochrome P450s, in liver microsomes, and in CYP expressing cell lines CYP2E1 is involved in glycidamide formation (21).

Acrylamide and glycidamide are conjugated to glutathione, which is the main pathway of acrylamide metabolism in both rodents and humans (10-12, 22-24). As products of this conjugation, the following mercapturic acid metabolites have been observed: N-acetyl-S-(2-carbamoylethyl)-cysteine (AAMA); N-acetyl-S-(2-carbamoylethyl)-cysteine-S-oxide; N-(R,S)-acetyl-S-(2-hydroxy-2-carbamoylethyl)-cysteine (GAMA); and N-acetyl-S-(1-carbamoyl-2-hydroxyethyl)-cysteine. Of these, AAMA, its sulfoxide, and GAMA have been unequivocally confirmed in humans (9-11). Whereas there are many glutathione-S-transferases (GST) in humans and rodents, which may be involved in these metabolic steps, nonenzymatic formation of glutathione conjugates is also possible and has been reported for acrylamide (25, 26). The role of GSTs in acrylamide metabolism in humans remains to be assessed.

Acrylamide and glycidamide have the capability to bind covalently to nucleophilic sites of proteins and nucleic acids (10, 20, 27, 28). Carcinogenicity studies in rats indicate that chronic acrylamide exposure increased the incidence of various tumor types (3, 5, 29) from the dose of 0.5 mg/kg/d, which would correspond to a human equivalent dose of 7.6 μg/kg/d (29). Based on rodent studies, the cancer risk for humans through the dietary acrylamide was estimated to be 0.3 × 10⁻³ to 2.4 × 10⁻³ by lifetime acrylamide uptake of 70 μg/d with ∼8-fold range of estimates driven by different cancer potency values and extrapolation models (6, 7, 13). Equivocal results on the relationship between dietary acrylamide and cancer incidence/mortality have been reported in epidemiologic studies (3, 13, 30, 31). A recently conducted case-control study showed a statistically significant correlation between levels of AA adduct to hemoglobin and breast cancer with an estimated incidence rate ratio of 2.7 (95% confidence interval, 1.1-6.6) per 10-fold increase in adduct level (32).

Based on urinary excretion data and levels of acrylamide and glycidamide adducts to hemoglobin, the extent of glycidamide formation and the estimated internal exposure to glycidamide, which is considered to play the main role in acrylamide carcinogenicity (16-18), are about two to four times lower in humans compared with rodents (9, 10). In humans, the urinary concentration ratio of AAMA and GAMA, which is considered as a metric to describe the extent of conversion of acrylamide to glycidamide (9) and thus reflects internal exposure, varies considerably between subjects (22, 23). It is unclear to which extent the activity and expression of metabolizing enzymes influences acrylamide metabolism and, therefore, contributes to differences in acrylamide disposition and carcinogenicity among species and between individuals (20, 33). The objectives of the present study were to clarify the role of CYP2E1 and GSTs and their genetic variants for the in vivo metabolism of acrylamide in humans and, thus, to improve the cancer risk assessment for humans through acrylamide intake with food.

The study was conducted according to the pertinent version of the Declaration of Helsinki and was approved by the Ethics Committee of the University of Cologne. All volunteers provided written informed consent. The study had a single-center, four-period cross-over design. In three of the four study periods, participants ingested a single acrylamide-containing test-meal (potato chips), which was given without comedication (reference period), after CYP2E1 inhibition with disulfiram (inhibition period) or after CYP2E1 induction with ethanol (induction period). Baseline CYP2E1 activity was determined in a fourth period using chlorzoxazone. The washout phase between treatments lasted 2 wk.

Despite the exploratory character of the study, the following considerations were made for selection of the sample size: On the basis of a conservative estimate of intrasubject multiplicative coefficients of variation (CV) of up to 30% for acrylamide metrics, a sample size of N = 16 would allow rejection of the respective null hypothesis “presence of a more than 0.7 or 1.43-fold change in acrylamide metrics by disulfiram or by ethanol intake” with α = 0.05 and a power of at least 80% for all parameters if the true mean changes were not outside the 0.95- to 1.05-fold range (34). To assess the relationship between chlorzoxazone and acrylamid kinetics, a sample size of 16 is sufficient to show an r² of 0.5 with α = 0.05 and a power of at least 80% (indicating that at least 50% in the variation of the respective acrylamide parameter would be related to individual variation in CYP2E1 activity). Thus, eight females and eight male Caucasian nonsmokers were included in the study; age (mean ± SD) was 29.8 ± 5.9 y and the body mass index was 22.9 ± 2.0 kg/m². Participants were required to be healthy as determined by medical history and physical examination including electrocardiogram, vital sign measurements, standard clinical chemistry, and hematologic evaluation. Main exclusion criteria were pregnancy or lactation in females, environmental or occupational exposure to acrylamide, alcohol consumption of >30 g/d, concomitant drug intake, and suspicion of hypersensitivity to chlorzoxazone and disulfiram. The volunteers were instructed to abstain from acrylamide-rich foodstuffs (in particular, potato chips, pan-fried potatoes, biscuits, crackers, and breakfast cereals), alcohol, methylxanthine-containing food and beverages, grapefruit flesh/juice, and any other comedication before and during the study.

The volunteers were hospitalized 11 h before acrylamide or chlorzoxazone administration and until 12 h (chlorzoxazone period) or 24 h (all other periods) thereafter. In each study period, they had to fast from 11 h before dosing (except disulfiram administration) and until 6 h post-dose, whereas fluid intake was standardized from −1 h to +6 h relative to the dose. Daily fluid intake up to 72 h post-dose should not exceed 3 liters. Acrylamide-containing food was prepared by frying batches of potato chips on the day before administration as previously described (9). Acrylamide concentration in the processed potato chips was determined by liquid chromatography/tandem mass spectrometry (LC-MS/MS) and appropriate portions were given to ensure the acrylamide target dose. In the acrylamide periods, each participant ingested potato chips with 240 mL of water. The true acrylamide doses administered reached a median value of 998 μg (range, 957-1,000 μg) or 14.0 μmol corresponding to 14.8 μg/kg body weight.

For CYP2E1 inhibition, volunteers received a single oral dose of 500 mg disulfiram (Antabus 0.5 Dispergetten, Altana Pharma AG) 10 h before acrylamide dosing and after at least 1 h of fasting. For CYP2E1 induction, participants drunk two bottles (0.5 liter) of beer every evening (Reissdorf Kölsch, Privat-Brauerei Heinrich Reissdorf & Co.) containing 48 mL of ethanol/L during the 7 d before acrylamide administration. Beer intake was ambulatory; regularity, completeness, timing, and tolerability of beer intake were assessed based on the subject's diary and checked by study personnel. The CYP2E1 phenotype was investigated by using a single oral dose of 250 mg chlorzoxazone (Chlorzoxazone, AstraZeneca).

Blood samples for phenotyping were collected just before chlorzoxazone dosing and 2 h thereafter using Li⁺-heparinized tubes (Sarstedt). The samples were centrifuged (2,000 × g, +4°C, 10 min), frozen, and stored at −80°C until assayed. Blood sampling for determination of acrylamide and glycidamide adducts to hemoglobin was carried out just before administration and at 24 h post-dose using EDTA-containing tubes (Sarstedt). The samples were initially centrifuged for isolation of erythrocytes (1,992 × g, +4°C, 5 min). For hemoglobin isolation, erythrocytes were washed thrice with 2.0 mL of 0.9% NaCl solution. Hemolysis was achieved by addition of pure water (2.0 mL) and pulse-vortexing (27). The sample was placed in a 15-mL Falcon tube and stored at −20°C or below until analysis. For genotyping, a blood sample from each volunteer was drawn at pre-study examination using EDTA-containing tube (Sarstedt) and stored at −20°C. For determination of urinary acrylamide and acrylamide derived metabolites, samples of urine were collected just before dosing and for the following intervals: 0-2, 2-4, 4-6, 6-8, 8-10, 10-12, 12-16, 16-24, 24-36, 36-48, and 48-72 h post-dose. Urine was kept at +4°C during the collection period and subsequently stored at −80°C until analysis.

The well-being of the study subjects was closely surveyed throughout the study by measuring vital signs and recording of any adverse event.

Acrylamide in chips samples was quantified by LC-MS/MS method (lower limit of quantification, LLQ: 140.7 nmol/L) as previously reported (35) using a Thermo Finnigan LC-MS/MS-System (Thermo Finnigan). Unmetabolized acrylamide in urine was analyzed by LC-MS/MS (LLQ: 7.0 nmol/L) using an Applied Biosystems API3000 LC-MS/MS device as previously described (8). Quantification of AAMA and GAMA in urine was done according to the method previously described (22, 36) with modifications. Briefly, 50 μL of 1 N HCl and 50 μL of labeled internal standard (d3-AAMA) were added to 1,200 μL of urine. The samples were vortexed and centrifuged (14,000 × g, 5 min). The supernatant (1,000 μL) was placed onto an SPE column [Isolute ENV+, 50 mg, 10 mL XL (Separtis)] preconditioned with 4 mL methanol, 2 mL formic acid (pH 2.0), and 2 mL water. Afterwards, the columns were washed with 1,200 μL HCl (pH 2.5). The analytes were eluted with 1,200 μL of 2% (v/v) formic acid in methanol and evaporated to dryness. The pellets were redissolved in 100 μL of 0.1% formic acid/10% methanol (v/v), and 20 μL were injected into the LC-MS/MS system (Thermo Finnigan). Chromatography was done in 6 min using a 50 × 2.1 mm, 5-μm Hypercarb column (Thermo Hypersil-Keystone) with a 10 × 4.0 mm, 5-μm Hypercarb pre-column and a linear gradient of methanol and formic acid (0.1%). AAMA and GAMA were detected by negative electrospray ionization in the selected reaction monitoring mode with ion transitions (m/z) 233.0→104.0 and 249.0→120.0 for AAMA and GAMA, respectively. The LLQs were 21.3 and 20.0 nmol/L for AAMA and GAMA, respectively. Calibration curves were linear between 20 and 20,000 nmol/L. Intraday and interday precision values did not exceed 11.3% and 10.3% for AAMA and GAMA, respectively; intraday and interday accuracy ranged from −3.7 to +6.8% for AAMA and from −0.7 to 6.8% for GAMA.

The analysis of acrylamide (AAVal) and glycidamide (GAVal) adducts to the NH₂-terminal valine of hemoglobin was carried out by gas chromatography-MS as previously described (37-41). The analysis was done using an Agilent Technologies 6890 N Network GC System [column, DB-XLB, 30 m × 0.25 mm ID, 0.25 μm film thickness (J&W Scientific); helium flow, 1.3 mL/min; injector, PTV injector used in pulsed splitless mode at 250°C; injection volume, 1 μL; temperature program: 90°C for 1 min, 25°C/min to 210°C, 10°C/min to 240°C, 25°C/min to 310°C for 5 min] coupled to 5973 inert Mass Selective Detector device (Agilent Technologies) operating in negative chemical ionization mode (reactant gas, methane (45%); ion source temperature, 150°C; quadrupole temperature, 150°C; transfer line temperature, 280°C). Hb adducts were determined as pentafluorophenylthiohydantoin (PFPTH) derivatives subsequent to the modified Edman degradation procedure (27, 38, 39). As internal standards for AAVal and GAVal, deuterium-labeled d₇-AAVal-PFPTH and d₇-acGAVal-PFPTH (quantification is based on the acetonized derivative d₇-acGAVal-PFPTH; ref. 39) were used. Standards were synthesized at the Department of Food Chemistry and Environmental Toxicology, University of Kaiserslautern, Germany as described (37, 42). The LLQs were 0.9 and 2.3 pmol/g Hb for AAVal and GAVal, respectively. Intraday and interday precision values were below 9.6% and 12.8% for AAVal and GAVal, respectively; intraday and interday accuracy ranged from 2.0% to 8.7% for AAVal and from 7.2% to 14.6% for GAVal. In the calibration range of 0.25 to 250 nmol/L, linearity was characterized by r² = 0.9981 for d₇-AAVal-PFPTH and r² = 0.9996 for d₇-acGAVal-PFPTH.

Chlorzoxazone and 6-OH-chlorzoxazone in plasma for CYP2E1 phenotyping were determined by a high-performance liquid chromatography method (43) with modifications. The high-performance liquid chromatography system consisted of a Waters 2690 Separations Module with a PDA Waters 996 photodiode array detector. Internal standard [3-isobutyl-1-methylxanthine (IBMX), 50 μL, 200 μg/mL] was added to 500 μL of plasma; plasma proteins were precipitated; samples were centrifugated (4°C, 16,000 × g, 20 min); and the supernatant was evaporated to dryness. Thereafter, the pellets were redissolved in 100 μL of mobile phase [acetonitrile/acetate buffer (20:80, v/v, 0.05 mol/L, pH 4.25)], and 30 μL were injected onto a 250 × 4.6 mm, 5-μm, Zorbax RX-C8 column (Duratec Analysentechnik) with a Zorbax RX-C8 12.5 × 4.6 mm pre-column. Chlorzoxazone, 6-OH-chlorzoxazone, and IBMX were detected at 279, 296, and 272 nm, respectively. Calibration curves were linear over a concentration range of 3.2 to 235.9 mmol/mL and 2.6 to 215.6 mmol/mL for chlorzoxazone and 6-OH-chlorzoxazone, respectively, and corresponding LLQs were 3.2 and 2.6 mmol/mL. The interday and intraday accuracy ranged from −7.8% to +6.1% for chlorzoxazone and from −1.4% to +2.7% for 6-OH-chlorzoxazone. Precision was below 5.7% and 1.9% for chlorzoxazone and 6-OH-chlorzoxazone, respectively.

Genetic analysis included the investigation of CYP2E1*5A (PstI+, RsaI−, DraI−), CYP2E1*5B (PstI+, RsaI−), and CYP2E1*6 (DraI−) alleles; null alleles of GSTM1 (GSTM1*0) and GSTT1 (GSTT1*0) genes; and GSTP1 313A>G and GSTP1 341T>C single nucleotide polymorphisms (SNP).

DNA was isolated from EDTA anticoagulated blood using the QIAamp DNA Blood Mini Kit (Qiagen) following the manufacturer's instructions. A 300-μL aliquot of a whole blood sample yielded, on average, 5 μg of DNA in 100 μL eluting buffer (50 ng/μL), which was in accordance with the value indicated by the manufacturer. All PCR amplifications were run in a MultiCycler programmable thermal cycler-200 (MJ Research Laboratories).

CYP2E1 genotyping was carried out with the PCR-RFLP method as previously reported (44) with modifications. The PCR reaction mixture (50-μL) consisted of genomic DNA (200 ng), PCR buffer (10×), MgCl₂ (25 mmol/L), deoxynucleotide triphosphates (10 mmol/L), each primer (50 μmol/L), GoTaq DNA-polymerase (2.5 units; Qiagen), and water for molecular biology analysis. For investigation of the PstI and RsaI restriction sites with a complete linkage disequilibrium (45), primer pair 5′-CCAGTCGAGTCTACATTGTCA-3′ and 5′-TTCATTCTGTCTTCTAACTGG-3′was used, which produced a 410-bp fragment. For the Dral restriction site, primer pair 5′-TCGTCAGTTCCTGAAAGCAGG-3′ and 5′-GAGCTCTGATGCAAGTATCGCA-3′ was used, which yielded a 995-bp DNA fragment. An initial denaturation (94°C, 1 min) was followed by 31 cycles of denaturation (94°C, 30 s), annealing (55°C, 1 min), and elongation (72°C, 1 min), with a final elongation at 72°C for 10 min. PCR products were separated electrophoretically on a 1.6% agarose gel (90 mV, 60 min). Each PCR product (17.5 μL) was digested with 10 units of restriction enzymes RsaI and PstI for 410-bp fragments and DraI for 996-bp fragments, respectively, in a Tango buffer (5×) at 37°C for 60 min. The incubation products were separated electrophoretically on a 1.6% agarose gel (90 mV, 60 min). The presence of the PstI restriction site yielded two fragments (290 and 120 bp) of the 410-bp PCR product. In contrast, the presence of the RsaI restriction site yielded two fragments (60 and 350 bp) with the 410-bp PCR product, whereas in this case it was not digested by PstI. The 995-bp PCR product with the polymorphic Dral restriction site and the wild-type allele was digested by DraI into two (874 and 121 bp) and three (572, 302, and 121 bp) fragments, respectively.

Genotyping for GSTs was done using direct sequencing methods (46, 47) with modifications. The PCR reaction mixture (50 μL) consisted of genomic DNA (100-200 ng), PCR buffer (10×), MgCl₂ (25 mmol/L), deoxynucleotide triphosphates (10 mmol/L), each primer (50 μmol/L; Table 1), GoTaq DNA-polymerase (2.5 units; Qiagen), and pure water. After amplification (Table 1), the PCR products were separated electrophoretically (90 mV, 60 min) on a 1.6% agarose gel. The allele-specific PCR method used for GSTM1 and GSTT1 genotyping allowed the identification of the active *A-allele(s) without discrimination between homozygous and heterozygous alleles. In such a case, the 650-bp (GSTM1) or 480-bp (GSTT1) DNA fragments were amplified by PCR. In the absence of active *A-allele(s) of these genes (null type, *0/*0), no DNA fragments were amplified. For GSTP1 genotyping, the PCR products were purified using Promega Wizard SV Gel and PCR Clean-Up System (Promega) according to the manufacturer's instructions. DNA sequencing was carried out using BigDye Terminator kit V.1.1 (Applied Biosystems). The reaction mixture (10 μL) for sequencing PCR consisted of purified PCR products (6 μL, 50 ng of DNA), sequencing primers [3.2 μmol/L; GSTP1 313A>G (ACCCCAGGGCTCTATGGGAA) and GSTP1 341T>C (GGCCAGATGCTCACCTGGTC)], and BigDye Terminator Mix (2 μL). PCR included 29 cycles, each consisting of denaturation (96°C, 30 s), annealing (50°C, 30 s), and elongation (60°C, 4 min), followed by final elongation (62°C, 10 min). The sequencing products were purified with Sephadex G50 (Sigma) on MultiScreen HV plates (MAHVN030, Millipore) according to the manufacturer's instructions. Capillary electrophoresis was carried out using an ABI Prism 3100 Genetic Analyzer sequencer (Applied Biosystems). Pairwise sequence comparison with reference sequences was done with MT Navigator PPC and Edit View 1.0.1 Software (Applied Biosystems).

For epoxidation of acrylamide to form glycidamide, a catalytical system [1,4,7-trimethyl-1,4,7-triazacyclononan and sodium ascorbate (Mn-TMTACN)], developed by Berkessel et al. (48, 49), was used. Acrylamide (2.84 g; 40.0 mmoL, 1 eq) was dissolved in 25.00 mL of acetonitrile and cooled down to 0°C. Afterwards, 20.00 mL of a 35% H₂O₂ solution were added to the mixture. Within 3 h, while stirring and cooling with ice, a mixture of 5.00 mL of a 40.00 mmol/L trimethyltriazacyclononane solution in acetonitrile, 2.50 mL of a 80.00 mmol/L aqueous sodium ascorbate solution, and 0.40 mL of a 160 mmol/L aqueous Mn (II)-acetate solution was added dropwise to the acrylamide solution. The clear solution turned brown. The solution was stirred overnight while warming from 0°C to ambient temperature. MnO₂ was added to the solution until gas production receded. The aqueous phase was filtered, extracted thrice with 20.00 mL ethyl acetate, the organic phases dried with MgSO₄, and the solvent in vacuo removed [glycidamide: melting point, 40°C; ¹H nuclear magnetic resonance (300 MHz, DMSO-d₆): δ = 2.74 (dd, J = 6, 2.4 Hz; 1 H, H-C3), 2.86 (dd, J = 6, 4.2 Hz; 1 H, H′-C3), 3.27 (dd, J = 4.2, 2.4, 1 H, H-C2), 7.36 (d, J = 41.7, 2 H, NH₂); ¹³C nuclear magnetic resonance (75 MHz, DMSO-d₆): δ = 45.44 (t, C-3), 48.26 (d, C-2), 169.98 (s, C-1); EA: C₃H₅NO₂: calculated: C, 41.38; H, 5.79; N, 16.09, O, 36.75; found: C, 41.19, H, 5.81, N, 15.37]. The yield of glycidamide was 43% and no impurities were present in the ¹H nuclear magnetic resonance spectrum. The subsequent synthesis of GAMA, AAMA, and d₃-AAMA was done as described (24).

Noncompartmental urinary toxicokinetic parameters of acrylamide and its metabolites were evaluated as described elsewhere (9). The molar plasma concentration ratio 6-OH-chlorzoxazone/chlorzoxazone at 2 h post-dose was applied as phenotyping metric for CYP2E1 (50, 51). Testing for an effect of disulfiram or ethanol on acrylamide toxicokinetics was handled as an equivalence problem; thus, point estimates and 90% confidence intervals were calculated for the ratios of toxicokinetic parameters between periods (52). An effect was considered to be statistically significant if the 90% confidence interval of the estimated true ratio μ_test/μ_reference did not include unity. Changes of hemoglobin adducts were assessed by a t test for paired data applied on log values. Linear regression analysis was used to investigate the relationship between CYP2E1 phenotype and acrylamide toxicokinetics and that between blood and urinary toxicokinetic parameters. Tests for differences between CYP2E1, GSTM1, T1, and P1 genotypes with regard to acrylamide toxicokinetics were carried out using the Kruskal-Wallis-test. Statistical analysis was done using SPSS for Windows version 15.0. Significance for the latter tests was assumed at P < 0.05.

Because of the ubiquitous exposure to acrylamide, in five of the pre-dose samples quantifiable amounts of acrylamide and/or its metabolites were found. In all study periods, following acrylamide intake, maximal urinary excretion rates during the 2 to 4 hours and 6 to 10 hours of collection period were observed for unchanged acrylamide and its metabolites (AAMA and GAMA), respectively (Table 2). One volunteer (#10) had unusual concentration-time profiles for GAMA in the reference and the inhibition periods; low GAMA excretion rates up to 12 to 16 hours post-dose with subsequent increasing until the end of urine collection period (72 hours post-dose) were observed. Because the excretion profiles of unchanged acrylamide and AAMA did not show any abnormalities in this volunteer and were similar to those in other study subjects, there was no clear evidence for an additional exposure to acrylamide after intake of the test meal. The reason for this discrepancy remains unexplained. Because a reasonable assessment of elimination half-life was not possible in this case, the data from this subject were excluded from parts of the calculation of some GAMA toxicokinetics and statistical evaluation (Table 2).

Compared with the reference period, disulfiram coadministration resulted in clear and significant changes in acrylamide toxicokinetics (Table 2; Fig. 1). Average cumulative acrylamide excretion extrapolated to infinity increased 1.34-fold and AAMA excretion increased 1.15-fold, whereas GAMA excretion was reduced to 0.44-fold of the respective values in the reference period. In contrast, CYP2E1 induction with ethanol did not result in significant changes in toxicokinetics (Table 2, Fig. 1). The ratios for the CYP2E1 induction period over the reference period were 1.00, 0.99, and 1.04 for acrylamide, AAMA, and GAMA, respectively. Likewise, the molar ratio of GAMA/AAMA for cumulative acrylamide excretion dropped significantly in the inhibition period but was similar in the reference and in the induction periods (Table 2). As the sum of unchanged acrylamide, AAMA, and GAMA, the cumulative urinary recovery reached 71% (CV 23%), 81% (CV 41%), and 70% (CV 31%) of the acrylamide dose. Thus, the changes in the CYP2E1 inhibition period resulted in clear increase in the recovery of the acrylamide dose.

At 24 hours after intake of the acrylamide-rich meal, increased mean concentrations of both AAVal and GAVal were observed in all study periods (Table 3). This increase was significant in all periods for AAVal when evaluated separately and was significant for GAVal only when evaluating all periods together. Compared with the reference period, CYP2E1 inhibition significantly increased the formation of AAVal to the 1.08-fold, whereas CYP2E1 induction had no effect (Table 3). No clear effects of CYP2E1 inhibition or induction was observed in the +24 h/pre-dose GAVal ratios.

The frequencies of CYP2E1 and GST genotypes detected by study subjects are shown in Table 4. The molar ratio of 6-OH-chlorzoxazone/chlorzoxazone in plasma used as a CYP2E1 phenotyping metric was (geometric mean and CV%) 0.33 (49%). This parameter showed no significant differences between CYP2E1 genetic variants. In addition, CYP2E1 variants had no clear effect on any toxicokinetic parameter of acrylamide, and baseline CYP2E1 activity had no significant correlation to toxicokinetic parameters of acrylamide (all P > 0.05). Furthermore, there was no obvious difference in toxicokinetics of acrylamide between the investigated GSTs genotypes (Fig. 2), and even the absence of alleles coding for active GSTM1 or GSTT1 variants had no effect. Comparison of all the toxicokinetic parameters between genotypes occasionally gave significant differences for individual study periods but with no consistent pattern, suggesting that this is a chance finding.

The acrylamide-containing meal and all study medications were well tolerated; there were no adverse events with relevant effect on the well-being of study participants or affecting study conduction.

In the present study, the toxicokinetic parameters of acrylamide were investigated following administration of a test-meal containing 1 mg acrylamide to healthy volunteers, and the role of CYP2E1 activity and GSTs and their genetic variants for the in vivo metabolism of acrylamide was evaluated. We found that acrylamide intake increases concentrations of acrylamide and glycidamide hemoglobin adducts and that cotreatment with disulfiram resulting in CYP2E1 blockade was related to a shift in acrylamide metabolism from glycidamide formation to increased formation of acrylamide mercapturic acid and acrylamide adducts to hemoglobin.

With respect to the metabolic pattern of acrylamide and total dose recovery in the reference period, the toxicokinetic parameters were in accordance to the data from other human studies with acrylamide dosing (9-12). For the disulfiram period, almost complete inhibition of CYP2E1 by the drug is expected. After a single oral dose of 500 mg disulfiram, CYP2E1 activity inhibitions of ≥90% at the time point of acrylamide administration and ∼50% at 72 hours post-dose have been described (53). Disulfiram coadministration resulted in an increase of the cumulative amounts of unchanged acrylamide and AAMA and a decrease of glycidamide excreted in urine, accompanied by an enhanced formation of AAVal. These results indicate that disulfiram inhibits the oxidative metabolism of acrylamide and, therefore, provide some in vivo evidence for the involvement of CYP2E1 in the conversion of acrylamide to glycidamide in humans, which is in concordance with recent in vitro data (21). Considering the quantitative changes in excretion of acrylamide and both metabolites by CYP2E1 inhibition, this enzyme may account for approximately one fourth of primary acrylamide metabolism. These data also show clearly that other enzyme(s) make a substantial contribution to glycidamide formation in humans. A caveat, however, is to be considered with regard to disulfiram selectivity, which is not only a selective inhibitor within the group of cytochrome P450 enzymes (54) but also inhibits (directly or via metabolites) aldehyde dehydrogenases (55), P-glycoprotein (56), and dopamine β-hydroxylase (57), as well as proteasome activity (58). Whereas the chemical nature of glycidamide as an oxidation product of acrylamide favors a role of CYP2E1 versus the other disulfiram targets, it cannot be excluded that inhibition of enzymes other than CYP2E1 may have contributed to the disulfiram effect on acrylamide toxicokinetics. Irrespective of the enzymes involved, it is currently under discussion whether disulfiram would be useful to diminish the formation of the more toxic glycidamide by subjects occupationally exposed to acrylamide (53); however, because disulfiram increases acrylamide adduct formation, it is doubtful whether this approach provides any benefit.

Interestingly, the overall recovery of an acrylamide dose when coadministered with disulfiram increased clearly. This prompts the speculation that additional undetected, probably small-molecule metabolites beyond glycidamide (including secondary glycidamide metabolites such as glycidamide diol) are formed (10, 23) by the enzymes inhibited by disulfiram (assumed to be CYP2E1). It is also possible that in the absence of disulfiram, a major fraction of acrylamide is metabolized to glycidamide, which then is bound rapidly to macromolecules and therefore is not available for excretion as GAMA. This is, however, not reflected in hemoglobin adducts but may occur in the tissues where glycidamide is formed.

To investigate the toxicokinetics of acrylamide on CYP2E1 induction as an independent approach to change CYP2E1 activity, ethanol was administered in the current study. Following ethanol consumption of 48 g/d for 1 week, an increase in CYP2E1 activity of ∼30% was expected in healthy subjects (59). We did not decide for a higher dose for tolerability reasons because we wanted neither to risk a potentially excessive formation of glycidamide nor to interfere too much with daily activities of the healthy volunteers. After 3 days of alcohol abstinence, the CYP2E1 activity would be assumed to return to baseline (59). We did not find any significant changes in toxicokinetics of either unchanged acrylamide or mercapturic acid metabolites and in formation of hemoglobin adducts on chronic exposure to ethanol. Assuming a predominant role of CYP2E1 in acrylamide oxidation and an ∼30% increased in enzyme activity under conditions used, expected changes in acrylamide toxicokinetics would be only one third of those observed for CYP2E1 inhibition by disulfiram. The lack of differences in acrylamide toxicokinetics on CYP2E1 induction of this magnitude, therefore, does not exclude a role of CYP2E1 in acrylamide metabolism because the intraindividual variability observed is too high for detection of small differences.

Baseline hemoglobin adduct concentrations were similar to those reported (27, 60-62). At 24 hours after intake of a single 1-mg acrylamide dose, an increased formation of both acrylamide and glycidamide adducts to hemoglobin was observed. This suggests that vice versa dietary interventions are helpful to reduce adduct formation, whereas a beneficial effect of a modulation of CYP2E1 activity is less clear. There were changes both in the GAMA/AAMA ratio of acrylamide excretion excreted in urine and in the ratio of GA-/AA-Val adducts to hemoglobin during disulfiram coadministration. Both parameters are proposed biomarkers of internal acrylamide exposure in humans (22, 27, 36), but it remains uncertain how closely they are linked to formation of tissue DNA adducts. In rodent studies, a good linear correlation was reported between glycidamide adducts to hemoglobin and liver glycidamide DNA adducts (16) and between acrylamide adducts to hemoglobin and acrylamide sperm DNA adducts (63). However, although the epoxide glycidamide is considered as the ultimate genotoxic agent, there may also be a direct risk caused by acrylamide, which may be as important as that conferred by glycidamide because glycidamide formation is not the main metabolic pathway of acrylamide. Thus, the use of ratios of GA- over AA-derived products probably are not useful biomarkers to predict acrylamide cytotoxicity and resulting health risk in humans for all situations.

For CYP2E1 phenotyping, chlorzoxazone was used in the present study and the plasma ratio of 6-OH-chlorzoxazone/chlorzoxazone at 2 hours post-dose served as a metric for enzyme activity. Good tolerability of the chlorzoxazone dose administered as well as validity of the applied phenotyping metric has been shown previously (50, 51). The plasma concentration ratio of 6-OH-chlorzoxazone/chlorzoxazone at 2 h post-dose was similar to that obtained in other phenotyping studies with similar design (51), and an apparently normal baseline CYP2E1 activity was present in all study subjects. Controversial data exist with respect to the CYP2E1 genotype-phenotype relationship. Haufroid and coworkers (64) observed in healthy Caucasians (n = 31) a trend to lower 6-OH-chlorzoxazone/chlorzoxazone plasma ratios in subjects with CYP2E1*6 allele(s) compared with wild-type subjects. In contrast, CYP2E1*5A, *5B, *6 alleles detected in 70 White Americans (65) as well as in 20 Caucasians and 20 Japanese (66) were not associated with differences in chlorzoxazone clearance. These conflicting results may arise due to the limited number of subjects with the rare genotypes in all studies, including the present investigation. The present study was not specifically designed to detect such differences for acrylamide pharmacokinetics but confirms that the presence of a single CYP2E1*6 allele has no obvious effect.

With respect to the frequencies of CYP2E1, GSTM1, GSTT1, and GSTP1 genotype distribution, the study population did not show an obvious deviation from the general Caucasian population (47, 67-74). Homozygous deletion of the entire GSTM1 and/or GSTT1 gene as well as a major single nucleotide polymorphism in GSTP1 did not influence the toxicokinetics of acrylamide (Fig. 2), showing that these individual enzymes play no exclusive role in the acrylamide and glycidamide conjugation with glutathione. Accordingly, an in vitro study showed no influence of GSTM1 or GSTT1 on hemoglobin adduct formation after exposure of human blood samples to acrylamide and glycidamide (75). However, recently the up-regulation of genes encoding GSTs was shown by exposure of Caenorhabditis elegans to 500 mg/L acrylamide, and proteomic analysis confirmed some involvement of GSTs (26). Negative results from our study on the influence of GSTs on acrylamide metabolism support the hypothesis of a primarily nonenzymatic conjugation of acrylamide and glycidamide with glutathione (75). In addition, this metabolic step could be mediated by other GSTs not investigated here.

In conclusion, the increased amounts of unchanged acrylamide and AAMA excreted in urine together with decreased acrylamide excretion of GAMA observed during CYP2E1 blockade by disulfiram supports the nonexclusive involvement of CYP2E1 in glycidamide formation in humans. Based on quantitative changes in acrylamide toxicokinetics on CYP2E1 inhibition, CYP2E1 probably accounts for one fourth of primary acrylamide metabolism in humans. Individual CYP2E1 activity, induction of CYP2E1, and polymorphisms of CYP2E1 had no clear effect on acrylamide toxicokinetics, most probably because the corresponding differences in enzyme activity are not large enough. GSTP1, GSTM1, and GSTT1 enzymes play no major role in the acrylamide and glycidamide conjugation with glutathione. No obvious genetic risks or protective factors with regard to acrylamide-metabolizing enzymes could be determined for exposed subjects.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.