Article Text
Abstract
Objectives Toluene diisocyanate (TDI) is used in the manufacturing process of polyurethane (PU) foams and is a potent inducer of occupational asthma. The objective of this study was to evaluate the correlation between the exposure to total TDI (2,4- and 2,6-TDI) in air and the corresponding biomarker concentration of total TDA (2,4- and 2,6-TDA) in hydrolysed urine. The aim was also to propose an appropriate biological exposure limit for total TDA in urine.
Methods 9 workers from two production lines in a PU foam producing plant were studied. Personal exposure to TDI during four representative production shifts was monitored by an active air sampling method (filter impregnated with 1-(2-methoxyphenyl)piperazine) and quantified by high-performance liquid chromatography and diode array detection (NIOSH n° 2535, 5521). In parallel, pre-shift and post-shift urinary samples were collected from the exposed workers, and TDA concentrations were determined by gas chromatography–mass spectrometry after alkaline hydrolysis. All samples were collected on four measuring days: two Fridays (end of workweek) and two Mondays (start of workweek) separated by a weekend without exposure.
Results Strong correlations between the personal air concentrations of total TDI and the corresponding biomarker levels of total TDA in urine (r=0.816) were observed. An increase of 18.12 μg TDA/l (post-shift minus pre-shift concentration) corresponds to an exposure of 5 ppb (37 μg/m3, the current American Conference of Governmental Industrial Hygienists threshold limit value) during the shift.
Conclusions The increase in TDA during the shift is a suitable biomarker for exposure to TDI during the same shift. Further research is needed to evaluate the use of start of week or end of week post-shift TDA in urine as biomarker since TDA was found to accumulate during the working week and thus the moment of sampling will clearly influence the result.
- TDI
- exposure assessment
- TDA
- biomonitoring
- biological monitoring
- hygiene/occupational hygiene
- toxicology
Statistics from Altmetric.com
What this paper adds
TDA concentrations in blood, plasma and urine are much investigated as biomarkers for toluene diisocyanate (TDI) exposure, but in most studies, a thorough characterisation of external exposure is unavailable.
None of these studies take a possible accumulation of TDA during the working week into account in their biological monitoring strategy.
This study was carried out to evaluate the measurement of TDA in urine pre- and post-shift and pre- and post-weekend.
Our results show that the increase in urinary TDA post-shift minus pre-shift correlates well with exposure to TDI during the shift and that post-shift TDA accumulates during the workweek.
Through regression analysis, a biological exposure limit of 18.12 μg/l for post-shift minus pre-shift TDA in urine corresponding to the current threshold limit value (5 ppb or 37 μg/m3 in environmental air) could be calculated.
More research is needed to evaluate start of week or end of week post-shift urine as a more practical tool to discriminate between workers exposed to TDI concentrations below or above the current threshold limit value.
Introduction
Toluene diisocyanate (TDI) is an important industrial chemical used in the production process of flexible and rigid polyurethane (PU) foams for furniture, bedding, construction, automotive and many other industries. The worldwide production of PU in 2007 was around 15.9 million tons and required a total TDI consumption of 1.9 million tons.1 The TDI used in industry is mainly an isomeric mixture of 2,4- and 2,6-TDI in a ratio of 80:20 or 65:35. Diisocyanates are extremely reactive compounds. During the PU foam production process, TDI violently reacts with polyols in presence of amines and water. In the human body of exposed workers, TDI covalently binds with carboxyl and amino groups of endogenous macromolecules such as haemoglobin and plasma proteins. Adducts to plasma albumin among workers exposed to TDI have been demonstrated.2 Skin irritation, irritative and allergic contact dermatitis,3–5 and respiratory impairments, such as occupational asthma, airway irritation and asthma-like symptoms (cough, wheezing and dyspnoea), are well known as occupational health problems for workers using TDI.6–9 Occupational asthma is the main problem with a prevalence up to 10% among workers exposed to TDI concentrations even below 15 ppb.10
Therefore, it is essential to monitor the occupational exposure to TDI and to keep it as low as reasonably achievable. Exposure surveillance is mostly performed by air monitoring. However, biological monitoring can hold several advantages. The sampling is more convenient, and biomarkers better reflect the absorbed dose that takes into account factors such as lung ventilation, genetic and acquired metabolic characteristics, the effects of personal protective equipment and alternative routes of exposure, such as skin exposure. Biomarker levels, rather than air levels, may better correlate with systemic adverse health effects.
Biological monitoring of TDI isomers is based on the assessment of the corresponding amines 2,4- and 2,6-toluenediamine (2,4- and 2,6-TDA) in hydrolysed biological fluids, such as urine or plasma.11–16 The hydrolysis method itself is known to strongly influence the amount of amines that will be released17 and complicates the comparison of data between different TDI exposure studies.
An indication of a two-phase elimination pattern in urine was found in a study of volunteers after 7.5 h test chamber exposure to a concentration of 40 μg/m3 total TDI. Half-lifes ranged from 1.6 to 1.9 h in the first phase to 5 h in the second phase and in total 8%–18% of the 120 μg inhaled TDI was excreted as TDA.18 These short half-lifes are in contrast with results from a study of chronically exposed workers, where mean urinary half-lifes of 5.8–11 days for 2,4-TDA and 6.4–9.3 days for 2,6-TDA were found. The hypothesis is that the first phase is directly related to the recent exposure, and the second and slower phase is related to the release of TDA from plasma albumin adducts in the body.19 Under different exposure conditions, TDA levels may reflect recent exposure during the past shift or either more long-term exposure from the past weeks.11 16
In ongoing attempts to reduce the health effects caused by TDI, airborne exposure limits are low. In 1983, the American Conference of Governmental Industrial Hygienists (ACGIH) has set a 0.005 ppm (37 μg/m3) threshold limit value (TLV, 8 h time weighted average) and 0.02 ppm (140 μg/m3) short time exposure limit value (15 min timed weighted average) for worker exposure to TDI in air, but until now, a corresponding biological exposure index for TDI has not been established.20 In 2006, the ACGIH proposed a further lowering of the TLV to 0.001 ppm (and to 0.003 ppm for the short time exposure limit), and since 2009, the TDI TLV is on the ACGIH ‘notice of intended change (NIC)’ list. The Health & Safety executive (HSE) in the UK introduced an end of shift biological monitoring guidance value (BMGV) for isocyanates of 1 μmol isocyanate-derived diamine per mole creatinine (which corresponds to 1.1 μg total TDA/g creatinine). This BMGV is not health based but is derived from the 90th percentile of biological monitoring data.21 The UK workplace exposure limit (WEL) for all isocyanates is 20 μg NCO-groups/m3 22 and corresponds to 41 μg TDI/m3.23 The results found in occupational studies of exposure to TDI during manufacture of PU foams14 24 show a good correlation between airborne TDI and TDA in urine and also suggest that after exposure to TDI at the UK WEL, urinary TDA will be in the range of 10–20 μmol total TDA/mol creatinine. So urinary TDA concentrations below the BMGV of 1 μmol/mol creatinine are indicative of exposure below the WEL, the current ACGIH TLV and the NIC.25 The Finnish Institute of Occupational Health (FIOH) defines a same action limit of 1 μmol total TDA/mol creatinine (end of shift) and advises to take paired samples post-shift end of workweek and pre-shift start of workweek.26 There is no corresponding external TDI exposure value for the HSE BMGV and FIOH action limit.
This study aimed at exploring the value of urinary TDA biomonitoring in assessing workers' total exposure to TDI, to propose an ideal sampling strategy (end of shift, end of workweek, paired sampling…) and a biological limit value appropriate for the sampling strategy and the established TLV.
Methods
Plant, production process, workers and sampling strategy
The study included nine workers from a flexible PU foam production plant using an 80:20 mixture of 2,4- and 2,6-TDI. There were two similar production lines, an older line 1 and a new line 2, with semi-enclosed tunnel systems. The foam slab stock production occurs through a process in which a mixture is continuously dispensed onto a carrier paper on a moving conveyer, which then draws the expanding foam into a ventilated tunnel. At the end of the tunnel, when the foam has stabilised and risen to the right height, the foam blocks are sawn into their appropriate lengths. Hereafter, the blocks are stored into a racking system for at least 24 h to cool down and to complete the curing process, before further processing.
Foam production generally took place for about 65% of the working day, and when not producing PU foam, workers were engaged in a variety of activities around the plant, such as preparing activities and cleaning the site. On average, there were four start-ups of different types of foaming mixes per shift.
All workers wore appropriate work clothing and safety shoes during the shifts. Respiratory protection was inconsistently used, and gloves were not consistently worn by all workers when handling hot blocks of uncured foam.
Nine workers, all male workers, were included in the study. Three of them (workers 1–3) were the same on the four measuring days and worked at line 2. The other three workers (workers 4–6), working at line 1, were only available during the first three shifts and were replaced by three others (workers 7–9) during the fourth and final measuring day.
Sampling took place on four representative shifts: two subsequent Friday and Monday shifts, separated by a weekend without exposure.
Sampling and analysis of TDI in air
On each of the four sampling days, personal air samples were collected during the whole foam production period, using a personal sampling pump (GilAir® -3, operating at a calibrated flow rate of 1 l/min) and double-glass fibre filters impregnated with 1-(2-methoxyphenyl) piperazine (2-MP method) in 25 mm filter cassettes. The filters were always attached in the breathing zone. Sampling times ranged from 189 to 372 min with an arithmetic mean (AM) of 313 min. Exposure to TDI was calculated for the sampling period only.
After each shift, air samples were immediately sent to the laboratory with refrigerated transport, where they were stored at 4°C until analysis. Air samples were analysed for the 2,4- and 2,6-TDI isomers, in the Laboratory of Industrial Toxicology (AZ Groeninge, Kortrijk, Belgium), through high-performance liquid chromatography and diode array detection (NIOSH n° 2535, 5521). The limit of detection was 0.03 μg/m3 and the limit of quantification (LOQ) was 0.17 μg/m3 for a sampling volume of 120 l air.
Sampling and analysis of TDA in urine
Urine samples were collected in polypropylene recipient cups immediately before the beginning and immediately after the end of the shift. Hereafter, 10 ml urine was transferred into a new recipient containing 0.5 ml HCl 6 M. After each shift, urine samples were immediately sent to the laboratory with refrigerated transport, where they were stored at 4°C until analysis. The analysis of the 2,4- and 2,6-TDA isomers was performed in the Laboratory of Industrial and Environmental Toxicology (Université Catholique de Louvain, Brussels, Belgium). Briefly, the urinary samples were hydrolysed overnight in 0.5 M of sodium hydroxide to release the TDI-related diamines 2,4- and 2,6-TDA. This extraction procedure was confirmed to be adequate to completely hydrolyse 2,4- and 2,6-diacetylated TDA previously synthesised in the laboratory (unpublished data). The diamines were extracted with toluene, and after derivatization with heptafluorobutyric anhydride, the derivatives were quantified by gas chromatography and mass spectrometry. To the best of our knowledge, external quality assessment schemes do not exist for 2,4- and 2,6-TDA. However, using this validated method, the laboratory has obtained successful results in external quality assessment schemes organised by the Institute for Occupational, Environmental and Social Medicine of the University of Erlangen, Germany (G-EQUAS program) for 4,4'-methylene dianiline, metabolite of 4,4'-methylenediphenyl diisocyanate. The LOQ is 1 μg TDA/l urine for this method, and the limit of detection for this method has not been assessed accurately.
Statistics
The analytical results of 2,4-TDI, 2,6-TDI, 2,4-TDA and 2,6-TDA were registered in an MS Excel 2003 file (Microsoft, Redmond, Washington, USA). TDA results below the LOQ were treated as equal to half the LOQ (0.5 μg/l). Total TDI and total TDA concentrations were calculated as the sum of the 2,4- and 2,6-isomers. Statistical analyses were performed using SPSS V.17 (SPSS Inc., Chicago, IL, USA).
As ambient and biological parameters showed skewed distributions, non-parametric tests were used on non-transformed values or parametric tests were used on values transformed with the natural logarithm (ln). After ln-transformation, normality was achieved.
The difference between end and start of workweek post-shift TDA was investigated with the Mann–Whitney U test on non-transformed values. The effect of the moment of sampling was confirmed in a covariance analysis with TDA (ln-transformed post-shift values) as dependent variable, TDI (ln-transformed values) as covariate and moment of sampling (Monday 1/Friday 2) as fixed factor.
The difference between pre- and post-shift samples was investigated with a paired samples t test on ln-transformed values.
The correlation between the biomarkers and the personal air exposure data was analysed through Pearson's correlation between ln-transformed values.
Statistical significance for all tests was set at p<0.05 (two tailed)
Biological exposure limit (BEL) values were calculated by means of linear regression analysis between ln-transformed values of air and urinary data. These limits are estimations of the biomarker levels that correspond to an average exposure to TDI at the current TLV or NIC.
Results
External and internal exposure at the end and start of the workweek
A total of 23 of the 24 air samples were analysed since one worker lost his sampling filter during the measurement. Two of the 24 pre-shift urinary samples could not be analysed due to problems with sample collection. Finally, 21 air and 21 pre- and post-shift sample pairs were retained.
All 2,4- and 2,6-TDI results were above the LOQ (0.17 μg/m3). All post-shift 2,4- and 2,6-TDA results were above the LOQ (1 μg/l). Three pre-shift samples (2/21 2,4-TDA results and 1/21 2,6-TDA result) were below the LOQ. External and internal sampling results for total TDI and total TDA are displayed in table 1. The external exposure is not significantly different between Friday (AM=39.45 μg/m3) and Monday (AM=21.94 μg/m3), and the internal exposure pre-shift is not significantly different between end (AM=10.95 μg/l) and start of workweek (AM=6.17 μg/l) (p=0.159 and p=0.057, respectively). In contrast, the internal exposure post-shift is significantly higher at the end of the workweek (p=0.009) (AM Friday=37.46 μg/l, AM Monday=13.47 μg/l). After adjusting for TDI exposure (lnTDI) in a covariance analysis, post-shift TDA (lnTDA) values still significantly differ between end (11/21) and start (10/21) of workweek (F=4.988, p=0.038).
Paired t test for pre-shift and post-shift total TDA
A paired samples t test with ln-transformed values of pre-shift (mean=1.96) and post-shift (mean=2.89) concentrations (n=21) learned that the total TDA concentration was higher in post-shift samples (p<0.001). This increase is likely to be explained by the exposure during the shift.
Relation between external and internal exposure
The increase in total TDA (difference post-shift minus pre-shift) is the accumulation of the internal dose that can be explained by exposure to total TDI during the shift. As total TDI and increase in total TDA showed skewed distributions, their ln-transformed values are used when performing least square regression analysis. The result of the analysis is shown in figure 1 (r=0.816, p<0.001). The regression line with equation lnTDA=0.922 lnTDI−0.432 can be used to calculate a BEL corresponding to the TLV. For a TDI concentration of 37 μg/m3, we get a mean value of 18.12 μg/l for the difference of post-shift minus pre-shift results. The 95% CI for this mean value (n=21) is 12.84 to 25.56 μg/l, while the 95% CI for individual values has a range of 4.92 to 66.78 μg/l. The same equation can be used to compute a corresponding BEL for exposure to 7 μg/m3 (the current NIC) and yields 3.90 μg/l (1.00 to 15.12 μg/l 95% CI for individual values or 2.39 to 6.38 μg/l 95% CI for the mean).
Sensitivity and specificity results
Using the 37 μg/m3 TDI and 18.12 μg/l TDA (difference post-shift minus pre-shift TDA) thresholds, 7/21 TDI results exceed 37 μg/m3 and 6/21 TDA results exceed 18.12 μg/l. These results are displayed in table 2. The sensitivity of the test is 71.4% and the specificity is 92.9%.
Using the 7 μg/m3 TDI and 3.90 μg/l TDA (difference post-shift minus pre-shift TDA) thresholds, 18 and 17 exceedances are retrieved, respectively (table 3). The specificity of this test is 33.3% and the sensitivity is 83.3%. Figure 2 shows the 95% CI of the weighted mean lnTDA (post-shift minus pre-shift TDA) for exposure under and above the current TLV of 37 μg/m3 (figure 2A) and NIC of 7 μg/m3, respectively (figure 2B). Post-shift minus pre-shift TDA can discriminate exposure below (n=14) and above (n=7) the TLV and below (n=3) and above (n=18) the NIC, at group level, independent of the day of sampling at the start or end of the workweek. The test towards 3.90 μg/l TDA, however, has insufficient specificity and will lead to false positives (table 3) in the range of exposure below the NIC.
Half-life calculation
Over weekend, the TDA concentration drops from 27.02 μg/l (GM) Friday post-shift to 4.99 μg/l (GM) Monday pre-shift (table 1) in a time range of 64 h. Exponential regression yields an equation of TDA(t)=27.02e−0.0264t. The combined half-life can be calculated as 26.3 h or 1.1 day.
Discussion
Through assessment of exposure to total TDI and pre- and post-shift biomonitoring of TDA, we find that an average concentration difference (post-shift minus pre-shift) of 18.12 μg/l corresponds to an average occupational exposure of 37 μg/m3. Hence, this value is our proposal for a BEL.
The exposure range we found is representative for TDI exposures found in other PU production plants.27 28 As a similar alkaline hydrolysis protocol was followed in our study, indicative comparisons for TDA levels can also be made: the urinary post-shift TDA levels observed in the present study are in the same range as levels found in other studies.27 28 The study of Tinnerberg and Mattsson28 compared sampling results from 2000 to 2005 in a similar PU plant in Sweden. Our post-shift TDA values are more like the values of the Swedish plant in 2000, before implementation of several preventive actions (median of the present study for 2,4-TDA is 4.6 μg/l and median of the Swedish study for 2,4-TDA is 8.1 μg/l in 2000 and 0.8 μg/l in 2005). In the meanwhile, several preventive actions are taken in the plant we studied in order to achieve a similar reduction in exposure.
In table 4, a comparison of our study with Sennbro et al27 is shown. The values of 2,4-TDA in post-shift urine are similar, but the values of 2,6-TDA are higher in our study. The isomeric mixtures in both studies might have been different and may account partially for this difference. Higher total TDA values in the present study are to be expected since the mean TDI exposure in our study was also higher (median 24.4 μg/m3 in our study vs 4 μg/m3 total TDI in the study of Sennbro et al).
When we compare the correlation coefficients of least square regression analysis results between non-transformed post-shift TDA levels and non-transformed exposure levels of TDI of our study (n=21) and Sakai et al15 (n=18), we achieve comparable results for the isomers: 0.917 (present study) and 0.905 (Sakai et al) for 2,6-TDA (micrograms per gram creatinine) versus 2,6-TDI and 0.842 (present study) and 0.639 (Sakai et al) for 2,4-TDA (micrograms per gram creatinine) versus 2,4-TDI, respectively. In both studies, the relation between air and urinary post-shift data is very strong.
From the corresponding regression equation TDA=0.941 TDI−3.222 (total TDA post-shift in micrograms per litre, non-transformed values), a BEL of 37.5 μg/l (n=21, Monday and Friday post-shift samples) can be calculated, which holds the mean between other values found in literature. In comparison with literature, Sennbro et al27 gets a regression equation TDA=2.2 TDI+0.1 and a less stringent BEL value of 79 μg/l (n=200, 95% CI 72 to 86 μg/l). But the exposure in the latter study was generally lower compared to our study. In the past, a more stringent BEL of 18 μg/g creatinine after exposure to 38 μg/m3 (n=9) has been found.14 From our data (total TDA post-shift in micrograms per gram creatinine, non-transformed values), the equation of TDA=0.547 TDI−1.636 yields a BEL of 19.15 μg/g creatinine after exposure to 38 μg/m3 (n=21). In contrast with other cited studies,27 28 Maitre et al14 used hydrochloric acid hydrolysis to release TDA. As alkaline hydrolysis has been shown to release twice as much TDA,17 it is rather strange that we do not see this difference reflected in the deducted BEL. In the determination of a BEL, however, the latter study as well as our study have the advantage that the occupational exposure was more in the range around the TLV.
None of the previous studies thoroughly studied the effect of the sampling day in relation to the workshift pattern. We find a significant difference between the post-shift urinary TDA concentrations on Monday (start of workweek) and Friday (end of workweek), although we do not find a significant difference between the air exposure values on Monday and Friday. Sennbro et al27 found no systematic differences between the biomarker levels in two urinary samples taken on two different days in two following weeks, but the study does not describe the day of measurement in the working week. A possible explanation for the difference we observed is that there is an accumulation during the working week, resulting in a higher post-shift value on Friday. Our estimation of a combined half-life of 1.1 day is in favour of the explanation that TDA might accumulate in the human body. It argues against the HSE advice to take post-shift urinary samples—anywhere in the workweek—as previous days with similar TDI exposure would not influence the results.21 Our half-life result is intermediary compared to previous results of 1.6–1.9 h in the first phase to 5 h in the second phase18 and half-lifes of 5.8–11 days for 2,4-TDA and 6.4–9.3 days for 2,6-TDA.19 While the first study deducted half-lifes from test chamber exposure results to TDI, the second study used an occupational setting. Lind et al19 calculated the half-lifes from data from four workers exposed to 10–120 μg TDI/m3, before and after a 5 days holiday. The exposure range was similar in our study (4–142 μg TDI/m3), but data from nine workers before and after a weekend without exposure were used in the calculation. Maybe the release from adducts in the body had a prolonging effect on the half-life in the study of Lind compared to our study.
With longer half-lifes, accumulation during the workweek occurs, but other studies did not consider this effect in their sampling strategy.
Our study has the advantage of using paired samples. We found a difference between pre- and post-shift paired urinary samples: the post-shift values were significantly higher. This can be explained by an increase during the shift because of the TDI exposure during the shift. Thus, the difference in TDA concentration of post- minus pre-shift urine is a relevant parameter to assess recent exposure to TDI, independent of the day of sampling or a possible accumulation effect during the workweek.
Although accumulation of TDI metabolites in chronically exposed workers during the workweek seems possible with half-lifes longer than a day, no recommendations for a suitable biomonitoring strategy are found. Only FIOH advises to take end of workweek and start of workweek paired samples but still sets a biomonitoring action limit for single post-shift urinary samples for exposed workers at 1 μmol total TDA/mol creatinine (1.1 μg total TDA/g creatinine).26 Following this guideline, all our post-shift TDA samples (2.4–112.9 μg TDA/g creatinine) would be above the action limit, while 60% of the exposure data were clearly below the TLV.
Using the proposed BEL of 18.12 μg TDA/l (post-shift minus pre-shift values), 71% of our data achieves compliance. The sensitivity (71.4%) and specificity (92.9%) of the test are reasonably high and indicate that on a group level, post-shift minus pre-shift TDA can be used to determine whether exposure is likely to be above or beneath the established TLV (figure 2). More research is needed to evaluate the usefulness of testing towards a BEL of 3.90 μg TDA/l (post-shift minus pre-shift values) in the exposure range below the NIC.
Despite the small sample size, this study supports evidence for a strong correlation between TDI in air and the increase in TDA in urine during the shift, independent of the day of sampling during the workweek. Moreover, it shows that the proposed value of 18.12 μg TDA/l for this increase is a suitable BEL at group level. More research is needed to evaluate start of workweek or end of workweek post-shift sampling of TDA and to determine appropriate BELs corresponding to the TLV and NIC for this more conventional way of sampling.
Acknowledgments
The authors wish to thank Bruno Ceunis (Provikmo, Bruges, Belgium) for excellent technical assistance. Bart Lepla and Frank Martens (AZ Groeninge, Kortrijk, Belgium) as well as Vincent Haufroid (Université Catholique de Louvain, Brussels, Belgium) are gratefully acknowledged for their flexible aid with the analyses of the different samples. Hendrik Veulemans (Katholieke Universiteit Leuven, Leuven, Belgium) and Steven Van den Eede (Provikmo, Bruges, Belgium) are acknowledged for critical reading of the manuscript and valuable statistical advice.
References
Footnotes
Funding TG and SD were financially supported by Provikmo to conduct this study. The workers agreed to participate for free in the study.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.