Annals of Occupational Hygiene Advance Access originally published online on September 1, 2005
Annals of Occupational Hygiene 2005 49(8):683-690; doi:10.1093/annhyg/mei028
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2005 British Occupational Hygiene Society Published by Oxford University Press
Original Article |
An Occupational Hygiene Investigation of Exposure to Acrylamide and the Role for Urinary S-Carboxyethyl-Cysteine (CEC) as a Biological Marker
1 Ciba Specialty Chemicals Water and Paper Treatment, Bradford, UK; 2 Health and Safety Laboratory, Sheffield, UK
* Author to whom correspondence should be addressed. Tel: +44(0)1274 417732; fax: +44(0)1274 417950; e-mail: peter.bull{at}cibasc.com
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Acrylamide has a range of toxicological hazards including neurotoxicity and reproductive toxicity; however, occupational risk management is driven by its genotoxic and carcinogenic potential (it is classified within the EU as a Category 2 carcinogen, R45 and Category 2 mutagen, R46). Since there is the potential for skin absorption and systemic toxicity, biological monitoring may be a useful aid for the assessment of exposure via inhalation, ingestion and dermal absorption. However, there are currently no biological monitoring guidance values (BMGVs). This study describes an extensive survey of potential workplace exposure to acrylamide at the Ciba (Bradford) site to gather data suitable for a BMGV. This manufacturing site is typical within the industry as a whole and includes a cross section of activities and tasks representative of acrylamide exposure. Acrylamide is used in the manufacture of polyacrylamide based products for applications in water treatment; oil and mineral extraction; paper, paint and textile processes. Workers (62 plus 6 controls) with varying potential exposures provided a total of 275 pre shift and 247 post-shift urine samples along with 260 personal air samples. A small non-exposed control group was similarly monitored. Urine samples were analysed for S-carboxyethyl-cysteine (CEC). Airborne, surface and glove samples were analysed for acrylamide. Inhalation exposures were well controlled with values consistently below one-tenth of the UK Workplace Exposure Limit. Engineering controls, personal protective equipment and work practice, all contributed to good control of occupational exposure. CEC was found in urine samples from both exposed workers and non-occupationally exposed controls. At the low levels of exposure found, smoking made a significant contribution to urinary CEC levels. Nevertheless a correlation between urinary CEC and airborne acrylamide was found. A mixed effects model incorporating inhalation concentrations of acrylamide and smoking habits could predict some of the variation in observed post-shift urine results but could be improved through the use of additional surface contamination data. However, the data does not suggest that dermal absorption was a major contributor to the systemic dose. Based on the 90th percentile of the data, inclusive of the effects of smoking and environmental factors, a value of 4 mmol mol–1 creatinine is proposed as a pragmatic BMGV associated with good occupational hygiene practice and control of workplace exposure. CEC in urine analysis has the utility for routine use as a means to estimate biological uptake where there is a potential for significant exposure or loss of workplace control.
Keywords: acrylamide • airborne exposure • biological benchmark guidance value (BMGV) • creatinine • dermal uptake • mercapturic acid • urinary metabolite
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Acrylamide is assigned a Workplace Exposure Limit (WEL) of 0.3 mg m–3 with a ‘skin’ notation in the UK and is classified within Europe as a Category 2 carcinogen and mutagen. The control of exposure is clearly important and biological monitoring for acrylamide may be a useful aid in the assessment of occupational exposure by all routes, but there are no biological monitoring guidance values (BMGVs) for acrylamide. The UK has established control based BMGVs for substances where a health-based value cannot be proposed. Such BMGVs are set at the 90th percentile of the occupational survey data where there is good exposure control. Once absorbed, acrylamide undergoes a complex metabolism. At least four different urinary metabolites have been identified in rats (Miller et al., 1982
), the most common being N-acetyl-S-(propionamide)-cysteine (APC), a mercapturic acid that accounts for 48% of the dose. A biological monitoring method based on the analysis of APC in urine involves acid hydrolysis of APC to produce S-carboxyethyl-cysteine (CEC). Acrylonitrile, present in acrylamide feedstock and cigarette smoke, is a known interference since hydrolysis also converts acrylonitrile mercapturic acid to CEC. Acrylamide is also metabolised into glycidamide and forms adducts with haemoglobin and exists as free acrylamide in plasma. All these forms of acrylamide are potential biomarkers of exposure (Calleman et al., 1994; Bergmark, 1997; Perez et al., 1999; Barber et al., 2001; Hagmar et al., 2001). The work described here is an occupational hygiene study to evaluate the non-invasive urinary biomarker CEC. This paper details a study of the major processes involving the manufacture, handling and polymerisation of acrylamide at Ciba (Bradford), a UK site typical within the industry comprising a cross section of activities and tasks representative of acrylamide exposure. Control of exposure is maintained via air extraction (LEV), good housekeeping, adequate supplies of suitable PPE, good personal hygiene, education and awareness via occupational health and hygiene of the professionals on the site. |
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Study design and sampling strategy
The aim of this occupational study was to assess the feasibility of establishing a biological benchmark guidance value (BMGV) based on a urine metabolite of acrylamide via a comprehensive survey of the occupational exposure during synthesis, polymerisation and product handling. The site had eight major processes (some at several locations) with potential exposure to acrylamide (Table 1). The sampling strategy aimed to recruit
10% of the potentially exposed workforce and get a minimum of six, and ideally ten, sets of data to assess exposure for each processes. A dataset comprised a pair of pre- and post-shift urine samples together with a personal air sample and any associated data on potential dermal absorption. Each worker provided a set of samples for at least two days. The numbers of workers on each process varied from a minimum of 1 to a maximum of 9 and, where possible, samples were collected from both smokers and non-smokers for each process. Where a process involved only one worker multiple sets of samples were collected. A group of office workers not occupationally exposed to acrylamide formed the ‘control’ group. Owing to the interference effects of smoking on the analysis, individuals monitored were split into two groups: smokers and non-smokers. The potential presence of other substances (e.g. acrylonitrile) that could yield CEC after hydrolysis was recorded at the time of sampling. Urine samples, collected at the start and end of each shift, were frozen below –20°C until analysed. Airborne exposure was monitored via personal air sampling throughout the shift (minimum 8 h). Swab samples were taken from the surfaces most frequently touched, and for selected shifts cotton gloves were worn throughout to estimate dermal exposure. Airborne, swab and glove samples were stored at <8°C prior to analysis. The study was approved by HSE's Research Ethics Committee.
|
Analysis of samples
Urine metabolite
The method was based on the method of Wu et al. (1993)
but modified to improve the derivatisation and sensitivity. The primary acrylamide metabolite, APC was a gift from Dr Wu, the Institute of Occupational Medicine, Chinese Academy of Preventative Medicine, Beijing, China. The method involved acid hydrolysis of the APC to produce CEC followed by derivatization with o-phthalaldehyde before the analysis by high performance liquid chromatography (HPLC) with fluorescence detection—excitation 340 nm, emission 450 nm. Mobile phase A was tetrahydrofuran:methanol:0.1 M sodium acetate (5:95:900), mobile phase B was 90% methanol. A gradient elution ran from 10 to 90% B over 15 min. The limit of detection (LOD) (3 times background noise) was 10 µmol l–1 as CEC and the method had a day-to-day coefficient of variation, determined from quality control data at 80 µmol l–1, of 12%.
Creatinine in urine
Creatinine concentrations were measured for urine samples according to the method of Jaffe (1886) using an automated method based on Bonsnes and Taussky (1945). Urine CEC concentrations were reported as mmol CEC per mole of creatinine. The method had a day-to-day coefficient of variation of 3.5%
Airborne monitoring and analysis
Sampling
Acrylamide was handled on the site as a solution, creating a potential for exposure via vaporization from the liquid surface or sublimation of dried deposits. The most suitable method of capturing airborne acrylamide was direct sampling onto silica gel filled glass tubes (SKC reference 226–10). The pump flow rate was set to 100 ml min–1 in order to achieve a LOD of 0.004 mg m–3 over the 8 h shift. Prior to monitoring, the personal sampling pump was calibrated and the actual flow rate recorded.
Analysis
The analysis using Ciba in-house method AC/OHHD/C10 (UKAS accredited) was based on MDHS and OSHA methodologies. Adsorbed acrylamide was desorbed from the tube packing with 1 ml methanol:water (50:50 v/v), analysed via HPLC with UV detection at 210 nm primary wavelength using a mobile phase of 0.1% phosphoric acid and acetonitrile. Calibration standards of acrylamide were prepared over the range of 4–200 µg ml–1. The LOD (lowest concentration with CV less than ±10%) was 0.2 µg ml–1and the coefficient of variation was 10% determined from the quality control data at a tube spike equivalent of 80 µg ml–1.
Swab testing
Sampling
Sterile cotton wool tipped swabs were immersed in 3 ml methanol:water (50:50 v/v) and used to wipe an area of 10 cm x 10 cm. The surface was wiped laterally with one side of the swab and then the swab was rotated through 180° to provide a fresh surface and the same area was wiped longitudinally. The swab tip was rinsed in the extraction solvent and the process repeated to ensure all potential contaminant is collected.
Analysis
Aliquots from the swab containers were transferred to vials for the analysis as per airborne samples. The detection limit was 0.06 µg for a 3 ml aliquot and the coefficient of variation was 10% based on quality control check solutions over the calibration range.
Cotton glove monitoring
The cotton gloves were worn to mimic bare skin as a means to measure the potential dermal exposure while performing activities where protective gloves (PPE) would not be used. Sampling was carried out over a single shift, concurrently with swab testing across all areas monitored. Gloves were not used for the control group.
Sampling
The gloves were labelled ‘L’ (left) and ‘R’ (right) and were worn throughout the time of personal monitoring, nominally 8 h.
Analysis
To minimize the volume of the solvent required, each glove had the wrist cuff removed prior to immersion in 50 ml methanol:water (50:50 v/v) contained in a screw top jar (6 oz). The acrylamide was extracted from the gloves via rotating the sealed jars for 16 h. Aliquots of the glove extracts were transferred to vials for the analysis as per airborne samples. The detection limit of 107.1 µg reflected the extraction volume and background interference (3 times background noise) from the blank gloves. The coefficient of variation was 10% based on quality control check solutions over the calibration range.
Statistical method
The statistical analysis of the results centred on establishing an association between the biological monitoring results and exposure. Two-way analysis of variance was employed to determine the confounding effects of smoking and exposed/control group. Spearman rank correlation coefficients were used to measure the tendency for the two quantities to increase together or be associated with one another. Linear mixed effects models have been used to relate individual urinary results to covariates such as personal airborne exposure levels and smoking habits (fixed effects) whilst taking account of repeated measurements on the same individual through random worker effects. Models were fitted to the un-transformed post-shift urinary data. Only the intercept term had an associated random effect i.e. no random effects were considered for smoking, airborne concentrations or any other covariate. Within worker errors were assumed to be independent and autocorrelation was not modeled. These models have been fitted using restricted maximum likelihood estimation using the software package S-Plus (Insightful Corporation 2001). Owing to the considerable number of samples below the LOD statistical analyses were repeated three times with biological monitoring values below the LOD set to zero, one half LOD and LOD, respectively.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Scope of sampling
A total of 68 volunteers (62 exposed and 6 controls) provided multiple samples. Out of them, 27 were smokers (24 exposed and 3 controls) and 41 were non-smokers (38 exposed and 3 controls). For each of the processes (the box and whisker plots in Fig. 1) a minimum of 7 and a maximum of 75 sets of samples (airborne personal and post-shift urine) were obtained. Most of the volunteers were monitored on several separate occasions in order to produce the required number of datasets for each category.
|
Airborne, glove and swab results
A total of 260 airborne samples were obtained with results ranging from <0.004 mg m–3 (detection limit) to 0.282 mg m–3 (all below the WEL of 0.3 mg m–3).
The distribution of airborne acrylamide results was highly skewed, where out of 240 results for exposed workers, over half (130) were <0.014 mg m–3. The mean airborne concentration was 0.028 mg m–3, the geometric mean 0.014 mg m–3 (1/20th of the WEL) and the geometric standard deviation was 3.34. The maximum concentration was 0.282 mg m–3. The mean airborne concentration for the control group by comparison was less than the detection limit of 0.004 mg m–3. Airborne monitoring results for other substances that could yield CEC were below detectable levels. The processes with the highest mean acrylamide air levels were acrylamide production and powder works areas (see Fig. 1a).
Surface swab samples were taken in each of the work areas (see Fig. 1b). Acrylamide swab results ranged from non-detected to 1295 µg. Mean acrylamide swab results in acrylamide production were over 10 times higher than in any other area. This location has both the highest mean airborne concentration and highest surface results. A Spearman rank correlation coefficient of 0.61 (P-value 0.036) indicates a strong association between airborne levels of acrylamide and the contamination of surfaces with acrylamide.
A total of 54 glove samples were taken (both left and right hand). The acrylamide contamination ranged from <107.1 µg (LOD) to 40.7 mg for the left hand and <107.1 µg to 37.9 mg for the right hand or from <214 µg to 79 mg for total hand contamination with a mean of 3 mg. The highest results were from acrylamide production work area as shown in Fig. 1c.
Urinary metabolite
There were 275 pre-shift and 247 post-shift urine samples provided. A summary of the post-shift urinary data classified according to control/exposed and smoker/non smoker is given in Table 2.
|
Over 43% of the results were below the LOD for the analytical method (10 µmol l–1 or
1 mmol mol–1 creatinine). The incidence of non-detected values differed markedly between smokers and non-smokers with only around 27% of samples from the smokers being below the LOD compared with 57% for non-smokers. This pattern is repeated with mean post-shift urine results: smokers recording mean post-shift urine results twice as that of non-smokers (1.92 mmol mol–1 versus 1.16 mmol mol–1). The exposed workforce exhibited slightly higher mean results than controls 1.64 mmol mol–1 in comparison to 1.40 mmol mol–1. A Spearman rank correlation coefficient (0.24) for pre- and post-shift urinary results is highly significant (P-value 0.0002). Post-shift urine results for each works area indicated that the highest CEC levels were from workers in acrylamide production and powder works areas in a similar pattern to the air monitoring results (Fig. 1d). In the UK, biological benchmark guidance values can be proposed based on the 90th percentile of the biological monitoring results obtained from a cross sectional study of workplaces with good control of exposure. The air monitoring, swab and glove samples, together with the occupational hygiene observations demonstrate a good control of exposure in this study and all the data obtained can be included in the calculation of this type of a guidance value. There are 227 post-shift urinary metabolite results for the exposed population and the 90th percentile is 3.4 mmol mol–1 creatinine. For the exposed smoking workforce (103 values) the 90th percentile is slightly higher, 3.9 mmol mol–1 creatinine. In one urine sample from a non-occupationally exposed control worker, the value exceeds 3.9 mmol mol–1 creatinine. This individual has been identified as a smoker.
Association between exposure and biological monitoring results
The analysis of the relationships between exposure, smoking and post-shift urinary results have been repeated substituting 3 different values for results less than the LOD (0, half LOD and LOD). The three analyses gave consistent conclusions and the results are only presented in the case of half LOD.
A two-way analysis of variance accounts for the possible effect of smoking when trying to ascertain the influence of exposure/control grouping. These results are shown in Fig. 2. Smoking was found to be a contributing factor in determining post-shift urine results (P-value < 0.001). At the low levels of exposure in this study, the higher mean urine results found in the exposed work force were not statistically significant compared with the controls. The exposed workforce as a whole in this study does not have significantly higher average urinary results. Through mixed effects models a more detailed understanding of the dependency of the biological monitoring results upon airborne, surface and dermal exposures can be attained.
|
A linear mixed effect model for the post-shift urine results incorporating smoking and airborne concentrations as fixed effects is given in Table 3.
|
This model predicts that smoking increases urinary results by
0.8 mmol mol–1, whilst an airborne concentration of 1 mg m–3 increases post-shift urine results by 8 mmol mol–1. Pre-shift urine values are not significant when added to this model. The within-worker variance component (1.424) is considerably larger than the between-worker variance (0.633). This may reflect day-to-day variation in workplace exposures, variation in dietary acrylamide or (for smokers) day-to-day variation in the number of cigarettes consumed. Mixed effects models have been fitted where the exposure group is added as a predictor. This analysis finds that only the two higher exposure groups in the manufacturing area are significantly different once smoking and inhalation have been allowed for. A possible explanation is that in these manufacturing groups, higher surface residues of acrylamide result in some uptake through dermal absorption. Table 4 shows the effect of replacing the exposure group with the (log) mean surface swab result (for each group). Adding the work area as a predictor to this model reveals no further significant results i.e. the higher urinary results in manufacturing areas are explained by the higher surface contamination in these areas.
|
It is possible to use models (1) and (2) to predict biological monitoring results for workers with a given level of exposure. Using the model (1) incorporating only airborne exposure and smoking habits, the predicted post-shift urinary result for smokers exposed to the maximum exposure level for acrylamide (0.30 mg m–3) is 4.26 mmol mol–1 creatinine, broadly in line with the suggested biological guidance value.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
At the start of this study the options for the biological monitoring of exposure to acrylamide were collection of blood samples for analysis of haemoglobin adducts (Bergmark et al., 1993
; Calleman et al., 1994) or collection of urine samples for the analysis of APC, the major acrylamide mercapturic acid (Calleman et al., 1994). The invasive nature of blood sampling and the complex analytical methods required for analysis of haemoglobin adducts were (and are) likely to limit the routine application of this approach. Non-invasive urine collection and a simpler analytical method for APC were potentially more useful. However, although it was known that this approach was capable of detecting high levels of exposure (Calleman et al., 1994) it was not known if the known interferences from cigarette smoke would undermine the utility of the method at low levels of exposure. The work reported here shows that in a company with a good control of occupational exposure to acrylamide in a wide range of processes, there is a weak correlation between exposure via all routes and the level of a CEC after hydrolysis of APC in urine has been established. A statistical mixed effect model incorporating only personal airborne exposure data and smoking habits predicted post-shift urinary levels of 4.26 mmol mol–1 creatinine for smokers and 3.46 mmol mol–1 creatinine for non-smokers exposed to acrylamide at the WEL (0.3 mg m–3). Whilst a statistically significant association was found between raised surface contaminations and higher urinary metabolites this did not represent a major contribution to systemic exposure compared with smoking or airborne exposure in this study.
Calculations show that smoking contributes 1 mmol CEC mol–1 creatinine to the excretion of CEC. Acrylamide and acrylonitrile are present in cigarette smoke and could result in elevated urinary metabolite levels in the ‘smokers’ group. At the low levels of exposure monitored in this study, the mean urinary metabolite levels in the smoking control group exceeded that of the non-smoking occupationally exposed workers. However, there are only three subjects in the smoking control group. So this finding should be interpreted with caution. The presence of the urinary CEC in the non-smoking control group suggests an additional source of CEC such as acrylamide in cooked food (WHO, 2002), which may account for >50% of the CEC measured in some cohorts. Other substances used on this site that could yield CEC on hydrolysis (acrylonitrile, substituted acrylamides, acrylic acid and acrylates) were below detectable levels, and did not contribute significantly to the CEC levels found. This study was limited by the minimal data collected on smoking habits. With hindsight, because occupational exposure to acrylamide was low in this study, more detailed information on smoking habits would have been useful and may have contributed to a refined understanding of the relationship between workplace exposures to acrylamide, smoking habits and urinary metabolites.
A recent paper by Boettcher et al. (2005) describes a method for N-acetyl-S-(carbamoylethyl)-1-cysteine and N-(R,S)-acetyl-S-(2-carbamoyl hydroxyethyl-1-cysteine (GAMA), the mercapturic acids of acrylamide and glycidamide formed by the metabolic oxidation of acrylamide. Although urinary concentrations of the glycidamide mercapturic acid are much lower than those of the acrylamide mercapturic acid, because glycidamide is thought to be the ultimate carcinogen, GAMA may be a better indicator of toxicity. This metabolite would also be less susceptible to interference from other acrylates such as acrylonitrile in the cigarette smoke. However, due to the presence of acrylamide in cigarette smoke even this metabolite may be limited in its ability to detect very low levels of occupational exposure to acrylamide.
Although background levels of CEC in urine limit its ability to detect very low levels of acrylamide exposure, it can be used to assess acrylamide exposures below current occupational limits.
A value of 3.4 mmol mol–1 creatinine (90th percentile) of CEC in urine samples collected post-shift was calculated for the exposed workers in this study. For the exposed smokers in the workforce the value was slightly higher at 3.9 mmol mol–1 creatinine.
Based on this study we propose 4 mmol CEC mol–1 creatinine, in urine samples collected at the end of shift, as a pragmatic BMGV associated with good occupational hygiene practice.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The authors would like to thank Phillip Kelley (Occupational Health Centre Manager), Michael Cain (Occupational Hygienist), Michelle Briggs (Assistant Occupational Hygienist), Nikki Rowntree (Occupational Health Administrator), Ciba, Bradford for their help with various aspects of the study.
Received October 14, 2004; in final form May 20, 2005
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Barber DS, Hunt J, LoPachin RM et al. (2001) Determination of acrylamide and glycidamide in rat plasma by reversed-phase high performance liquid chromatography. J Chromatogr B Biomed Sci Appl; 758: 289–93.[Medline]
Bergmark E, Calleman CJ, He F et al. (1993) Determination of haemoglobin adducts in humans occupationally exposed to acrylamide. Toxicol Appl Pharmacol; 120: 45–54.
Bergmark E. (1997) Hemoglobin adducts of acrylamide and acrylonitrile in laboratory workers, smokers and nonsmokers. Chem Res Toxicol; 10: 78–84.
Bonsnes RW, Taussky HH (1945) On the colorimetric determination of creatine by the Jaffe reaction. J Biol Chem; 158: 581–7
Boettcher MI, Schettgen T, Kutting B, Pischetsrieder M, Angerer J (2005) Mercapturic acids of acrylamide and glycidamide as biomarkers of the internal exposure to acrylamide in the general population. Mutation Research; 580: 167–176.[Web of Science][Medline]
Calleman CJ, Wu Y, He F et al. (1994) Relationships between biomarkers of exposure and neurological effects in a group of workers exposed to acrylamide Toxicol Appl Pharmacol; 126: 361–71[CrossRef][Web of Science][Medline]
Hagmar L, Tornqvist M, Nordander C et al. (2001) Health effects of occupational exposure to acrylamide using hemoglobin adducts as biomarkers of internal dose. Scand J Work Environ Health; 27: 219–26.[Web of Science][Medline]
Jaffe M (1886) Uber den niederschlag, welchen pikriksaure in normalen harn erzeugt und uber eine neue reaction des kreatinins. Z Physiol Chem; 10: 391.
Miller MJ, Carter DE Sipes LG (1982) Pharmacokinetics of acrylamide in fisher 334 rats. Toxicol Appl Pharmacol; 63: 36–64.
Perez HL, Cheong HK, Yang JS et al. (1999) Simultaneous analysis of hemoglobin adducts of acrylamide and glycidamide by gas chromatography-mass spectrometry. Anal Biochem; 274: 59–68.[CrossRef][Web of Science][Medline]
WHO. (2002) Health implications of acrylamide in food. Joint FAO/WHO consultation, Geneva, Switzerland, 25–27 June 2002. World Health Organization. ISBN 02 4 156218 8. Available at: http://www.who.int/foodsafety/publications/chem/acrylamide_june2002/en/
Wu Y-Q, Yu A-R, Tang, X-Y et al. (1993) Determination of acrylamide metabolite mercapturic acid by HPLC. Biomed Environ Sci; 6: 273–80.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||