|Home | About | Journals | Submit | Contact Us | Français|
When working with hot mix asphalt, road pavers are exposed to polycyclic aromatic hydrocarbons (PAHs) through the inhalation of vapors and particulate matter (PM) and through dermal contact with PM and contaminated surfaces. Several PAHs with four to six rings are potent carcinogens which reside in these particulate emissions. Since urinary biomarkers of large PAHs are rarely detectable in asphalt workers, attention has focused upon urinary levels of the more volatile and abundant two-ring and three-ring PAHs as potential biomarkers of PAH exposure. Here, we compare levels of particulate polycyclic aromatic compounds (P-PACs, a group of aromatic hydrocarbons containing PAHs and heterocyclic compounds with four or more rings) in air and dermal patch samples from 20 road pavers to the corresponding urinary levels of naphthalene (U-Nap) (two rings), phenanthrene (U-Phe) (three rings), monohydroxylated metabolites of naphthalene (OH-Nap) and phenanthrene (OH-Phe), and 1-hydroxypyrene (OH-Pyr) (four rings), the most widely used biomarker of PAH exposure. For each worker, daily breathing-zone air (n=55) and dermal patch samples (n=56) were collected on three consecutive workdays along with postshift, bedtime, and morning urine samples (n=149). Measured levels of P-PACs and the urinary analytes were used to statistically model exposure–biomarker relationships while controlling for urinary creatinine, smoking status, age, body mass index, and the timing of urine sampling. Levels of OH-Phe in urine collected postshift, at bedtime, and the following morning were all significantly associated with levels of P-PACs in air and dermal patch samples. For U-Nap, U-Phe, and OH-Pyr, both air and dermal patch measurements of P-PACs were significant predictors of postshift urine levels, and dermal patch measurements were significant predictors of bedtime urine levels (all three analytes) and morning urine levels (U-Nap and OH-Pyr only). Significant effects of creatinine concentration were observed for all analytes, and modest effects of smoking status and body mass index were observed for U-Phe and OH-Pyr, respectively. Levels of OH-Nap were not associated with P-PAC measurements in air or dermal patch samples but were significantly affected by smoking status, age, day of sample collection, and urinary creatinine. We conclude that U-Nap, U-Phe, OH-Phe, and OH-Pyr can be used as biomarkers of exposure to particulate asphalt emissions, with OH-Phe being the most promising candidate. Indications that levels of U-Nap, U-Phe, and OH-Pyr were significantly associated with dermal patch measurements well into the evening after a given work shift, combined with the small ratios of within-person variance components to between-person variance components at bedtime, suggest that bedtime measurements may be useful for investigating dermal PAH exposures.
Asphalt (also known as bitumen) is a highly viscous petroleum product that, when combined with additional materials (e.g. coal tar and gravel) to form hot mix asphalt, is commonly used for paving roads. While the chemical composition of asphalt varies according to the source of petroleum and other ingredients, constituents of asphalt include a complex array of hydrocarbons (NIOSH, 2000). Of these, studies of asphalt-exposed workers have focused primarily upon the polycyclic aromatic hydrocarbons (PAHs), a group of fused-ring aromatic compounds, which includes several carcinogenic forms (IARC, 1987). Levels of PAHs in hot mix asphalt can vary depending upon the source of raw materials and whether coal tar was used in its production (Burstyn and Kromhout, 2000; Burstyn et al., 2002).
During application of hot mix asphalt, workers are exposed to PAHs through the inhalation of asphalt vapor and fumes [condensed vapor in the liquid and solid form, reported here as particulate matter (PM)] and through dermal contact with PM via deposition or direct contact with contaminated surfaces. The vapor-phase emissions from hot mix asphalt generally contain the smaller and more volatile PAHs ranging in size from two to four rings [e.g. naphthalene (Nap), phenanthrene (Phe), and pyrene (Pyr)], whereas the particulates generally contain the larger four to six rings species (e.g. chrysene, benzo(a)pyrene, and benzo(g,h,i)perylene). Because the particulates from hot mix asphalt contain the most carcinogenic PAHs, health professionals are primarily concerned about asphalt workers’ exposures to PM via inhalation and dermal contact.
Although epidemiological studies point to increased cancer risks among asphalt workers that are likely related to PAH exposures (Partanen and Boffetta, 1994; Boffetta et al., 1997; Hooiveld et al., 2002; Boffetta et al., 2003a,b; Burstyn et al., 2003; Randem et al., 2004; Burstyn et al., 2007), the quantitative relationship between exposures to asphalt chemicals and cancer risk is uncertain (Chiazze et al., 1991; NIOSH, 2000). This uncertainty stems largely from difficulties in assessing exposures to asphalt emissions, particularly the PAH content, via mixed routes (i.e. air and dermal contact, as well as ingestion) and physical forms (i.e. vapors and particulates). To address this uncertainty, previous investigations have quantified asphalt exposures using both air samples and dermal patch samples to gauge the amounts of PAHs potentially inhaled and deposited on the skin (Jongeneelen et al., 1988; McClean et al., 2004a; Väänänen et al., 2005). Analytes included specific PAHs as well as several nonspecific measures of PAH exposure, including PM, benzene-soluble PM, and polycyclic aromatic compounds (PACs, a group of aromatic hydrocarbons containing primarily four to six ring PAHs and heterocyclic compounds). Urinary biomarkers have also been used to assess exposures and internal doses of PAHs in asphalt-exposed workers. Urinary levels of unmetabolized PAHs and of monohydroxylated PAH metabolites, notably those of Nap, Phe, and Pyr, have been associated with airborne and dermal exposures to PAHs in asphalt-exposed populations (Väänänen et al., 2003; McClean et al., 2004b; Campo et al., 2006a,b; Väänänen et al., 2006; Buratti et al., 2007). (These urinary analytes are often considered in PAH biomonitoring studies because they are much more abundant than urinary metabolites of the larger, more carcinogenic PAH species). However, only a few studies have reported concentrations of multiple urinary PAH biomarkers and the corresponding levels of PAHs or PACs in air and on dermal patch samples (Väänänen et al., 2005; Väänänen et al., 2006).
In a study involving 20 road pavers who worked with hot mix asphalt, PACs were measured in air samples and on dermal patches (McClean et al., 2004a). Levels of the PACs varied significantly according to the specific tasks performed by the pavers, and also differed between air measurements and dermal patch measurements, suggesting that PAC exposures were differentially expressed regarding the air and skin (McClean et al., 2004a). In a follow-up investigation, these same workers were further examined using urinary OH-Pyr as a biomarker of exposure to asphalt emissions (McClean et al., 2004b). The authors concluded that the effect of dermal exposure to asphalt emissions on OH-Pyr levels was ~eight times that of inhalation exposure and that the effect of inhalation exposure was more apparent immediately after the work shift, while the effect of dermal exposure was more apparent during the period of 8–16 h after the work shift (McClean et al., 2004b).
More recently, urinary levels of Nap and Phe, and of monohydroxylated metabolites of Nap, Phe, and Pyr, were analyzed in archived urine from the same 20 road pavers (Sobus et al., 2009a). Results showed that urinary biomarker measurements significantly varied by work assignment, as had been observed for PAC measurements in the same workers (McClean et al., 2004a; Sobus et al., 2009a). Thus, the results indicated that urinary biomarkers of Nap, Phe, and Pyr may be suitable exposure surrogates for hot mix asphalt emissions (Sobus et al., 2009a). The purpose of the current investigation is to statistically model the relationships between levels of urinary PAH analytes and the corresponding particulate polycyclic aromatic compound (P-PAC) measurements in air and dermal patch samples from these 20 asphalt-exposed workers. This analysis compares urinary analytes that may be useful biomarkers of exposure to particulate PAHs among asphalt-exposed workers and further delineates the differences between air samples and dermal patch samples as measures of exposure to P-PACs. These results should aid in the design of future studies to assess exposures to particulate PAHs in asphalt emissions for both hazard control and epidemiologic investigations.
Twenty male road pavers, all residing in the Greater Boston area of the USA, were included in this study. Workers were recruited with informed consent according to a protocol approved for the protection of human subjects. Information about smoking habits and body size were obtained by questionnaire.
Starting at the beginning of the workweek, daily air samples and dermal patch samples were collected from each worker over three consecutive days. Personal air measurements of PACs were obtained according to National Institute for Occupational Safety and Health (NIOSH) Method 5506 (NIOSH, 1998), using a personal sampling pump operating at 2 l min−1 to draw air through a Teflon filter (for airborne P-PACs [aP-PACs]) followed by a XAD-2 sorbent tube for vapor polycyclic aromatic compounds (V-PACs). Levels of P-PACs on dermal patch samples (dP-PACs) were measured using a method described by Jongeneelen et al. (1988) and VanRooij et al. (1993), with minor modifications. A soft polypropylene filter was attached to an exposure pad to create a dermal patch with an effective surface area of 8.71 cm2. Using an adhesive backing, the dermal patch was attached to the underside of each wrist (the average of measured PAC levels on the left and right wrists were used here). Levels of PACs in air and dermal patch samples were analyzed using liquid extraction coupled with high-pressure liquid chromatography as described previously (McClean et al., 2004a,b). The limits of detection (LOD) for total airborne PACs (V-PACs + aP-PACs) and dP-PACs were 0.2 μg m−3 and 38 ng cm−2, respectively (McClean et al., 2004a). Method precision was evaluated using the intraclass correlation coefficients (ICCs) determined from the analysis of replicate samples. The ICCs were 0.96 for V-PACs (based on eight replicate samples), 0.97 for aP-PACs (based on 20 replicate samples), and 0.93 for dP-PACs (based on 78 replicate samples), indicating that assay variability was minimal (≤7% of the total variability in sample measurements).
Urine samples were collected from all workers before work (morning), after work (postshift), and at bedtime of each workday. Morning samples collected on the first day of the workweek (after a work-free weekend) were treated as baseline samples for each subject. Each morning sample, collected on a subsequent workday, was treated as the final observation for the previous workday. All urine samples were collected in sterilized polypropylene containers and stored at −20°C for up to 7 years prior to analysis. Urinary creatinine concentrations were determined using a colorimetric procedure (Sigma, 1984). Concentrations of urinary Nap and Phe (U-Nap and U-Phe, respectively) were determined using head space-solid-phase microextraction coupled with gas chromatography–mass spectrometry (Waidyanatha et al., 2003; Sobus et al., 2009b). The estimated LOD for both U-Nap and U-Phe was 0.40 ng l−1 and the estimated coefficients of variation (CVs) were 0.25 and 0.26, respectively (Sobus et al., 2009b). Levels of 1- and 2-hydroxynaphthalene (1- and 2-OH-Nap), 1-, 2-, 3-, 4-, and 9-hydroxyphenanthrene (1-, 2-, 3-, 4-, and 9-OH-Phe), and 1-hydroxypyrene (1-OH-Pyr) were determined using solid-phase extraction coupled with liquid chromatography–tandem mass spectrometry (Onyemauwa et al., 2009). The limits of quantitation (LOQs) for these analytes were as follows: 10 ng l−1 for 1-OH-Nap; 1.0 ng l−1 for 2-OH-Nap; 2.0 ng l−1 for (2+3)-OH-Phe (these isomers could not be chromatographically separated and were quantified together); and 5.0 ng l−1 for 1-, 4-, and 9-OH-Phe and 1-OH-Pyr (Onyemauwa et al., 2009). The estimated CVs for these analytes ranged from 0.053 to 0.27 (Onyemauwa et al., 2009). In this investigation, we consider the summed values of these monohydroxylated metabolites, such that ‘OH-Nap’ is the sum of 1- and 2-hydroxynaphthalene, ‘OH-Phe’ is the sum of 1-, 2-, 3-, 4-, and 9-hydroxyphenanthrene, and ‘OH-Pyr’ is 1-hydroxypyrene.
All statistical tests were performed using SAS statistical software (v. 9.1, SAS Institute, Cary, NC, USA). Measurements of PACs in air and dermal patch samples and urinary analyte levels were evaluated after natural log transformation to satisfy normality assumptions and to remove heteroscedasticity. Considering the repeated measures study design, linear mixed-effects models were used (Proc MIXED) to evaluate exposure–biomarker relationships. Correlation coefficients for exposure data were determined according to Hamlett et al. (2003).
In preliminary analyses of the exposure measurements, levels of aP-PACs and V-PACs were found to be highly correlated (r=0.92, P-value <0.0001) whereas levels of aP-PACs and dP-PACs were essentially uncorrelated (r=0.32, P-value=0.2). Thus, to remove problems related to collinearity in statistical models, and for ease of interpretation, we used only aP-PACs and dP-PACs in our full mixed models. By focusing upon the larger (four rings to six rings) carcinogenic PAHs that were common to both aP-PACs and dP-PACs (Herrick et al., 2007), we were able to directly compare P-PAC levels in air and dermal patch samples. (Very similar results were obtained when V-PACs were used instead of aP-PACs in statistical models).
Variations in urine dilution for the 20 workers were first evaluated using a mixed-effects model with creatinine as the response variable and the period of sample collection (i.e. dummy variables for postshift, bedtime, or morning) as the predictor variables. We then included creatinine as in independent variable in our full mixed-effects models to allow for adjustments on covariates as well as the dependent variables (Barr et al., 2005).
In our full mixed-effects models, postshift, bedtime, and morning urinary levels were individually regressed on the corresponding measurements of aP-PACs and dP-PACs, along with covariates. The following numbers of observations were available for the different sampling times: U-Nap and U-Phe models included 51 postshift, 50 bedtime, and 48 morning measurements, while OH-Nap, OH-Phe, and OH-Pyr models included 49 postshift, 48 bedtime, and 46 morning measurements (six samples could not be analyzed due to insufficient urine volume). Baseline urine samples were not included in the full mixed-effects models because they did not have corresponding measurements of aP-PACs and dP-PACs. For measurements of dP-PACs, McClean et al. (2004a) defined the LOD as three times the standard deviation of field blank values. Using this criterion, ~30% of dP-PAC levels used in this analysis were below the LOD. However, because values were reported for these observations, we used the reported levels in our analyses and did not impute values below the LOQ.
The full mixed-effects model for an exposure–biomarker relationship at the hth sampling time (i.e. postshift=1, bedtime=2, and morning=3) is defined as:
where Xhij represents the analyte concentration for the jth measurement of the ith individual and Yhij is the natural logarithm of the individual measurement Xhij. The coefficient β0h represents the intercept for the hth sampling time. The coefficients β1–β8 represent the coefficients for fixed effects, which include: CREATININEhij [creatinine concentration (g l−1) for the jth measurement of the ith individual at the hth sampling time]; aP-PACshij (ng m−3) and dP-PACshij (ng cm−2) which correspond to the jth air and dermal patch measurements, respectively, for the ith individual at the hth sampling time; DAYhij and HOURhij (hour: postshift=0 h; bedtime and morning values equal elapsed hours since postshift sample) for the jth measurement of the ith individual at the hth sampling time; and AGEi, BMIi, and SMOKEi which represent the ith subject's age (years), body mass index (kg m−2), and smoking status (nonsmoker=0), respectively.
In this full model, bhi and ϵhij are independent random variables; bhi represents the random effect for the ith individual at the hth sampling time and ϵhij represents the random error for the jth measurement of the ith individual at the hth sampling time. It is assumed that bhi and ϵhij are normally distributed with means of zero and variances of and , representing the between-person and within-person components of total variance (i.e., ), respectively, for sampling time h. Linear mixed-effects models were constructed using restricted maximum likelihood estimates of variance components. Fixed effects were retained in the models at a significance level of α=0.10, using manual backwards stepwise elimination. Bayesian Information Criterion (BIC) values were used to select between competing models. Compound symmetry covariance matrices were used; these generally yielded the lowest BIC diagnostic values compared to those from homogeneous autoregressive and heterogeneous autoregressive matrices. Observations with missing values (i.e. air or dermal patch measurement, urinary analyte measurement, or significant covariate) were not included in the final models.
Linear mixed-effects models including only β0h, bhi, and ϵhij were also constructed for each urinary analyte. The estimate of from each null model (no additional fixed effects), designated , was used along with the estimate of from each full model (including significant fixed effects), designated , to calculate the percent of the variance explained by fixed effects, where: [(−)/] × 100 = % explained.
Summary statistics for the urinary analytes (i.e. U-Nap, U-Phe, OH-Nap, OH-Phe, and OH-Pyr), P-PAC measurements (aP-PACs and dP-PACs), and covariates (CREATININE, BMI, AGE, HOUR, and SMOKE) are listed in Table 1. The long period of sample storage (7 years) did not appreciably affect our urinary analyte measurements, based on a comparison [n=194 common observations (including baseline measurements)] of our OH-Pyr measurements [geometric mean (GM)=1130 ng l−1] to earlier OH-Pyr measurements (GM=1100 ng l−1) (McClean et al., 2004b). The correlation coefficient for the paired OH-Pyr measurements from these two studies was 0.89 (P-value < 0.0001), and a regression coefficient of 1.02 (standard error = 0.044) was observed when OH-Pyr measurements from the current study were regressed on those from the original study in log scale. Additional descriptive statistics for the observations of U-Nap, U-Phe, OH-Nap, OH-Phe, OH-Pyr, and PACs (dP-PACs and total airborne PACs) have been reported elsewhere (McClean et al., 2004a,b; Sobus et al., 2009a).
The results of the mixed-effects model with creatinine as the response variable and the period of sample collection (i.e. dummy variables for postshift, bedtime, or morning) as the predictor variables showed a significant effect of the period of sample collection (P-value < 0.0001) on creatinine levels. The highest creatinine levels were observed in postshift samples (mean=1.92 g l−1), followed by bedtime samples (mean=1.53 g l−1), and then morning samples (mean=1.37 g l−1). These changes in urinary creatinine levels are consistent with dehydration of the workers during hot works shifts.
Table 2 shows the final models for urinary analytes in postshift samples. Levels of aP-PACs, dP-PACs, and CREATININE were significant predictors of levels of U-Nap, U-Phe, OH-Phe, and OH-Pyr, and BMI was a significant predictor OH-Pyr. The levels of aP-PACs, dP-PACs, and CREATININE collectively explained 66, 54, 42, and 41% of the variability in levels of OH-Phe, U-Nap, U-Phe, and OH-Pyr, respectively. When CREATININE was removed from the models, aP-PACs and dP-PACs explained 53, 47, 37, and 35% of the variability in levels of OH-Phe, U-Nap, U-Phe, and OH-Pyr, respectively. Conversely, when aP-PACs and dP-PACs were removed from the models, CREATININE explained 13, 8, 6, and 5% of the variability in levels of OH-Phe, U-Nap, U-Phe, and OH-Pyr, respectively. This indicates much weaker effects of CREATININE on analyte levels than those of aP-PACs and dP-PACs in postshift urine. After adjusting for CREATININE, aP-PACs, and dP-PACs, a significant effect of BMI was observed on levels of OH-Pyr only. The negative regression coefficient suggests that OH-Pyr levels were lower in individuals with higher BMIs. Interestingly, the model for OH-Nap levels was quite different from those for all other analytes in postshift urine. For OH-Nap, aP-PACs and dP-PACs were not significant predictors, and the final model included only CREATININE, DAY, AGE, and SMOKE. Together, these variables explained 74% of the postshift variability in OH-Nap levels.
Table 2 shows, for each analyte, the estimated between-person and within-person (log scale) variance components, designated and , respectively, along with corresponding values of , representing the estimated ratio of the within-person variance component to the between-person variance component (i.e. = /). For U-Phe, = 0 indicating that all of the between-person variation in this biomarker was explained by the significant fixed effects, namely, aP-PACs, dP-PACs, and CREATININE. For U-Nap and OH-Pyr, after adjusting for significant fixed effects, was greater than , as shown by values of 2.23 and 1.29, respectively. Values of and were similar for OH-Phe = 0.881), whereas was considerably smaller than for OH-Nap (= 0.483).
Table 3 shows the final models for analytes in bedtime urine samples. In these samples, OH-Phe was the only analyte significantly affected by the exposure variables, aP-PACs and dP-PACs, as well as CREATININE. These combined effects explained 56% of OH-Phe variability. When CREATININE was considered as the only fixed effect, it explained 22% of the variability in OH-Phe levels. When aP-PACs and dP-PACs were considered as predictors of OH-Phe without CREATININE, they explained 23% of OH-Phe variability. Thus, in bedtime samples, the exposure variables (aP-PACs and dP-PACs) and CREATININE contributed equally to the levels of OH-Phe and were the only significant determinants of OH-Phe levels.
As with postshift samples, OH-Nap levels in bedtime samples were highly affected by CREATININE and SMOKE (explaining 65% of the variability in OH-Nap) but not by aP-PACs and dP-PACs. For OH-Pyr, U-Nap, and U-Phe, significant effects of CREATININE and dP-PACs were observed, together explaining 50, 30, and 30% of the respective analyte variability. When included in the models without other fixed effects, CREATININE explained 25, 15, and 14% of the variability in OH-Pyr, U-Phe, and U-Nap levels, respectively. When CREATININE was removed from each model, dP-PACs explained 15, 12, and 11% of the variability in OH-Pyr, U-Nap, and U-Phe levels, respectively. After adjusting for dP-PACs and CREATININE, a moderately significant effect of SMOKE on U-Phe was observed (P-value=0.07). When included in the model, the combined effects of CREATININE, dP-PACs, and SMOKE explained 36% of variability in U-Phe levels.
Results from the final models for bedtime urine samples showed that all analytes except U-Phe ( = 1.13) had values <1 (U-Nap=0.528; OH-Nap=0.698; OH-Phe=0.604; OH-Pyr=0.483), indicating greater between-person variability compared to within-person variability at bedtime.
Model results for morning samples are shown in Table 4. Only the levels of OH-Phe were significantly affected by aP-PACs and dP-PACs, as well as CREATININE; together, these three variables explained 65% of the variability in morning OH-Phe levels. CREATININE alone explained 38% of the variability in OH-Phe levels while aP-PACs and dP-PACs jointly explained 25% of OH-Phe variability.
For U-Nap and OH-Pyr, dP-PACs (but not aP-PACs) and CREATININE were significantly associated with analyte levels, together explaining 36 and 35% of the variability in OH-Pyr and U-Nap levels, respectively. When dP-PACs were removed from the models, CREATININE alone explained 20 and 28% of the variability in OH-Pyr and U-Nap levels, respectively. When dP-PACs were solely considered, 7 and 1% of the variability in OH-Pyr and U-Nap levels was explained, respectively. After adjusting for CREATININE and dP-PACs, a significant negative BMI effect was observed for OH-Pyr. In total, the effects of CREATININE, dP-PACs, and BMI explained 46% of the variability in OH-Pyr levels. Levels of dP-PACs and aP-PACs from the previous workday were not significant determinants for either OH-Nap or U-Phe. In fact, CREATININE and DAY were the only significant determinants for U-Phe, explaining 33% of the variability, and CREATININE, DAY, and SMOKE were the only significant determinants for OH-Nap, together explaining 61% of the variability.
In morning urine samples, more variability in OH-Pyr levels was observed between subjects than was observed within subjects (= 0.744). Estimates of λ were close to one for U-Nap and OH-Phe ( = 1.06 and 0.961, respectively) and were >1 for U-Phe and OH-Nap ( = 3.31 and 2.60, respectively). This indicates comparatively larger within-person variability than between-person variability for U-Phe and OH-Nap compared to U-Nap, OH-Nap and OH-Pyr in morning urine samples.
The purpose of this investigation was to evaluate and compare urinary PAH analytes as biomarkers of exposure to PAHs among asphalt-exposed workers and to further delineate the differences between air samples and dermal patch samples as measures of exposure to P-PACs. From our expanded analysis of urinary PAH biomarkers in these paving workers, we found that aP-PACs and dP-PACs were each strongly associated with the urinary PAH analytes (see Table 2), but were essentially uncorrelated with each other (r=0.32, P-value=0.2). We conclude that air and dermal patch measurements contributed independently to the levels of the urinary biomarkers, and that aP-PACs and dP-PACs provide reasonable measures of particulate PAH exposure by both routes at the time of urine collection.
Figure 1 shows the percentages of the variances of the various urinary analytes that were explained by aP-PACs and dP-PACs in postshift, bedtime, and morning samples. The variables aP-PACs and dP-PACs explained much more of the variability of biomarker levels in postshift urine samples (up to 53%), compared to bedtime (up to 23%) and morning (up to 25%) samples. In postshift samples, the strongest associations with aP-PACs and dP-PACs were observed for OH-Phe (53% of explained variance) followed by U-Nap (47%), U-Phe (37%), and OH-Pyr (35%). In bedtime and morning urine samples, aP-PACs and dP-PACs were much weaker predictors of urinary biomarker levels than they were in postshift urine samples. Nonetheless, levels of OH-Phe were significantly associated with both aP-PACs and dP-PACs in bedtime and morning samples, levels of U-Nap and OH-Pyr were significantly associated with dP-PACs (but not aP-PACs) in bedtime and morning samples, and levels of U-Phe were associated with dP-PACs in bedtime (but not morning) samples (Tables 3 and and4).4). The consistent influence of dP-PACs on levels of PAH biomarkers in bedtime urine and morning urine samples suggest that dermal contact with Nap, Phe, and Pyr persisted well into the evening after a given work shift. Thus, results from our models are consistent with the earlier findings of McClean et al. (2004b), based upon measurements of Pyr and OH-Pyr, and also suggest that levels of U-Nap, U-Phe, OH-Phe, as well as OH-Pyr, are suitable biomarkers of PAH exposure emanating from hot mix asphalt via both inhalation and dermal contact.
The fact that these PAH biomarkers were more highly associated with exposure variables in postshift urine samples, compared to bedtime and morning samples, indicates that Nap, Phe, and Pyr were rapidly absorbed, metabolized, and eliminated from the body during and immediately following the work shift. Since aP-PACs, dP-PACs, and the urinary biomarkers were measured over three consecutive days, we included a workday effect into our models (i.e. DAY) to test the possibility that biomarker levels accumulated over the workweek. After adjusting for aP-PACs, dP-PACs, and other significant covariates, no significant workday effects were observed for U-Nap, OH-Phe, or OH-Pyr. A significant positive workday effect was observed for U-Phe in morning samples only. While this indicates a linear increase in morning U-Phe levels for 3 days, our overall results reinforce the conclusion that these urinary analytes primarily reflect PAH exposures on the current day, and thus would be regarded as short-term biomarkers of exposure (Rappaport and Kupper, 2008). A significant positive workday effect was also observed for OH-Nap in postshift samples and morning samples. However, because the variables aP-PACs and dP-PACs were not included in the final models (most likely due to the confounding effects of smoking), it is difficult to interpret the observed workday effect for OH-Nap.
In addition to assessing workday effects, we used a fixed effect for time in h postshift (i.e. HOUR) because bedtime and morning collection times varied from day to day and from person to person. The median collection time for bedtime samples was 6.09 h [range: 2.50–14.3 h (see Table 1)] and the median collection time for morning samples was 15.3 h [range: 11.8–18.6 h (see Table 1)]. Results from each model showed no significant effect of HOUR on analyte levels. This indicates that, after making adjustments for exposure levels and other covariates, subtle variations in sampling times did not affect urinary analyte levels.
In preliminary models, urinary creatinine concentrations were significantly affected by the period of sample collection (i.e. postshift, bedtime, or morning). This finding indicates that urine volumes varied considerably within and between the road pavers under investigation, perhaps reflecting a diurnal excretion pattern (Boeniger et al., 1993) as well as the dehydration of workers during the hot work shifts. In the full mixed-effects models, levels of urinary creatinine (CREATININE) were significant predictors of all urinary analytes. The positive regression coefficient in each model (Tables 2, ,3,3, and and4)4) indicates increased analyte levels with increased creatinine levels. Figure 2 shows the percent of each analyte's variance explained by CREATININE in urine collected postshift, at bedtime, and in the morning following exposure. Overall, U-Nap and U-Phe were less affected by CREATININE than OH-Nap and OH-Phe; this would be expected because unmetabolized organic compounds tend to enter urine via diffusion from blood rather than via glomerular filtration in the kidney, the latter mechanism being responsible for elimination of polar metabolites (Boeniger et al., 1993). However, Fig. 2 also shows an increase in the strength of association between urinary analytes and CREATININE from postshift to morning. This trend runs counter to that observed between P-PACs and urinary analytes (Fig. 1), indicating that occupational exposure dominates urinary analyte variability in postshift samples, and that urine dilution dominates the variability in morning samples. This result is easily explained considering the short biological half-lives of these urinary analytes and has important implications for biomonitoring studies in which first morning void samples are often used to infer chemical exposures on the previous day.
In postshift, bedtime, and morning samples, neither aP-PACs nor dP-PACs were significantly associated with levels of OH-Nap. However, smoking status (i.e. SMOKE) was a highly significant determinant of OH-Nap levels, as has been observed elsewhere (Serdar et al., 2003a,b; Väänänen et al., 2006; Buratti et al., 2007). We determined that smoking and significant covariates (CREATININE, DAY, and AGE) were able to explain 74, 65, and 61% of the variability in OH-Nap levels in postshift, bedtime, and morning samples, respectively. However, smoking status was not a significant determinant of U-Nap levels in postshift, bedtime, or morning samples. This result was somewhat surprising given that Nap is present in cigarette smoke (Rustemeier et al., 2002). However, our results for U-Nap corroborates those from previous studies where smoking status had little effect on U-Nap levels, particularly in moderately to highly exposed workers (Serdar et al., 2003a; Waidyanatha et al., 2003). Serdar et al. (2004) showed that, at low Nap exposure levels, smokers produced higher OH-Nap levels than nonsmokers, and, at high Nap exposure levels, smokers produced lower OH-Nap levels than nonsmokers and speculated that these results reflected induction of microsomal epoxide hydrolase enzymes by cigarette smoke at low Nap exposure levels. A similar mechanism could have influenced OH-Nap levels in this study. We also observed a significant smoking effect on U-Phe levels in bedtime samples. Considering the lack of a smoking effect for U-Phe levels in postshift samples, the observed elevation of U-Phe levels at bedtime due to smoking seems to be relatively modest compared to those of aP-PACs and dP-PACs.
We observed a significant negative BMI effect on OH-Pyr levels (the only biomarker affected), as previously reported by McClean et al. (2004b), based upon independent measurements of OH-Pyr in these same 20 subjects. This observation may indicate increased storage of Pyr in adipose tissue with elevated exposures, thereby limiting the amount of Pyr in the systemic circulation that is available for metabolism.
Tables 2–4 list estimated values of the within-subject and between-subject variance components obtained from the mixed-effects models of each urinary biomarker, as well as the corresponding estimated variance ratio in each case. Lin et al. (2005) used values of to choose between air measurements and biomarker measurements in selecting the least biasing surrogate for exposure in estimating the slope of an exposure–response relationship. In the current study, we can use values to choose among postshift, bedtime, and morning urine samples to find the minimally biased biomarker of exposure for an epidemiological investigation. Interestingly, values for each of the biomarkers, save OH-Nap (which was affected primarily by smoking), were minimal in bedtime urine samples (see Tables 2–4). Indeed, for U-Nap (=0.53), OH-Phe (=0.60), and OH-Pyr ( = 0.48), the variance ratios were in the range of 0.5, which were much smaller than those for the same analytes in postshift urine samples where values of = 2.2, 0.88, and 1.3, respectively. In light of the earlier discussion concerning the possibility that these biomarkers reflect continued effects of dP-PACs, this finding opens the possibility that bedtime measurements of these biomarkers could be useful in investigating dermal exposures to PAH. This conjecture will require further experimental confirmation.
Finally, we use our results to suggest the most promising urinary PAH biomarker for studies of asphalt-exposed workers. Urinary OH-Pyr is a widely used PAH biomarker in occupational studies (Jongeneelen, 2001), and we observed significant associations between P-PAHs (as measured by aP-PACs and dP-PACs) and OH-Pyr levels in our subjects. This result was expected because Pyr, a four-ring PAH, is often abundant in measurements of PM from asphalt and combustion sources. Somewhat more surprising were the significant associations between Nap and Phe biomarkers (namely U-Nap, U-Phe, and OH-Phe) and P-PAC measurements. Naphthalene, a two-ring PAH, occurs almost exclusively in the gas phase, whereas Phe, a three-ring PAH, is partitioned between the gas and particulate phases. Our results suggest that these abundant two-ring and three-ring PAHs may also be good surrogates for particulate exposures stemming from hot asphalt sources. Moreover, the results from our postshift models indicate stronger associations between levels of U-Nap, U-Phe, and OH-Phe and the P-PAC levels in air and dermal patch samples than the corresponding associations for OH-Pyr. Overall, we consider OH-Phe as the most promising candidate biomarker for evaluating particulate asphalt exposures. We do not recommend use of OH-Nap as a biomarker of exposure to particulate asphalt emissions, due to the apparent confounding by smoking.
National Institute of Environmental Health Sciences (training grant T32ES07018, research grant P42ES05948, and center grant P30ES10126).
Disclaimer—The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Institute of Environmental Health Sciences or the National Institutes of Health.