human experimental exposure study on the uptake and urinary elimination of n ...
TRANSCRIPT
Arch Toxicol (2007) 81:335–346
DOI 10.1007/s00204-006-0161-6TOXICOKINETICS AND METABOLISM
Human experimental exposure study on the uptake and urinary elimination of N-methyl-2-pyrrolidone (NMP) during simulated workplace conditions
Michael Bader · Renate Wrbitzky · Meinolf Blaszkewicz · Christoph van Thriel
Received: 18 September 2006 / Accepted: 10 October 2006 / Published online: 14 November 2006© Springer-Verlag 2006
Abstract A human experimental study was carriedout with 16 volunteers to examine the elimination ofN-methyl-2-pyrrolidone (NMP) after exposure to thesolvent under simulated workplace conditions. TheNMP concentrations were 10, 40 and 80 mg/m3 for 2 £4 h with an exposure-free interval of 30 min. Addition-ally, a peak exposure scenario (25 mg/m3 baseline,160 mg/m3 peaks for 4 £ 15 min, time-weighted aver-age: 72 mg/m3) was tested. The inXuence of physicalactivity on the uptake and elimination of NMP wasstudied under otherwise identical exposure conditionsbut involving moderate workload on a bicycle ergome-ter (75 W for 6 £ 10 min). The peak times and biologi-cal half-lives of urinary NMP and its main metabolites5-hydroxy-N-methyl-2-pyrrolidone (5-HNMP) and2-hydroxy-N-methylsuccinimide (2-HMSI) in urinewere analysed as well as the interrelationshipsbetween exposure and biomarkers. All analytesshowed a close correlation between their post-shiftpeak concentrations and airborne NMP. An exposureto the current German workplace limit value of80 mg/m3 under resting conditions resulted in urinarypeak concentrations of 2,400 �g/L NMP, 117 mg/gcreatinine 5-HNMP and 32 mg/g creatinine 2-HMSI
(workload conditions: 3,400 �g/L NMP, 150 mg/g cre-atinine 5-HNMP, 44 mg/g creatinine 2-HMSI). Moder-ate workload enhanced the total uptake of NMP byapproximately one third. DiVerences between the esti-mated and the observed total amount of urinarymetabolites point to a signiWcant contribution of der-mal absorption on the uptake of NMP. This aspect,together with the inXuence of physical workload,should be considered for the evaluation of a biologicallimit value for NMP.
Keywords N-Methyl-2-pyrrolidone · Biomonitoring · Experimental exposure
Introduction
N-Methyl-2-pyrrolidone (NMP, CAS-No. 872-50-4)is a cyclic amide that covers a broad spectrum oftechnical and industrial applications due to its excel-lent miscibility with water and most organic solvents.NMP is used on a large scale for the extraction ofunsaturated and aromatic hydrocarbons in petro-chemistry and as a solvent for agriculture chemicals(HSE 1997). Workplace studies reported on the useof NMP for surface cleaning such as graViti and resinremoval (Anundi et al. 1993, 2000; Langworth et al.2001; Bader et al. 2006) or in the microelectronicsindustry (Beaulieu and Schmerber 1991). NMP alsoserves as a solvent for coating polymers such as acry-lates, epoxy resins, polyurethane lacquers, printingdyes and insulator plastics (HSE 1997; Åkesson2001). The annual production of NMP in the Euro-pean Union amounts to approximately 38,000 tons(Bader et al. 2002).
M. Bader (&) · R. WrbitzkyDepartment of Occupational Medicine, Hannover Medical School, OE 5370, Carl-Neuberg-Str. 1, 30625 Hannover, Germanye-mail: [email protected]
M. Blaszkewicz · C. van ThrielInstitute for Occupational Physiology at the University of Dortmund (IfADo), Ardeystr. 67, 44139 Dortmund, Germany
123
336 Arch Toxicol (2007) 81:335–346
NMP is a mild irritant to the eyes, the mucous mem-branes and the skin (Greim 2006). While the relativelylow overall toxicity of the compound has resulted, forinstance, in the recommendation of NMP as a substi-tute for dichloromethane in paint strippers in Ger-many, the observation of prenatal toxicity in animalstudies is an issue of ongoing scientiWc debate. NMP isclassiWed as a Group C compound in Germany, mean-ing that “there is no reason to fear a risk of damage tothe embryo or fetus when MAK (= maximum work-place concentrations) and BAT values (= biological tol-erance values) are observed” (DFG 2006). Recently,the Commission for the Investigation of Health Haz-ards in the Work Area of the Deutsche Forschungs-gemeinschaft (DFG) has conWrmed a workplace limitvalue for NMP in Germany of 20 ppm (82 mg/m3, cur-rently in force: 80 mg/m3).
From an occupational–medical point of view, thepotential dermal uptake of NMP is of particular interestfor the implementation of safety measures. NMP hasgot a skin notation in Germany and a relevant part ofthe overall uptake at the workplace might originatefrom dermal exposure. In this case, air monitoring andcompliance with workplace limit values is of limited sig-niWcance for the estimation of the internal dose possiblycausing systemic health eVects. Therefore, biologicallimit values for NMP are currently under discussion inscientiWc working groups such as the German Commis-sion for the Investigation of Health Hazards in theWork Area of the DFG, the ScientiWc Committee onOccupational Exposure Limits of the European Union(SCOEL) and the Biological Exposure Indices (BEI)Committee of the American Conference of Govern-mental Industrial Hygienists (ACGIH).
The metabolism of NMP in man after inhalational,oral and dermal uptake has been intensively investi-gated in recent years (e.g. Åkesson and Jönsson 1997,2000; Åkesson and Paulsson 1997; Akrill et al. 2002;Ligocka et al. 2003; Åkesson et al. 2004; Bader et al.2005). According to these studies, NMP is sequentiallyoxidised after bodily uptake by cytochrome P450enzymes to 5-hydroxy-N-methyl-2-pyrrolidone (5-HNMP), N-methylsuccinimide (MSI) and 2-hydroxy-N-methylsuccinimide (2-HMSI) (Fig. 1). Recently,Carnerup et al. (2006) reported on a fourth metabolite,2-pyrrolidone (2-P).
Two experimental inhalation studies with humanvolunteers by Åkesson and coworkers (e.g. Åkessonand Jönsson 2000; Jönsson and Åkesson 2003; Carn-erup et al. 2006) have revealed close correlationsbetween airborne NMP levels and the concentrationsof NMP and its metabolites in blood and urine. NMPitself also showed a general suitability for biomonitor-
ing purposes, but due to the risk of sample contamina-tion at the workplace this parameter seems notadvisable for occupational–medical surveillance.
Not particularly designed for the evaluation ofworkplace limit values but rather to study the toxicoki-netics of NMP and its metabolites, the experiments byÅkesson and co-workers comprised only six individu-als and three NMP concentrations or six individualsexposed four times to one NMP concentration in twodiVerent air humidity scenarios.
To improve the database on the correlation betweenexternal and internal exposure to NMP, especially withrespect to the evaluation of workplace limit values inbiological material, an extended human experimentalstudy was carried out including 16 volunteers and fourdiVerent NMP concentrations in a concentration rangefrom 10 mg/m3 up to the current German workplacelimit value of 80 mg/m3. Further emphasis was put onstudying the inXuence of physical workload and short-term peak exposures on the uptake and elimination ofNMP. Urinary NMP, 5-HNMP and 2-HMSI were cho-sen as parameters for the biological monitoringbecause the studies by Åkesson and co-workersrevealed no particular advantages of plasma over urineanalyses, and non-invasive sampling is generally betteraccepted for occupational–medical surveillance at theworkplace. To propose appropriate time frames forworkplace surveillance and to gain toxicokinetic data,the time courses of the elimination of NMP and its twomain metabolites in urine were recorded. Additionally,mass balance calculations were carried out to get anestimate of the body doses of NMP.
Materials and methods
Study group
Sixteen male volunteers (average age: 26.5 § 2.4 years)were recruited at the University of Dortmund, Ger-many. The participants were examined by a physician
Fig. 1 Structural formulas of N-methyl-2-pyrrolidone (NMP)and its main metabolites 5-hydroxy-N-methyl-2-pyrrolidone (5-HNMP) and 2-hydroxy-N-methylsuccinimide (2-HMSI)
N
CH 3
O N
CH3
OHO N
CH3
OO
OH
NMP 5-HNMP 2-HMSI
123
Arch Toxicol (2007) 81:335–346 337
to check their general Wtness for bicycle ergometry andto exclude subjects with respiratory restrictions, skindiseases or cardiovascular diseases (e.g. hypertension).The study was carried out in accordance with the Hel-sinki Declaration and was approved by the EthicsCommittee of the Institute for Occupational Physiol-ogy at the University of Dortmund (IfADo). All par-ticipants volunteered for the study and provided theirinformed consent.
Exposure chamber
The exposure laboratory at IfADo is built of glassand stainless steel with spatial dimensions of4.80 m £ 2.65 m £ 2.27 m (»29 m3). The room temper-ature was kept constant at 23°C (§ 6%), the relativehumidity was 40% (§13%) and a ventilation rate of9 h¡1 was maintained throughout the experiments. TheNMP concentration in air was adjusted by vapourisa-tion of liquid NMP on a heater platform followed bydispersion through a branched pipe system. Four sam-ple devices for online air monitoring were located atthe ceiling and samples were taken every 80 sec. Thequasi-continuous measurement of NMP in air was car-ried out by a standardised photoacoustic infrared spec-trometer (Gas-Monitor 1412, Innova, Ballerup,Denmark). Spot checks by means of stationary air sam-pling on silicagel tubes as described by Bader et al.(2006) conWrmed the accuracy and compliance of theNMP concentrations with their target values.
Experimental exposure scenarios
The exposure conditions included three diVerent NMPconcentrations (10, 40, 80 mg/m3) that were kept con-stant throughout the experiments and one variableexposure scenario with a baseline concentration of25 mg/m3 NMP and four 15 min periods with anincreased concentration of 160 mg/m3 and inter-peakintervals of 2 h. The eVective NMP concentrations forevery scenario are summarised in Table 1. The meanconcentration of airborne NMP in the experimentswith variable exposure was 72 mg/m3. In accordance
with typical work-shifts, the exposure times were set to2 £ 4 h with a 30 min exposure-free break after the Wrstinterval. A second exposure series was carried out withthe same NMP concentrations but included moderatephysical workload in 6 £ 10 min periods of exercise ona bicycle ergometer (75 W). Four participants wereexposed at a time. In both experimental series at restand with additional workload, the study participantswere regularly occupied with neuropsychological testbatteries and ratings assessing possible chemosensoryeVects due to NMP exposures (van Thriel et al., inpreparation).
Sample collection and storage
Every urine sample voided immediately before, duringand up to 40 h after the exposure sessions was collectedseparately in 500 mL polyethylene containers (totalsampling interval: 48 h). The individual sample vol-umes were recorded and the specimens were stored at4°C for a maximum interval of 24 h. Two aliquots ofurine were then transferred into 10 mL polyethylenecontainers and kept frozen at -18°C until transport toHannover Medical School (MHH) and subsequentanalysis.
Determination of NMP in urine
Urinary NMP was analysed according to Åkesson andPaulsson (1997), except for the introduction of an iso-tope-labelled internal standard to compensate fordiVerent extraction yields. All chemicals were of bestavailable analytical grade from Merck (Darmstadt,Germany) or from Fluka (Buchs, Switzerland). Ultra-pure water was prepared in-house (Quantum puriWersystem; Millipore, Schwalbach, Germany). d9-NMPwas purchased from Cambridge Isotope Laboratories(Andover, MA, USA).
Each urine sample (2 mL) was transferred into a10 mL screw capped vial and mixed together with 4 mLof 12 M aqueous potassium hydroxide solution and100 �L of internal standard (10 mg/L d9-NMP). Thesamples were then extracted with 2 mL toluene for
Table 1 Target concentra-tions and eVective NMPconcentrations in air (mean values, ranges in brackets)
Constant Constant Variable Constant
Target exposure (mg/m3) 10 40 25/160 = 72 80EVective exposure (mg/m3)-resting-
10.7 (10.3–11.0)
40.9 (39.7–43.8)
71.9 (71.1–73.0)
79.9 (79.7–80.2)
EVective exposure (mg/m3)-workload-
10.4 (10.0–11.1)
40.4(39.1–42.1)
72.3 (69.8–74.5)
79.4 (79.0–79.8)
123
338 Arch Toxicol (2007) 81:335–346
10 min on a laboratory shaker. After centrifugation(5 min, 3,500 £ g, room temperature) to separate theorganic layer from the aqueous solution, 1 mL of thetoluene extract was transferred into 1.8 mL samplevials and subjected to gas chromatography/mass spec-trometry (GC/MS).
The analyses were carried out on a 6890/5973 GC/MSD (Agilent, Palo Alto, CA, USA) equipped with acooled inlet system (CIS 4) and an MPS 2 autosam-pler (Gerstel, Mülheim, Germany). The CIS waskept at 120°C for injection (1 �L) and was thenheated at a rate of 12°C/s to 300°C. GC oven: 80°Cinitial, 10°C/min to 160°C, 25°C/min to 260°C, 6 minisothermal, interface 260°C, column: HP InnoWax30 m £ 0.25 mm, 0.25 �m Wlm, carrier gas: helium 5.0,1.2 mL/min constant Xow, split 1:10, ion source 230°C,electron energy 70 eV, quadrupole 150°C, electronmultiplier according to autotune settings. The massfragments scanned were 99 and 98 m/z (d9-NMP: 108and 106 m/z) with a dwell time of 100 ms for each ion.The retention times of NMP and d9-NMP were 7.14and 7.12 min, respectively.
Calibration standards were prepared from a pooledurine sample of non-exposed individuals in a concen-tration range between 10 and 10,000 �g/L. The intra-and interassay imprecision for the analysis of a spikedurine sample (200 �g/L) were 4% (ten analyses of thesame sample) and 6% (one analysis on 24 consecutivedays). A limit of quantiWcation of 10 �g/L was calcu-lated. NMP concentrations below this value were gen-erally adjusted to 5 �g/L.
Determination of 5-HNMP and 2-HMSI in urine
Urinary 5-HNMP and 2-HMSI were analysed accord-ing to Jönsson and Åkesson (1997) with minor modiW-cations. All chemicals were of best available analyticalgrade from Merck (Darmstadt, Germany) or fromSigma-Aldrich (Deisenhofen, Germany). 5-HNMP, d4-5-HNMP, d3-2-HMSI were from Synthelec (Lund,Sweden), 2-HMSI was from Fluka (Buchs, Switzer-land). Ultra-pure water was prepared in-house (Quan-tum puriWer system; Millipore, Schwalbach, Germany).
Urine (1 mL) was added to 4 mL water and 100 �Linternal standard (d4-5-HNMP, d3-2-HMSI; c = 500mg/L). One mL of this solution was then transferred toa solid phase extraction column (200 mg ENV+, Separ-tis, Germany) preconditioned with 5 mL methanoland 10 mL water. The sample was slowly passed throughthe column to allow adsorption of 5-HNMP and2-HMSI. The column was then dried by applying amoderate vacuum for 10 min and a subsequent centri-fugation (1,500 £ g, 10 min, room temperature). The
metabolites were eluted with 2 mL ethyl acetate/meth-anol (80/20, v/v) and transferred into a GC vial. Theextract was evaporated to dryness under a gentlestream of nitrogen and the residue was re-dissolved in50 �L bis(trimethylsilyl) triXuoroacetamide. The sam-ple was tempered at 110°C for 60 min. Finally, 1 mLethyl acetate was added to the vial and subjected toGC/MS analysis.
One microlitre was injected with a split of 1:20into a GC/MSD 6890/5973 system (Agilent Technol-ogies, Palo Alto, CA, USA) equipped with a CIS 4injector and an MPS 2 autosampler (both from Ger-stel, Mülheim, Germany) and a HP 50+ capillary col-umn (30 m £ 0.25 mm, 0.25 �m Wlm). The CIS waskept at 120°C for injection and was then heated at arate of 12°C/s to 280°C (isothermal for 1 min). Ovenprogram: 70°C initial, 5°C/min to 125°C, 25°C/min to250°C, isothermal for 5 min. Carrier gas: helium 5.0,constant Xow of 1.2 mL/min. The interface tempera-ture was set to 260°C. Characteristic ion fragmentsof 5-HNMP (186 and 172 m/z) and of d4-5-HNMP(190 and 176 m/z) were monitored after electronimpact ionisation at 70 eV with a dwell time of100 ms per ion (2-HMSI: 186 and 144 m/z, d3-2-HMSI:189 and 147 m/z). The electron multiplier wasoperated according to autotune settings. The reten-tion times were 12.53 min for 5-HNMP and 12.48 minfor d4-5-HNMP (2-HMSI: 12.60 min, d3-2-HMSI:12.56 min).
Calibration standards were prepared from pooledurine samples of non-exposed individuals (5-HNMPand 2-HMSI not detectable) in concentration ranges of2.5–200 mg/L 5-HNMP and 1.25–100 mg/L 2-HMSI.The imprecision of the analysis of 5-HNMP (150 mg/L)was 2% (intra-assay, 10 samples analysed within oneseries) and 10% (inter-assay, one sample analysed on24 consecutive days). In the case of 2-HMSI (75 mg/L),the imprecision was 1% (intra-assay) and 3% (inter-assay), respectively. A limit of quantiWcation of 2 mg/Lwas calculated in both cases. Analytical results belowthis value were set to 1 mg/L.
Determination of creatinine in urine
Urinary creatinine was analysed photometricallyaccording to the JaVé reaction. Urine (100 �L) wasadded to 2 mL saturated picric acid and 150 �L 10%sodium hydroxide solution. After 10 min at room tem-perature, the samples were diluted with 7.75 mL waterand left for another 5 min at room temperature. Onemillilitre of each sample was subsequently transferredinto a polypropylene cuvette and the absorption of thesample was measured (� = 546 nm).
123
Arch Toxicol (2007) 81:335–346 339
Quality control
Quality control (QC) samples were prepared fromurine specimens of non-exposed individuals (200 �g/LNMP, 20 and 150 mg/L 5-HNMP, 10 and 75 mg/L2-HMSI). Every analytical series was accompanied byQC samples to verify the stability of the methods andsamples throughout the study. The accuracy of thedetermination of 5-HNMP, 2-HMSI and creatinine wassuccessfully tested in three round robins of the Deut-sche Gesellschaft für Arbeitsmedizin und Umweltmed-izin (DGAUM) (c/o Institute for Occupational, Socialand Environmental Medicine of the University ofErlangen-Nuremberg, Germany) during the study.
Data processing and calculations
All calculations were carried out with Microsoft Excel2000® or SPSS 13.0©. The Mann–Whitney U-test (inde-pendent samples) or the Wilcoxon signed rank test(paired samples) were applied for the analysis of diVer-ences between subgroups. Linear regression was usedfor correlation analyses between pairs of variables. Thereported P-values are two-sided and statistical signiW-cance was accepted for � < 0.05. Polynomic trend lineswere used to visualise the time courses of the elimina-tion curves. Assuming Wrst-order elimination kineticsfor all biomarkers, linear regression analyses of semilog-arithmic concentration versus time plots were appliedto calculate the biological half-lives (t1/2 = ln 2/k, withk = slope of the linear regression curve). Urinary creat-inine was analysed to correct for diuretic eVects. Toavoid a bias from high or low concentrated urines, onlysamples with creatinine contents between 0.5 and 2.5 g/L were used for kinetic calculations (peak times, uri-nary half-lives) or for the correlation analyses. Thecomplete set of samples, however, was accepted fortotal uptake and mass balance calculations. In the caseof metabolically unchanged NMP, the volume-relatedconcentrations were used as opposed to the creatininerelation of 5-HNMP and 2-HMSI (Åkesson and Jöns-son 2000). Both 5-HNMP and 2-HMSI are consideredto undergo glomerular Wltration like creatinine, whileunmetabolised NMP is subject to tubular absorption(Carnerup 2004).
Results
Elimination kinetics
The time-dependent urinary elimination of NMP andits metabolites is shown in Figs. 2, 3, 4. Urinary NMP
increased rapidly after the onset of exposure and anelimination peak within the Wrst hour post-exposurewas generally observed (Fig. 2). The elimination ofNMP in urine was completed after 24 h post-exposure(t = 32 h). Only in isolated cases, irregularly high NMPconcentrations were seen in the time period after 16 h,mostly associated with concentrated urine samples(creatinine > 2.0 g/L). The mean biological half-life ofurinary NMP was 3.8 § 0.2 h, calculated from the dataof the time interval between 8 and 32 h. No signiWcantinXuence of physical workload, intensity of exposure orvariable exposure on the peak times, half-lives andtotal elimination times was observed.
As summarised in Table 2, the peak concentrationsof NMP increased with the intensity of the exposurefrom a median 265 �g/L (mean § standard deviation:295 § 93 �g/L; 10 mg/m3 NMP) up to 2,192 �g/L (mean2,296 § 759 �g/L; 80 mg/m3) under resting conditions.
Fig. 2 Elimination of NMP in urine after constant exposure un-der resting conditions (triangles 10 mg/m3, crosses 40 mg/m3, cir-cles 80 mg/m3)
0
1000
2000
3000
4000
0 8 16 24 32 40 48time (h)
NM
P (
µg/L
)
Fig. 3 Elimination of 5-HNMP in urine after constant exposureunder resting conditions (triangles 10 mg/m3, crosses 40 mg/m3,circles 80 mg/m3)
0
40
80
120
160
200
0 8 16 24 32 40 48
time (h)
5-H
NM
P (
mg/
g cr
eatin
ine)
123
340 Arch Toxicol (2007) 81:335–346
Moderate physical workload increased the NMP peakconcentrations by 45% on average, but due to the indi-vidual variability (»30%) the eVect of workload wasstatistically signiWcant only for the 80 mg/m3 conditions(P < 0.05, Wilcoxon signed rank test).
The concentrations of 5-HNMP in urine showed aless steep increase as compared to NMP with post-exposure peak concentrations of 2–4 h in all experi-mental scenarios (Fig. 3). The urinary elimination of5-HNMP was generally completed after 32 h post-exposure (t = 40 h), although in some cases 5-HNMPconcentrations between 2 and 15 mg/g creatinine werestill detectable. The average biological half-life of5-HNMP in urine of 7.4 § 1.0 h (data from the timeinterval between 10 and 34 h) was not inXuenced bythe intensity or variation of exposure or by workload.A longer half-life of 9.8 h was calculated for the 10 mg/m3 exposure under resting conditions, but the datawere more scattered in this case than in the other
exposure conditions (correlation coeYcient r = 0.671 vs.mean r = 0.815) and thus the determination of t1/2 isless accurate. As shown in Table 2, the maximum con-centrations increased with the intensity of exposurefrom a median 16 mg/g creatinine (mean 16 § 5 mg/gcreatinine) for an 8 h exposure to 10 mg/m3 NMPunder resting conditions to 106 mg/g creatinine (mean:120 § 41 mg/g creatinine) for 80 mg/m3. Workloadincreased the urinary 5-HNMP peak concentrations by36% on average. In this case, the diVerences betweenthe resting and the workload experiments were signiW-cant for the 40 and 72 mg/m3 designs (P < 0.05, Wilco-xon signed rank test).
The urinary elimination of 2-HMSI in urine is shownin Fig. 4. Throughout the exposure time the 2-HMSIconcentrations were generally below 10 mg/g creati-nine (limit of quantiWcation: 2 mg/L). The individualpeak times were broadly distributed in a rangebetween 16 and 24 h post-exposure (t = 24–32 h) withan average of about 18 h post-exposure (t = 26 h). Theelimination was not completed after the 48 h totalobservation time of the study, but the peak time of2-HMSI and the Wrst half-life interval were covered,thus allowing at least a rough estimate of the biologicalhalf-life of 2-HMSI in urine of 24 § 4 h (data from thetime interval between 28 and 48 h) (Table 2). No inXu-ence of workload, intensity of exposure or variableexposure on the peak-times and biological half-lives wasobserved. Under resting conditions, the peak concen-trations of urinary 2-HMSI increased with the intensityof the exposure from a median 5.4 mg/g creatinine(mean: 5.7 § 1.6 mg/g creatinine) for an 8 h exposureto 10 mg/m3 up to 30.6 mg/g creatinine (mean:32.6 § 7.4 mg/g creatinine) for an exposure to 80 mg/m3.The average peak concentrations of 2-HMSI in theworkload designs were 21% higher as compared to the
Fig. 4 Elimination of 2-HMSI in urine after constant exposureunder resting conditions (triangles 10 mg/m3, crosses 40 mg/m3,circles 80 mg/m3)
0
10
20
30
40
50
0 8 16 24 32 40 48
time (h)
2-H
MS
I (m
g/g
crea
tinin
e)
Table 2 Biological half-lives and median peak concentra-tions of NMP and its main metabolites in urine
Exposure 10 mg/m3 40 mg/m3 72 mg/m3 80 mg/m3
Half-life t1/2
Peak concentration Median
NMPa Resting 3.6 h265
3.9 h1031
3.8 h2295
3.5 h2192
Workload 4.1 h411
3.5 h1647
4.0 h2532
3.7 h3422
5-HNMPb Resting 9.8 h16
7.5 h57
6.9 h113
6.6 h106
Workload 7.1 h21
7.5 h81
7.2 h133
6.8 h161
2-HMSIb Resting 20 h5.4
28 h18.7
63 hc
29.622 h30.6
Workload 23 h5.8
26 h21.2
19 h36.2
30 h43.0
a NMP given in �g/Lb 5-HNMP and 2-HMSI given in mg/g creatininec Regression not signiWcant
123
Arch Toxicol (2007) 81:335–346 341
same exposures under resting conditions (Table 2), butstatistical signiWcance was reached only for the 72 and80 mg/m3 design (P < 0.05, Wilcoxon signed rank test).
Mass balance calculations
The absolute amount of NMP and its metabolites inevery urine specimen was calculated from theirconcentrations and the volume of the correspondingsample. After conversion into molar masses, the summa-tion of these data yielded the absolute molar amountseliminated within the 48 h observation interval. Addi-tionally, a correction factor of 1.03 for metabolites thatwere not analysed (N-methylsuccinimide, 2-pyrroli-done) was applied. In the case of 2-HMSI, the elimina-tion was not completed during the sampling interval.According to Åkesson and Jönsson (1997), the fractionof 2-HMSI eliminated in urine after 48 h accounts forabout 21% of the total 2-HMSI excretion after oraluptake of 100 mg NMP (5.4 mg out of 25.7 mg). Thus,the total molar mass of 2-HMSI in this study wascorrected for the missing percentage. In a Wnal calcula-tion, the molar amounts of NMP and its metaboliteswere transformed into mass equivalents of the parentcompound NMP to give an estimate for the whole bodydose. The results of the mass balance calculations aresummarised in Table 3.
The total amount of urinary eliminated NMPincreased with the intensity of the external exposurefrom a median 3 up to 26 �mol under resting condi-tions and from 4 up to 24 �mol following moderateworkload. The corresponding mean values showed largestandard deviations of about 35% (range 24–57%), thus
indicating a broad range of interindividual results. Theratio between the highest and the lowest value in everyexposure series ranged from 3 to 7. In consequence, thediVerences between the conditions “resting” and“workload” generally did not reach statistical signiW-cance for unmetabolised NMP in urine, neither on agroup basis nor on an individual level. Nevertheless,the average ratio between the median values at restand with physical workload was 1.2, at least suggestinga slight increase of 20% due to workload.
In the case of 5-HNMP, the total eliminated amountvaried from 217 �mol after an exposure to 10 mg/m3
(resting conditions) to 1400 �mol for 80 mg/m3 (work-load conditions). The standard deviation of the meanvalues was 34% (22–44%) and the ratios between thehighest and the lowest results ranged from 2 to 5. The5-HNMP elimination was increased on average by afactor of 1.2 following moderate physical workload ascompared to the resting conditions. This diVerence wasstatistically signiWcant for the exposures to 40, 72 and80 mg/m3 (P < 0.05, Wilcoxon signed rank test).
The total amount of 2-HMSI in urine alsoincreased with the intensity of the exposure to air-borne NMP and with workload from 103 �mol(10 mg/m3, resting conditions) to 756 �mol (80 mg/m3,workload conditions). The standard deviation of themean values was 26% (range 18–38%) and thus lowerthan for the two other parameters. Individual resultswere less scattered and the ratios between the highestand the lowest values were generally between 2 and 3.The 2-HMSI elimination was by a factor of 1.3 higherin the workload experiments as compared to the rest-ing scenarios, but the diVerence was signiWcant onlyfor the 80 mg/m3 conditions (P < 0.05, Wilcoxonsigned rank test).
The total excreted amount of NMP equivalents wasproportional to the intensity of the external exposure(Table. 3). Under resting conditions, between 33 and171 mg NMP were recovered in urine within 48 h.Moderate physical workload increased the amount ofNMP equivalents in urine. This eVect was more pro-nounced in the scenarios with higher exposure inten-sity (37% for variable 72 mg/m3, 35% after constant80 mg/m3) as compared to the lower concentrations.The diVerences were signiWcant for the 40, 72 and80 mg/m3 conditions (P < 0.05, Wilcoxon signed ranktest).
Correlation analyses
The correlations between the exposure to airborneNMP and the peak concentrations of the three biomar-kers in post-exposure urine were analysed by linear
Table 3 Mass balance calculations for the urinary elimination ofNMP and its main metabolites
Median values, a corrected for 21% missing 2-HMSIb Corrected for 3% metabolites not analysed
10 mg/m3 40 mg/m3 72 mg/m3 80 mg/m3
NMP (�mol)Resting 3 11 22 26Workload 4 17 21 24
5-HNMP (�mol)Resting 217 565 878 1052Workload 174 670 1213 1400
2-HMSIa (�mol)Resting 103 255 456 569Workload 113 378 589 756
Total amountb (�mol)Resting 331 933 1406 1725Workload 311 1055 1929 2333
Total amount (mg)Resting 33 92 139 171Workload 31 104 191 231
123
342 Arch Toxicol (2007) 81:335–346
regression. All analyses revealed signiWcant correla-tions (P < 0.05) and the regression coeYcients werebetween r = 0.810 and 0.913.
In the case of unmetabolised NMP (Fig. 5), theregression analysis pointed to a urinary concentrationof (31 £ 80) - 80 = 2,400 �g/L in post-shift urine for an8 h exposure to the current German workplace limitvalue of 80 mg/m3. The 95% conWdence interval forthis condition ranged from 1,500–3,300 �g/L. Moderateworkload increased the average peak concentration inurine to 3,400 �g/L NMP (95% conWdence interval:1,800–5,000 �g/L).
For 5-HNMP, an 8 h exposure to 80 mg/m3 NMPresulted in a calculated elimination of 117 mg 5-HNMP/gcreatinine in 2–4 h post-shift urine under resting condi-tions with a 95% conWdence interval of 80–160 mg/gcreatinine (Fig. 6). An increase to 150 mg/g creatininedue to moderate workload was observed (95% conW-dence interval: 110–190 mg/g creatinine).
The linear regression analysis for 2-HMSI showed aconcentration of 32 mg/g creatinine in 18 h post-expo-sure urine (t = 26 h) under resting conditions (95%conWdence interval: 20–45 mg/g creatinine) and 44 mg/g creatinine (95% conWdence interval: 30–58 mg/g cre-atinine) after moderate workload (Fig. 7).
Discussion
This human experimental study broached the issue ofNMP uptake and elimination under simulated workplace
conditions, i.e. an 8 h exposure interrupted by a shortbreak of 30 min. The concentrations of airborne NMPcovered a range between 10 mg/m3 up to the currentGerman workplace limit of 80 mg/m3. Particularemphasis was put on the frequent and complete sam-pling of urine for the estimation of an appropriate sam-pling time for workplace surveillance and also, from atoxicological point of view, for mass balance calculationsand correlation analyses. The eVect of a moderate physi-cal workload on the uptake and elimination of NMP wasstudied. Workload enhances the minute volume andexposure scenarios under resting conditions may there-fore underestimate the NMP uptake at workplaces.
Elimination kinetics
The basic toxicokinetic data obtained in this studyaccord with those reported e.g. by Jönsson and Åkes-son (2003) and by Carnerup et al. (2006). Unmetabo-lised NMP as well as 5-HNMP showed an eliminationmaximum shortly after the end of an exposure andshould be analysed in post-shift urine for workplacesurveillance. However, the short peak time interval ofNMP renders this parameter more susceptible to varia-tions in the sampling time. Additionally, urine samplesare prone to external contamination with NMP espe-cially if the workplace and sampling venue are notstrictly separated or if the sampling material is storedunder unfavourable conditions. The sampling time of5-HNMP is less critical and an external contaminationor subsequent formation is unlikely.
Fig. 5 Correlation between NMP in air and urinary NMP(dashed lines 95% conWdence intervals). Left side resting condi-tions, NMP (�g/L) = 31 (mg NMP/m3) - 80, r = 0.829, P < 0.05.
Right side workload conditions, NMP (�g/L) = 45 (mg NMP/m3) -196, r = 0.810, P < 0.05
806040200
NMP (mg/m3)
7000
6000
5000
4000
3000
2000
1000
0
NM
P (
µg/L
)
806040200
NMP (mg/m3)
7000
6000
5000
4000
3000
2000
1000
0
NM
P (
µg/L
)
123
Arch Toxicol (2007) 81:335–346 343
In the case of 2-HMSI, the delayed peak maximumof 16–24 h post-exposure and the long biological half-life make this parameter especially suitable for the sur-veillance of accumulative eVects during a workingweek. The accumulative eVect is less pronounced for5-HNMP as more than three half-lives elapse betweentwo post-shift peaks. It has to be noted, however, that
the observation interval in this study was too short tocover more than one half-life of 2-HMSI and theresults for t1/2 are necessarily less accurate than thosefor NMP and 5-HNMP. According to Jönsson andÅkesson (2003) and Carnerup et al. (2006), the biologi-cal half-life of 2-HMSI after inhalational uptake ofNMP is 17–18 h. The question of accumulation was
Fig. 6 Correlation between NMP in air and urinary 5-HNMP(dashed lines 95% conWdence intervals). Left side resting condi-tions, 5-HNMP (mg/g creatinine) = 1,47 (mg NMP/m3) - 0.40,
r = 0.869, P < 0.05. Right side workload conditions, 5-HNMP (mg/g creatinine) = 1,90 (mg NMP/m3) - 2.45, r = 0.913, P < 0.05
806040200
NMP (mg/m3)
200
150
100
50
0
5-H
NM
P (
mg/
g cr
eatin
ine)
806040200
NMP (mg/m3)
200
150
100
50
0
5-H
NM
P (
mg/
g cr
eatin
ine)
Fig. 7 Correlation between NMP in air and urinary 2-HMSI(dashed lines: 95% conWdence intervals). Left side restingconditions 2-HMSI (mg/g creatinine) = 0.38(mg NMP/m3) + 2.0,
r = 0.829, P < 0.05. Right side workload conditions 2-HMSI(mg/g creatinine) = 0.54 (mg NMP/m3) + 0.3, r = 0.889, P < 0.05
806040200
NMP (mg/m3)
60
50
40
30
20
10
0
2-H
MS
I (m
g/g
crea
tinin
e)
806040200
NMP (mg/m3)
60
50
40
30
20
10
0
2-H
MS
I (m
g/g
crea
tinin
e)
123
344 Arch Toxicol (2007) 81:335–346
not particularly addressed in this NMP study butwould make an interesting aspect for future works on2-HMSI.
The toxicokinetics of the biomarkers (peak times,biological half-lives, elimination interval) were notaVected by variable exposure (25 mg/m3 with 160 mg/m3 short-term peaks). The variable exposures corre-sponded to a time-weighted average concentration of72 mg/m3 and the data from these experiments were inline with those from constant exposures. This observa-tion further supports the applicability and reliability ofNMP biomonitoring for workplace surveillance.
Mass balance calculations
The molarity-based calculations allowed an estimateof the total body dose of NMP. An 8 h exposure to80 mg/m3 NMP under resting conditions resulted inthe elimination of NMP and metabolites equivalent to171 mg NMP. With an average minute volume of 5 Lat rest (Greim and Lehnert 1998), a human test personwould inspire 5 L £ 60 min £ 8 h = 2,400 L = 2.4 m3
per 8 h exposure. Assuming a pulmonal retention fac-tor of 0.90 (Åkesson and Jönsson 2000), about2.4 m3 £ 80 mg/m3 £ 0.90 = 173 mg NMP would beabsorbed. This estimation is in good accordance withthe observed 171 mg. However, the retention factor of0.90 (0.88–0.93) should be regarded as a rough esti-mate that was calculated from the diVerences betweeninhaled and exhaled air. The underlying experimentswere not described in detail and therefore diYcult toassess.
In the case of moderate workload, the minute vol-ume should increase to about 26.5 L for a 75 W load(Greim and Lehnert 1998), resulting in a time-weighted average of (6 £ 10 min with 26.5 L/min) +(7 £ 60 min with 5 L/min) = 7.7 L in this study. A totalinspiration volume of 7.7 L/min £ 60 min £ 8 h =3,700 L = 3.7 m3 would yield an uptake of 266 mg NMPfor an exposure to 80 mg/m3 (pulmonal retention of90%). The diVerence to the observed amount of NMPequivalents of 266-231 mg = 35 mg NMP is relativelysmall, given the uncertainty of the assumptions madebefore.
However, this Wrst calculation approach disregardsimportant inXuence factors that contribute to theabsorption and elimination of NMP. Åkesson and Jöns-son (1997) reported a total dose recovery of only 65%in urine following an oral uptake of NMP in a humanexperimental study and concluded that either gastroin-testinal absorption was incomplete or that yet unidenti-Wed NMP metabolites were missing. In contrast, animalexperiments have shown recoveries of 95% in rat urine
after oral administration of NMP (Midgley et al. 1992).Given that NMP is rapidly absorbed across the skin(Akrill et al. 2002, Åkesson et al. 2004, Bader et al.2005), it seems more likely that gastrointestinal uptakeis high and that missing fractions represent NMP dis-tributed in the body compartments or transformed intoyet unknown metabolites. In contrast to the results ofthe oral uptake experiments, Åkesson and Jönsson(2000) calculated an almost complete body dose recov-ery after inhalational exposure of human volunteers. Inthis case, a longer biological half-life of 5-HNMP (»7 h)was noted after inhalational exposure as comparedto oral uptake (»4 h) and the elimination curve of5-HNMP showed a biphasic pattern. These resultsmight reXect an additional dermal uptake of NMP fromcontaminated surfaces and clothings or from thevapour phase and the calculated respiratory uptakewould underestimate the total body dose by a factor of35% (Carnerup et al. 2006).
If a urinary recovery rate of 65% is applied to thecalculated NMP uptake, about 173 mg £ 0.65 = 112 mgNMP equivalents would be expected for an 8 h expo-sure to 80 mg/m3 under resting conditions. In the caseof workload during the exposure, a total excretedamount of 266 mg £ 0.65 = 173 mg NMP equivalentsshould be found. The diVerences between these calcu-lated and the experimentally observed amounts of171 - 112 mg = 59 mg (resting conditions) and 231 -173 mg = 58 mg (workload conditions) are equal and
thus point to a constant inXuence factor on the NMPuptake in both designs. Transdermal absorption mightbe such a factor, at least if the exposed skin area is ofthe same size and enhanced blood circulation or tran-spiration can be ruled out. These observations supportthe view of a signiWcant contribution of the dermalroute of as much as one third to the NMP uptake afterwhole-body exposure.
Linear regression analyses
While unmetabolised urinary NMP is not recommend-able for workplace surveillance due to possible samplecontamination, the parameter shows a close correla-tion to airborne NMP. At a concentration of 80 mg/m3
NMP, approximately 2,400 �g/L NMP in urine werecalculated for an 8 h exposure under resting conditions,and 3,400 �g/L following moderate physical workload.No calculated data on post-exposure NMP concen-trations in urine were reported from an inhalationexperiment by Åkesson and Paulsson (1997), but aWgure-based estimation points to approximately 1,000 �g/L after an 8 h exposure to 25 mg/m3 NMP. An extrapola-tion to 80 mg/m3 yields (1,000 �g/L £ 80/25) = 3200 �g/
123
Arch Toxicol (2007) 81:335–346 345
L, a value close to the results of the workload condi-tions in this study.
The linear regression analysis for NMP in air versuspost-exposure 5-HNMP in urine showed an averageconcentration of 117 mg/g creatinine for an exposureto 80 mg/m3 NMP under resting conditions. The regres-sion curve reported by Åkesson and Jönsson (2000) fora series of 8 h exposures to 10, 25 and 50 mg NMP/m3
(y = 2.23 (mg NMP/m3) - 0.066 (with y = mg 5-HNMP/g creatinine) was considerably steeper than the curvecalculated in this study. Apart from the small eVect ofthe constant term, the ratio between the Swedish dataand this study is (2.23/1.47) = 1.5. A comparison of theworkload experiments with the data from Åkesson andJönsson (2000) shows a ratio of 1.2 with respect to theslopes of the regression curves (2.23/1.90).
Linear regression analyses between airborne NMPand urinary 2-HMSI peak values (16–24 h post-expo-sure) yielded concentrations of 32 mg/g creatinineunder resting conditions and 44 mg/g creatinine aftermoderate workload for an 8 h exposure to 80 mg/m3
NMP. Comparable to the situation with 5-HNMP, theresults from an inhalation study by Jönsson and Åkes-son (2003) showed higher urinary 2-HMSI concentra-tions than seen in this study. The authors found 51 mg/g creatinine on the morning after an 8 h exposure to50 mg/m3 NMP, corresponding to 82 mg/g creatininefor 80 mg/m3 NMP. The ratios between the concentra-tions of 2-HMSI are (82/32) = 2.6 (resting conditions)and (82/44) = 1.9 (moderate workload).
In conclusion, the results of the Swedish studies andthis experimental design have revealed diVerences inthe analysed or estimated concentrations of NMP andits metabolites in urine. While these diVerences are rel-atively small in the case of unmetabolised NMP andmight have been attributed only to interindividual vari-ations, the 5-HNMP results of this study are approxi-mately 30% lower than the Swedish data, and roughlyhalf as high in the case of 2-HMSI.
Some inXuence factors that vary between the studiesmay contribute to these diVerences. The Swedish expo-sure chamber described by Åkesson and Paulsson(1997) had a total volume of 6 m3, a ventilation rate of20 h¡1, and two volunteers were exposed at a time. Theexposure laboratory used in the current study was aroom of 29 m3 with large window panes (ventilationrate 9 h¡1) and four persons were exposed at a time.Apart from this diVerent exposure environment, indi-vidual factors such as age, general Wtness and bicycleroutine might have an eVect on physiological parame-ters, e.g. the minute volume. Also, the workloaddesigns in this study have shown that physical activityplays a central role for the uptake of NMP. Therefore,
even small diVerences in the baseline activities of theparticipants may have had a signiWcant eVect on theinternal exposure to NMP. Not at least, dermal absorp-tion of NMP might be considered to explain the dis-crepancies between the studies. NMP is rapidlyabsorbed across the skin and any contact to NMP con-taminated surfaces is likely to result in a dermal uptakeof the solvent (Bader et al. 2005; Carnerup et al. 2006).
Conclusions
The three biomarkers NMP, 5-HNMP and 2-HMSI inurine have shown close correlations between theirpost-shift peak concentrations and the exposure toNMP. The most promising candidates for biomonitor-ing purposes and workplace surveillance are 5-HNMPand 2-HMSI. Between rounded 120 and 150 mg5-HNMP/g creatinine and 30–45 mg 2-HMSI/g creati-nine correspond to an exposure to 80 mg/m3 NMP,depending on physical activity. The diVerencesbetween the estimated and observed total amount ofurinary metabolites in this study point to a signiWcantcontribution of dermal absorption on the overalluptake of NMP. This aspect, together with the inXu-ence of physical workload, should be considered forthe evaluation of a biological limit value for NMP.
Acknowledgments The authors thank the participating labora-tory and scientiWc staVs of IfADo and MHH for their excellenttechnical assistance. This study was Wnancially supported bygrants of the NMP Producers Group, c/o Bergeson & Campbell,Washington DC, USA.
References
Åkesson B (2001) N-Methyl-2-pyrrolidone. Concise interna-tional chemical assessment document 35, International pro-gramme on chemical safety, World Health Organization,Geneva
Åkesson B, Carnerup M, Jönsson BAG (2004) Evaluation ofexposure biomarkers from percutaneous absorption of N-methyl-2-pyrrolidone. Scand J Work Environ Health30:306–312
Åkesson B, Jönsson BAG (1997) Major metabolic pathway forN-methyl-2-pyrrolidone in humans. Drug Metab Dispos25:267–269
Åkesson B, Jönsson BAG (2000) Biological monitoring of N-me-thyl-2-pyrrolidone using 5-hydroxy-N-methyl-2-pyrrolidonein plasma and urine as the biomarker. Scand J Work EnvironHealth 26:213–218
Åkesson B, Paulsson K (1997) Experimental exposure of malevolunteers to N-methyl-2-pyrrolidone (NMP): acute eVectsand pharmacokinetics of NMP in plasma and urine. OccupEnviron Med 54:236–240
Akrill P, Cocker J, Dixon S (2002) Dermal exposure to aqueoussolutions of N-methylpyrrolidone. Toxicol Lett 134:265–269
123
346 Arch Toxicol (2007) 81:335–346
Anundi H, Langworth S, Johanson G, Lind ML, Åkesson B, FriisL, Itkes N, Söderman E, Jönsson BAG, Edling C (2000) Airand biological monitoring of solvent exposure during graVitiremoval. Int Arch Occup Environ Health 73:561–569
Anundi H, Lind ML, Friis L, Itkes N, Langworth S, Edling C(1993) High exposures to organic solvents among graVitiremovers. Int Arch Occup Environ Health 65:247–251
Bader M, Keener SA, Wrbitzky R (2005) Dermal absorption andurinary elimination of N-methyl-2-pyrrolidone. Int Arch Oc-cup Environ Health 78:673–676
Bader M, Rosenberger W, Rebe Th, Keener S, Brock TH, Hem-merling H, Wrbitzky R (2006) Ambient monitoring and bio-monitoring of workers exposed to N-methyl-2-pyrrolidone(NMP) in an industrial facility. Int Arch Occup EnvironHealth 79:357–364
Bader M, Will W, Rossbacher R, Nasterlack M, Wrbitzky R (2002)N-Methyl-2-pyrrolidon - Arbeitsmedizinische Bedeutung undToxikologie. Arbeitsmed Sozialmed Umweltmed 9:422–428
Beaulieu HJ, Schmerber KR (1991) M-Pyrol(TM) (NMP) use in themicroelectronics industry. Appl Occup Environ Hyg 6:874–880
Carnerup MA (2004) Analysis, metabolism, eVects and biologicalmonitoring of N-methyl-2-pyrrolidone (NMP). Thesis, Fac-ulty of Medicine, Lund University, Sweden
Carnerup MA, Spanne M, Jönsson BA (2006) Levels of N-meth-yl-2-pyrrolidone (NMP) and its metabolites in plasma andurine from volunteers after experimental exposure to NMPin dry and humid air. Toxicol Lett 162:139–45
DFG (Deutsche Forschungsgemeinschaft) (2006) List of MAKand BAT values 2006. Report No. 42, Wiley-VCH, Wein-heim
Greim H (ed.) (2006) N-Methyl-2-pyrrolidon (Dampf). Maxi-male Arbeitsplatzkonzentrationen. Arbeitsmedizinisch-
toxikologische Begründungen. Senatskommission zur Prü-fung gesundheitsschädlicher ArbeitsstoVe. Deutsche Fors-chungsgemeinschaft. 41. Lieferung, Wiley-VCH, Weinheim
Greim H, Lehnert G (eds.) (1998) Die Bedeutung der Arbeitsleis-tung für die Inhalationskinetik. Toxikologisch-arbeitsmedizini-sche Begründungen von MAK-Werten, Senatskommission zurPrüfung gesundheitsschädlicher ArbeitsstoVe. DeutscheForschungsgemeinschaft. 27. Lieferung, Wiley-VCH, Wein-heim
HSE (Health and Safety Executive) (1997) N-Methyl-2-pyrroli-done. Risk assessment document. EH72/10, ISBN0717615286, HSE Books, Sudbury, UK
Jönsson BAG, Åkesson B (1997) Determination of 5-hydroxy-N-methylpyrrolidone and 2-hydroxy-N-methylsuccinimide inhuman urine. J Chrom B 694:351–357
Jönsson BAG, Åkesson B (2003) Human experimental exposureto N-methyl-2-pyrrolidone (NMP): toxicokinetics of NMP,5-hydroxy-N-methyl-2-pyrrolidone, N-methylsuccinimideand 2-hydroxy-N-methylsuccinimide (2-HMSI) and biologi-cal monitoring using 2-HMSI as a biomarker. Int Arch Oc-cup Environ Health 76:267–274
Langworth S, Anundi H, Friis L, Johanson G, Lind ML, Söder-man E, Åkesson B (2001) Acute health eVects common dur-ing graViti removal. Int Arch Occup Environ Health 74:213–218
Ligocka D, Lison D, Haufroid V (2003) Contribution of CYP2E1to N-methyl-2-pyrrolidone metabolism. Arch Toxicol77:261–266
Midgley I, Hood AJ, Chasseaud LF, Brindley CJ, Baughman S,Allan G (1992) Percutaneous absorption of co-administeredN-methyl-2-[14C]pyrrolidinone and 2-[14C]pyrrolidinone inthe rat. Food Chem Toxic 30:57–64
123