ENVIRONMENTAL RESEARCH 57, 133--148 (1992)
Urinary Mutagenic Activity in Workers Exposed to Diesel Exhaust MARC B . SCHENKER, 1'* NORMAN Y . K A D O , * ' t S. KATHARINE H A M M O N D , ~ STEVEN J. SAMUELS,* SUSAN R . WOSKIE,~ AND THOMAS J. SMITH~
*Division of Occupational~Environmental Medicine and Epidemiology and ~Department of Environmental Toxicology, University of California, Davis, California 95616-8648, and ~Department of Family and Community Medicine, Environmental Health Science Program, University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester, Massachusetts 01605 Received June 19, 1991 We measured postshift urinary mutagenicity (mutu; n = 306 samples) on a population of railroad workers (n = 87) with a range of diesel exhaust exposures. Postshift urinary mutagenicity was determined by a sensitive microsuspension procedure using Salmonella strain TA 98 +- $9. Number of cigarettes smoked on the study day and urinary cotinine were highly correlated with postshift urinary mutagenicity. Diesel exhaust exposure was measured over the work shift by constant-flow personal sampling pumps. Respirable particle concentrations were adjusted for the contribution of environmental tobacco smoke, as estimated from nicotine concentration on treated filters. The relative ranking of jobs by this adjusted respirable particle concentration (ARP) was correlated with relative contact the job groups have with operating diesel locomotives. After adjustment for cigarette smoking (active and passive) in multiple regressions, there was no independent association of diesel exhaust exposure, as estimated by ARP, with postshift urinary mutagenicity among smokers or nonsmokers. An important finding is the detection of "baseline" mutagenicity in most of the nonsmoking workers. Despite the use of individual measurements of diesel exhaust exposure, the absence of a significant association in this study may be due to the low levels of diesel exposure, the lack of a specific marker for diesel exhaust exposure, and/or urinary mutagenicity levels from diesel exposure below the limit of sensitivity for the mutagenicity assay. © 1992AcademicPress, Inc.
INTRODUCTION In the late 1970s, the projected increase in diesel engine use in the United States and the known mutagenicity of diesel exhaust raised concerns about increased cancer from diesel exhaust exposure (National Research Council, 1981; Pepelko et al., 1979). Epidemiologic studies of lung cancer among workers with occupational exposure to diesel exhaust had previously given conflicting results, but these studies were often limited by small sample sizes, inadequate duration of exposure, limited follow-up, or confounding exposures (Schenker and Speizer, 1979; Schenker, 1980). Several new epidemiologic studies were begun to evaluate the effects of long-term occupational exposure to diesel exhaust, and these studies have generally shown a consistent, small increase in lung cancer attributable to diesel (Garshick et al., 1987, 1988; Howe et al., 1983; Schenker et al., 1984; Schenker, 1989). These epidemiologic study findings are consistent with long1 To whom requests for reprints should be addressed. 133 0013-9351/92 $3.00 Copyright© 1992by AcademicPress. Inc. All rightsof reproductionin any formreserved.
134
SCHENKER ET A L .
term animal exposure studies which have confirmed the carcinogenicity of diesel exhaust (Ishinishi et al., 1986; Mauderly et al., 1987). The International Agency for Research on Cancer has determined that diesel engine exhaust is a probable human carcinogen (IARC, 1989). A major weakness in epidemiologic studies of occupational carcinogens, including the early diesel studies, is limited or inaccurate estimation of exposure or dose. Even accurate measurement of exposure may not be an accurate estimate of the dose received because of individual differences in the handling and response to similar exposures. Biologic markers of exposure are a potential method to characterize more accurately exposure in epidemiologic studies, and thereby increase study sensitivity (Hulka and Wilcowsky, 1988; Griffith et al., 1989). Urinary mutagenicity is convenient for epidemiologic studies because of its low cost and ease of sample handling. Increases in urinary mutagenicity have been observed with some occupational exposures and other enumerated toxins such as cigarette smoke (Everson, 1986; Falck et al., 1980; Kado et al., 1983; Pasquini et al., 1989; De Meo et al., 1987; Scarlet et al., 1990). Mutagenicity is particularly useful for complex mixtures such as diesel exhaust because there are no specific markers for diesel exhaust. We therefore measured urinary mutagenicity in a population of railroad workers who were exposed occupationally to diesel exhaust. Potential confounding exposures such as cigarette smoke and dietary factors were estimated by other biologic markers or by questionnaires. Diesel exhaust and environmental tobacco smoke exposures also were measured by personal sampling. METHODS Study Population
All subjects worked for a single railroad in the northeast United States. Sampling was done during six field visits in the winter because our previous work had shown that diesel exhaust exposure was highest during that season (Woskie et al., 1988a; Hammond et al., 1988). Study subjects were selected, based on job category and work location, from five job categories known to provide a range of diesel exhaust exposures among employees of similar socioeconomic status (Table 1). Employees in management and administrative positions were not studied. TABLE 1 SUBJECTS PARTICIPATING IN FIELD STUDY BY JOB CATEGORY
Subjects,
Collection days,
Age,
Non-tobacco users,
Job category
n
n
E ± SD
n (%)
Brakers Carmen Clerks Engineers Shop workers Total
17 18 21 8 23 87
37 35 60 16 158 306
45.3 45.3 51.0 55.7 43.2 47.2
+ + + + + +
9.7 9.9 7.6 5.2 10.2 9.7
11 (65) 5 (28) 15 (71) 4 (50) 17 (74) 52 (60)
U R I N A R Y M U T A G E N I C I T Y A N D DIESEL E X H A U S T
135
All subjects were studied on at least two consecutive workdays. Because of the limited population size and to maximize study power by oversampling the highest diesel exposure groups, several subjects provided more than one 2-day sample set (Table 1). Subjects were recruited after informed consent. Personal sampling pumps with filter cassettes were attached at the beginning of the work shift, and individual logs were kept of work activities, locations, and field conditions on the study days. Pumps and samples were collected at the end of the work shift, cigarettes smoked were counted, and a spot urine was obtained. The same procedure was repeated on the second day, and the subjects also completed a questionnaire on their medical history, nonworkplace exposures, diet, cigarette smoking, and other lifestyle factors. Several nonworkplace exposure variables were tested as predictors of urinary mutagenicity, including clinical conditions, dietary intake in the previous 2 days, and nonoccupational chemical exposures. A specific dietary variable of "protective food" consisted of individuals who reported ingesting cabbage, brussel sprouts, or fish in the previous 2 days. Diesel Exhaust and Environmental Tobacco Smoke Exposure
Personal exposure to diesel exhaust and other air contaminants was measured by constant-flow personal sampling over full work shifts, which ranged from 7 to 10 hr. The sampling method has been described in detail elsewhere (Woskie et al., 1988a; Hammond et al., 1987, 1988). Briefly, respirable particles were collected on 37-mm Teflon-coated fiber filters (Pallflex Corp, Putnam, CT) preceded by a 10-mm nylon cyclone. Particle mass was determined gravimetrically in a room maintained at controlled temperature (70 -+ 5°F) and humidity (50 -+ 10%). Air flow through the sampling train was drawn at 1.7 liters per minute (lpm), and field calibrated (+5%) with a rotameter. A second filter downstream from the particle filter was treated with an acid, sodium bisulfate, to bind with nicotine vapor, which is alkaline. Nicotine vapor loss by this collecting system was measured by studies of environmental tobacco smoke (ETS) exposure in an environmental chamber, and was less than 1% of the nicotine collected on the treated filters (Hammond et al., 1987). Nicotine was measured by aqueous desorption, pH adjustment, extraction into heptane and gas chromatography with nitrogen phosphorus detection. Recovery of 0.5 ~g nicotine spikes on clean filters was 98 --+ 2%. The limit of detection for field sampling at 1.7 lpm for 8 hr is an average nicotine concentration of 0.2 p~g/m3, based on GC limits of quantitation and linearity. Exposure to diesel exhaust among diesel-exposed workers was estimated from respirable particle concentration (RSP) adjusted for respirable particles from ETS as estimated by the nicotine concentration on each sample (Woskie et al., 1988a; Hammond et al., 1987). The correction factor for ETS particles from nicotine concentrations was derived from environmental chamber studies and confirmed by regressing RSP on nicotine among workers in this study who were non-diesel exposed. Both methods yielded a factor of 8.6 as the coefficient of nicotine concentration, so ETS particle concentration was determined by multiplying nicotine
136
SCHENKER ET AL.
concentration by 8.6, This value was subtracted from individual filter RSP results to yield an "adjusted" RSP (ARP). Mutagenicity Testing We used a microsuspension (preincubation) procedure, previously developed to detect mutagens in the urine of cigarette smokers and non-tobacco users (Kado et al., 1983, 1986). The method is at least l0 times more sensitive than the standard plate incorporation test based on absolute amounts of compound added per plate (Ames et al., 1975). Briefly, Salmonella were grown overnight in oxoid nutrient broth, harvested by centrifugation, and resuspended in iced phosphate-buffered saline (PBS, 0.15 M, pH 7.4) to a concentration of 1 x l0 l° cells/ml. For the microsuspension assay the following ingredients were added in order to a 12 x 75-ram glass culture tube placed in an ice bath: 0.1 ml $9 (Ames et al., 1975), 0.005 ml urine extract in dimethyl sulfoxide (DMSO), and 0.1 ml bacteria in PBS (1 × 109 bacteria per tube). The mixture was incubated at 37°C for 90 min with shaking. After incubation, the tubes were placed in an ice water bath and removed singly from the ice bath, and 2 ml of molten top agar containing 90 nmole of both histidine and biotin was added. The molten suspensions were mixed immediately with a Vortex mixer and poured onto minimal glucose plates. The plates were incubated at 37°C in the dark for 48 hr and counted using an automatic colony counter. Genetic markers for the strains were routinely verified. All extraction and testing were carried out in a room fitted with yellow fluorescent lights to minimize potential photooxidation. Urine samples were collected in washed, sterile amber-colored glass or polyethylene bottles and immediately placed on ice and stored at -213°C. Urines were extracted with XAD-2, following the methods of Yamasaki and Ames (1977) and Putzrath and co-workers (1981). Urine extracts were tested for mutagenic activity in batches with their own PBS controls. All urine samples or PBS control samples were tested in duplicate at a minimum of three urine equivalent doses, 2.5, 5, and l0 ml, with and without $9. The mutagens 2-aminofluorene and benzo[a]pyrene served as positive controls for + $9, and 2-nitrofluorene and 4-nitroquinolone-Noxide served as positive controls for - $ 9 assays. All positive control mutagens were tested at three doses in triplicate. Four or five plates were routinely used for the DMSO standard which provided the 0.0 concentration. Only single values are presented for all replicate samples. Thus, the sample size for all samples reflects the number of specimens collected. Mutagenicity (revertants per milliliter equivalent of urine) was determined from a straight-line least-squares regression, followed by subtraction of the slope from the extraction control PBS sample. Points starting from the highest dose were dropped until the t test for quadratic curvature was "nonsignificant" at p > 0.15. Urinary concentration of cotinine was determined by P. Jacob, L. Yu, and N. Benowitz, using a multistep extraction procedure followed by gas chromatographic analyses (Jacob et al., 1981). Concentrations of urine markers were standardized for urine creatinine concentration. Statistics Diesel exhaust exposure was evaluated as a predictor of urinary mutagenicity in
URINARY MUTAGENICITYAND DIESEL EXHAUST
137
three ways: (1) by evaluation of mean urinary mutagenicity among samples from workers classified by a priori levels of diesel exposure as estimated by job categories; (2) by evaluation of urinary mutagenicity levels among ordered strata of diesel-exposed workers as estimated by ARP; and (3) by a full multiple regression model in which the effect of "diesel exposure" was expressed as an interaction between the amount of respirable particles and nicotine in the air and work at jobs with known potential for diesel exposure. Results were adjusted for other predictors of mutagenicity as determined by multiple linear regressions. To accommodate the multiple observations per person that were taken in the study, an estimated generalized least-squares program (BMDP3V) was used, which allowed for within- and between-person variance components. All analyses were done by SAS 5.16 (SAS Institute, Inc., 1985) or BMDP (BMDP Statistical Software, 1983) procedures on a VAX 11/750 computer. RESULTS
The 87 subjects studied had a mean age of 47.2 -+ 9.7 (x -+ SD) and mean years of railroad work of 16.1 + 10.8 (Table 1). Engineers tended to be older than subjects in other job categories, which is consistent with railroad career paths. Thirty-six percent of subjects (n = 31) were cigarette smokers. Four nonsmokers and two smokers used other forms of tobacco. This yielded 95 urine samples for cigarette smokers (32%), 10 samples for cigarette smokers who used other tobacco products (3%), 189 non-tobacco user samples (62%), and 12 samples from subjects who used other tobacco products (4%). Results for cigarette smokers who used other tobacco products were similar to those for cigarette smokers and, therefore, were analyzed together. Results for urine analyses of non-cigarettesmoking subjects who used other tobacco products were excluded or considered separately. The highest rate of tobacco use was among the carmen studied (72% of subjects), with lower rates of smoking present for the other job categories (2%50% of subjects) (Table 1). Among smokers, the average number of cigarettes smoked during the day prior to urine sample collection was 14.4 -+ 8.1. For most analyses the number of cigarettes smoked during the study day up to the time of urine collection was used, since it was felt to better reflect markers present in the urine at the end of the shift than the number of cigarettes smoked only during the shift. Some samples were missing or not analyzed from each source in this large, multidisciplinary field investigation. The reasons were different for the specific variables. The number of samples not analyzed ranged from 4% for respirable particle concentrations to 12% for urinary cotinine. The high number of samples not analyzed for cotinine occurred because it was measured after all other urine assays, and there was often an insufficient quantity of urine remaining. Mutagenicity was not measured with $9 in 6% of samples and without $9 in 10% of samples, in most cases because of an inadequate quantity of urine. Diesel Exhaust and ETS Exposure A total of 303 (99% of collection days) personal exposure samples were collected. However, 8 samples were missing particulate mass values due to filter
138
S C H E N K E R ET A L.
damage and 23 were missing nicotine values. Use of medians for exposure variables (Table 2) adjusts for the skewness commonly seen in the distribution of airborne exposures, which was also present in our data. The carmen had the highest exposures to RSP, even though they have no diesel exhaust exposure. This resulted in part because some of them were exposed to particulate matter generated by welding or by using acetylene torch for cutting and burning chassis of train cars during repairs, an activity that could occupy 1/2 to 4 hr per day. A large difference in RSP was seen in the average exposures of the carmen welders and nonwelders (Table 2), confirming the contribution of weldingassociated exposures to the RSP exposures among carmen. Despite the expected difference in diesel exhaust exposure between the shop workers and the clerks, the RSP concentrations of these job groups (Table 2) were similar, which is consistent with our previous findings from U.S. railroads (Woskie et al., 1988a; Hammond et al., 1988). This similarity of RSP concentrations between job groups was largely due to the RSP mass contributed by ETS for the clerks. Median concentrations of ETS were highest among the carmen, who had the highest prevalence of smoking, with high concentrations also present among the clerks and the engineers. The relative ranking of the jobs by ARP (Table 2), excluding carmen, was close to that expected by the relative contact these groups have with operating diesel locomotives, assuming that all workers are exposed to a background air concentration of about 30 p,g/m 3. Specifically, highest ARPs were seen for the shop workers, who have the highest exposure to diesel exhaust, and lower ARP levels were obtained for the other job categories, which have little or no diesel exposure. This ranking and ARP levels are similar to those obtained in our previous study of railroad workers (Woskie et al., 1988a; Hammond et al., 1988). However, this cannot explain all of the differences seen between the job groups. Daily exposures for individual study days were combined into strata of ARP for regression analyses of urinary mutagenicity and diesel exposure. Workers in groups known a priori not to be diesel exposed (clerks, carmen) were placed in the unexposed strata (Woskie et al., 1988a; Hammond et al., 1988). Cutoff values for the strata of ARP for smokers and non-tobacco users were 77 and 155 ixg/m3, selected to create approximately equal-size strata. The median levels of ARP for the low, intermediate, and high strata of nonsmoking diesel-exposed workers analyzed by regression were 50, 110, and 181 Ixg/m3, respectively. A similar relationship between ARP and diesel exposure was seen for the smoking workers, with median levels of ARP in the three strata of 10, 115, and 222 txg/m3. Urinary Mutagenicity and Diesel Exhaust Exposure Urinary mutagenicity was first evaluated by job group among nonsmokers (Table 3). No association was present for job grouping and urinary mutagenicity +$9 according to a priori classification of diesel exhaust exposure. Highest levels of urinary mutagenicity among the nonsmokers were seen in the carmen welders, probably due to their welding exposures. Similarly, no association was present among cigarette smokers for mutagenicity +-$9 and diesel exposure as estimated by job category (Table 3). Regression models were run separately for smokers and for nonsmokers, for
URINARY
MUTAGENICITY
AND
DIESEL
139
EXHAUST
r-=_-
0
o
2
0
u.l
? o
0.05). None of the differences among the three strata of diesel-exposed workers or comparisons with the unexposed workers was statistically significant. Regressions were also run excluding all carmen, with results similar to those presented (results not shown). Mutagenicity - $9 also failed to demonstrate a dose-response relationship with diesel exposure strata. The mean level of adjusted mutagenicity ( - $ 9 ) in the non-diesel-exposed and three strata of ARP (low to high) among cigarette smokers were 0.48, 0.20, 0.20, and 0.34, respectively (Table 4). As with the results with $9, TABLE 4 CRUDE AND ADJUSTED MUTAGENICITY FROM MULTIVARIATEREGRESSION MODELS BY STRATA OF ADJUSTED RESPIRABLE PARTICLE (ARP) CONCENTRATIONS
Mean
2SE interval of adjusted
Crude ~
Adjustedb
0.33
0.46
0.26, 0.66
0.37 0.37 0.34
0.37 0.34 0.12
0.17, 0.56 0.19, 0.49 - 0 . 1 5 , 0.38
0.14
0.16
- 0.01, 0.33
0.06 0.11 0.10
0.10 0.12 0.03
- 0 . 0 6 , 0.26 - 0.00, 0.25 - 0.18, 0.24
4.16
3.18
2.37, 4. l0
3.78 3.91 4.30
3.12 3.93 3.54
2.06, 4.39 2.85, 5.17 2.09, 5.33
0.46
0.48
0.23, 0.72
0.28 0.18 0.30
0.20 0.20 0.34
- 0 . 1 0 , 0.50 - 0.06, 0.46 - 0 . 0 4 , 0.71
Non-tobacco users +$9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP - $9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP
Smokers +$9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP -$9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP
a Average observed values. b Non-tobacco user values adjusted for urinary cotinine, ambient nicotine, and positive batch con-
trol 2-aminofluorene. Smoker values adjusted for urinary cotinine and cigarettes smoked on study day and positive batch control 2-aminofluorene.
142
SCHENKER ET AL.
there was no suggestion of an association between diesel exposure and mutagenicity without $9 among cigarette smokers. Among non-tobacco users, multiple regressions using stratification of ARP also revealed no evidence of a dose response for mutagenicity with or without $9 (Table 4). Mean adjusted urinary mutagenicities (+$9), for the non-dieselexposed and low, medium, and high strata of ARP (diesel) were 0.46, 0.37, 0.34, and 0.12, respectively (Table 4). A similar absence of dose response was seen among non-tobacco users for mutagenicity - $ 9 . Similarly, in the full regression models (not shown), no additional contribution was made to mutagenicity ___$9 by dichotomous diesel exposure status (yes/no) or by the interaction of diesel exposure status (yes/no) with respirable nicotine or with total respirable particles. Among non-tobacco users the estimated between-person variances were approximately one-fourth the magnitude of the within-person (between-sample) variances, indicating a definite lack of independence in observations taken from the same worker. To examine more closely the possible effects of diesel exposure independent of individual differences and the effects of ETS, we sought to identify a subset of workers who (1) had jobs with potentially high diesel exposure, (2) were nonsmokers, (3) were exposed to little or no ETS (based on nicotine on personal samplers), and (4) had been sampled repeatedly. Three repair shop workers met these criteria, each of whom had been sampled on 8 to l0 different days. Plots for the repeated samples on these three workers of urinary mutagenicity +-$9 (revertants per micromole of creatinine) versus the marker of diesel exposure (ARP) showed no suggestion of a positive association (Figs. 1 and 2). This provides additional support for the regression results showing no association between diesel exposure and urinary mutagenicity among nonsmokers. 2-
O
E i []i
• m&
!
•
1(3 170
•
IO 206
[]
ID 285
~gD 0
.0
0
100
1
-100
200
300
400
500
6OO
Adjusted RSP (pg/m 3) FIG. 1. Urinary mutagenicity +$9 versus respirable particle concentration adjusted for environmental tobacco smoke (ARP, see test) for three nonsmoking shop workers with repeated measurements and low environmental tobacco smoke exposures.
143
URINARY MUTAGENICITY AND DIESEL EXHAUST
== L~ ®
E ::3.. t3
=& • •
o
o~
t
-l-
=
I
¢=
a
0
100
•
1D 1 7 0
•
ID206
a
ID285
C
-1 -100
200
300
400
500
600
Adjusted RSP ( p g / m 3 ) FIG. 2. Urinary mutagenicity - $ 9 versus respirable particle concentration adjusted for environmental tobacco smoke (ARP, see text) for three nonsmoking shop workers with repeated measurements and low environmental tobacco smoke exposures.
The "protective food" variable (ingestion of cabbage, brussels sprouts, or fish in the prior 2 days) was associated with lower urinary mutagenicity only with $9 among smokers and non-tobacco users, but inclusion of this variable did not alter the absence of an association of diesel exposure category and mutagenicity (data not shown). DISCUSSION Based on the use of exposure measurements to estimate diesel dose and markers of exposure to potentially confounding exposures (active and passive smoking), there was no association between postshift urinary mutagenicity and diesel exhaust exposure in this population. This absence of an association does not mean that diesel-exposed workers did not excrete diesel-associated mutagens. Factors which are important in evaluating this result include the actual dose of dieselassociated mutagens, limits of the mutagenicity assay sensitivity, pharmacokinetics of mutagen handling including absorption and excretion, interindividual variability, and uncontrolled sources of urinary mutagenicity such as diet. The median level of RSP among nonsmoking subjects in our highest exposed job group (shop workers) was only 132 txg/m3, and 75% of the samples were below 200 txg/m 3. These levels are lower than many historical measurements of diesel exhaust exposure among railroad workers, and probably reflect the decline in U.S. railroad activity with associated reductions in diesel engine use (Woskie et al., 1988b). The use of personal measurements for RSP concentrations adjusted for ETS improved the accuracy of our diesel exposure estimates, but some error in exposure assessment likely remained (Hammond et al., 1988). Another problem with
144
S C H E N K E R ET A L.
using ARP as a measure of diesel exhaust exposure is that RSP has sources other than diesel and ETS, and particles from those sources are included in ARP. This is illustrated by the high ARP levels among carmen, who are known not to be exposed to diesel exhaust. It is likely that this imprecision would reduce the ability to observe an association of diesel exhaust with urinary mutagenicity if one were present, but it is not possible to quantify the magnitude of this effect. The development of improved markers of diesel exhaust exposure such as elemental carbon or mutagenic activity of respirable particles (Kado et al., 1987) should increase the ability to detect an association of exposure and biologic markers of dose. However, accurate markers for complex mixtures such as diesel exhaust remain a difficult challenge (Hammond et al., 1988; Hulka and Wilcowsky, 1988). It is unlikely that a unique marker will be identified for diesel exhaust, similar to the relationship of nicotine to tobacco smoke, but improved estimates of diesel dose will improve the sensitivity of epidemiologic studies. Recently, Zaebst and co-workers (1991) demonstrated that the elemental carbon content of particles is a promising marker for exposure to diesel exhaust. Diesel exhaust-associated mutagens may have been present in the urine, but below the level of detection in the afternoon urines. For example, a worker exposed to a mutagen concentration of 1000 revertant equivalents per cubic meter of air who breathed 7-10 m 3 during the 7-hr exposure period received a total mutagen dose of approximately 7000 to I0,000 revertant equivalents (1000 rev/m 3 x 7 to 10 m3). The 1000 revertant equivalents per cubic meter of air was estimated from a diesel shop worker's personal filters (Kado et al., 1987). If we assume a small fraction of mutagens are excreted in the urine in mutagenic form, then the measurable revertant equivalents can be approximated. This may not be an unreasonable assumption since the mutagens absorbed on diesel particulate matter will have to be desorbed in physiologic fluids. The bioavailability of mutagens on diesel particulate matter has been reported to be a relatively slow desorption process (King et al., 1981). Belisario and co-workers (1984), who injected rats with whole diesel particles and measured urinary mutagenic activity, found that only a small fraction of measurable mutagens (in this case, less than 5%, approximately) are excreted in a 24-hr period. For illustrative purposes, if 1% of the total revertant equivalents are excreted in the afternoon urine sample collected, 70 to 100 revertant equivalents would be in 100 to 200 ml of urine (0.01 x 7000 to 10,000 rev eq). The concentration would therefore be at the most 0.7 to 1 rev eq/ml urine (70 to 100 rev eq/100 ml urine). This is the level of activity generally determined in the urine of nonsmokers (Kado et al., 1983). Another important consideration in evaluating these results is the unknown kinetics of diesel exhaust absorption and excretion in humans and the limited ability to evaluate them in the field setting. For example, it was feasible in this field study to collect only a single spot urine at the end of the work shift. While the spot urine samples showed a good dose-response relationship with smoking dose on the workday, if diesel mutagens were excreted more slowly than mutagens from cigarette smoke, the assay might not have been sensitive enough to detect the diesel exhaust mutagens. It is also possible that the diesel mutagens are in part
URINARY MUTAGENICITYAND DIESEL EXHAUST
145
excreted by another route, such as the feces, or are excreted in a form which is not mutagenic in the urine. The diesel-associated mutagens may be conjugated to polar molecules such as glucuronides or sulfates, which may not be mutagenic and may not be efficiently extracted by XAD-2. Finally, the study power may not have been great enough to observe an association between diesel exhaust and urinary mutagenicity, if one were present, but we do not believe this explains our negative findings. Study power was evaluated for analyses -+$9 in both smokers and nonsmokers. Multiple regression models with four diesel exposure groups were employed. Power was calculated using the three-degrees-of-freedom partial F test of differences in group means, after adjustment for other factors. For non-tobacco users, the mean mutagenic activity (+ $9) was about 0.3 rev/~mole creatinine with a standard deviation of 0.4. The adjusted + $9 means (Table 4) had a maximum difference of about 0.1 rev/p~mole creatinine. If the unexposed group is considered as baseline, assuming true mean differences from baseline of 0.1, 0.2, and 0.3 corrected counts in the three ordered diesel groups, these differences would have been detected with 93% probability in this study. For smokers the maximum difference between adjusted means (Table 4) was about 0.4 rev/wmole creatinine, and 75% of individual observations fell between 1.9 and 5.4 rev/p~mole creatinine. If the true differences from the baseline mean were + 0.2, + 1.5, a n d + 2.0 (lowest to highest exposure), statistical significance at the 0.05 level would have been declared 84% of the time. The above estimates of power are liberal for nonsmokers because of the independence assumption. They are conditional on the observed values of the covariates in our data and on the estimated coefficients for the covariates. Thus, we do not think that a lack of statistical power can explain our findings, but we cannot exclude that possibility. Our findings are consistent with those of Willems et al. (1989) who reported on the urinary and fecal mutagenicity of car mechanics exposed to diesel exhaust. They used the standard plate incorporation test (Ames et al., 1975), and found that no exposure to diesel exhaust (based on job category) resulted in increased mutagenicity in the urine or feces. The net result of the factors discussed above is that detection of elevated urine mutagenicity resulting specifically from diesel exhaust exposure was not possible despite adequate statistical power in this large field study. However, some important findings on urinary mutagenicity in working populations derived from the scope of this study. First, measuring detectable mutagenicity "baseline" was possible for most of the nonsmoking subjects. Second, we were able to estimate personal exposure to diesel exhaust in a field study by adjustment of respirable particle exposure for particulate mass from environmental tobacco smoke. The use of a biologic marker such as urinary mutagenicity to measure diesel exhaust represents a common difficulty in epidemiologic studies of exposure to complex mutagen and carcinogen mixtures of this type. More specific and sensitive assays including molecular markers could increase the chances of success. This study provides no basis upon which to draw conclusions regarding cancer risk associated with diesel exhaust exposure. Other epidemiologic and animal studies have directly addressed this issue (Schenker, 1989; Ishinishi et al., 1986;
146
SCHENKER ET AL.
M a u d e r l y e t al., 1987; G a r s h i c k e t al., 1987, 1988), a n d a r e m o r e a p p r o p r i a t e t o evaluate such an association.
ACKNOWLEDGMENTS Research described in this article was conducted under contract to the Health Effects Institute (HEI), an organization that supports the conduct of independent research and is jointly funded by the U.S. Environmental Protection Agency (EPA) and automotive manufacturers. Publication here implies nothing about the view of the contents by HEI or its research sponsors. Although the work described in this document was funded in part by the U.S. Environmental Protection Agency under Assistance Agreement X808859 with HEI, the contents do not necessarily reflect the views and policies of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
REFERENCES Ames, B. N., McCann, J., and Yamasaki, E. (1975). Methods for detecting carcinogens and mutagens with the Salmonella/mammalian-microsome mutagenicity test. Mutat. Res. 31, 347-364. Atkinson, A. C. (1985). "Plots, Transformations and Regressions." Oxford Univ. Press (Clarendon), London. Belisario, M. A., Buonocore, V., DeMarinis, E., and DeLorenzo, F. (1984). Biological availability of mutagenic compounds absorbed onto diesel exhaust particulate. Mutat. Res. 135, 1-9. BMDP Statistical Software (1983). "Printing with Additions." University of California Press, Berkeley. De Meo, M. P., Dumenil, G., Botta, A. H., Laget, M., Zabaloueff, V., and Mathias, A. (1987). Urine mutagenicity of steel workers exposed to coke oven emissions. Carcinogenesis 8, 363-367. Everson, R. B. (1986). Detection of occupational and environmental exposures by bacterial mutagenesis assays of human body fluids. J. Occup. Med. 28, 647-655. Falck, K., Sousa, M., and Vainio, H. (1980). Mutagenicity in urine of workers in rubber industry. Mutat. Res. 79, 45-52. Garshick, E., Schenker, M. B., Munoz, A., Segal, M., Smith, T. J., Woskie, S. R., Hammond, S. K., and Speizer, F. E. (1987). A case-control study of lung cancer and diesel exhaust exposure in railroad workers. Am. Rev. Respir. Dis. 135, 1242-1248. Garshick, E., Schenker, M. B., Munoz, A., Segal, M., Smith, T. J., Woskie, S. R., Hammond, K., and Speizer, F. E. (1988). A retrospective cohort study of lung cancer and diesel exhaust exposure in railroad workers. Am. Rev. Respir. Dis. 137, 820-825. Griffith, J., Duncan, R. C., and Hulka, B. S. (1989). Biochemical and biological markers: Implications for epidemiologic studies. Arch. Environ. Health 44, 375-381. Hammond, S. K., Leaderer, B. P., Roche, A. C., and Schenker, M. (1987). Collection and analysis of nicotine as a marker for environmental tobacco smoke. Atmos. Environ. 21, 457--462. Hammond, S. K., Smith, T. J., Woskie, S. R., Leaderer, B. P., and Bettinger, N. (1988). Markers of exposure to diesel exhaust and cigarette smoke in railroad workers. Am. Ind. Hyg. Assoc. J. 49, 516-522. Howe, G. R., Fraser, D., Lindsay, J., Presnal, B., and Yu, S. (1983). Cancer mortality (1965-1977) in relation to diesel fume and coal exposure in a cohort of retired railroad workers. J. Natl. Cancer Inst. 70, 1015-1019. Hulka, B. W., and Wilcowsky, T. (1988). Biologic markers in epidemiologic research. Arch. Environ. Health 43, 83-89. International Agency for Research on Cancer (IARC) (t989). "IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Diesel and Gasoline Engine Exhausts and Some Nitroarenes," Vol. 46. IARC, Lyon. Ishinishi, N., Koizumi, A., McClellan, R. O., and Stober, W. (Eds.) (1986). "Carcinogenicity and Mutagenicity of Diesel Engine Exhaust." Elsevier Science, New York. Jacob, P., Wilson, M., and Benowitz, N. L. (1981). Improved gas chromatographic method for the determination of nicotine and cotinine in biologic fluids. J. Chromatogr. 222, 61-70.
URINARY MUTAGENICITY AND DIESEL EXHAUST
147
Kado, N. Y., Guirguis, G. N., Flessel, C. P., Chan, R. C., Chang, K., and Wesolowski, J. J. (1986). Mutagenicity of fine (
Urinary Mutagenic Activity in Workers Exposed to Diesel Exhaust MARC B . SCHENKER, 1'* NORMAN Y . K A D O , * ' t S. KATHARINE H A M M O N D , ~ STEVEN J. SAMUELS,* SUSAN R . WOSKIE,~ AND THOMAS J. SMITH~
*Division of Occupational~Environmental Medicine and Epidemiology and ~Department of Environmental Toxicology, University of California, Davis, California 95616-8648, and ~Department of Family and Community Medicine, Environmental Health Science Program, University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester, Massachusetts 01605 Received June 19, 1991 We measured postshift urinary mutagenicity (mutu; n = 306 samples) on a population of railroad workers (n = 87) with a range of diesel exhaust exposures. Postshift urinary mutagenicity was determined by a sensitive microsuspension procedure using Salmonella strain TA 98 +- $9. Number of cigarettes smoked on the study day and urinary cotinine were highly correlated with postshift urinary mutagenicity. Diesel exhaust exposure was measured over the work shift by constant-flow personal sampling pumps. Respirable particle concentrations were adjusted for the contribution of environmental tobacco smoke, as estimated from nicotine concentration on treated filters. The relative ranking of jobs by this adjusted respirable particle concentration (ARP) was correlated with relative contact the job groups have with operating diesel locomotives. After adjustment for cigarette smoking (active and passive) in multiple regressions, there was no independent association of diesel exhaust exposure, as estimated by ARP, with postshift urinary mutagenicity among smokers or nonsmokers. An important finding is the detection of "baseline" mutagenicity in most of the nonsmoking workers. Despite the use of individual measurements of diesel exhaust exposure, the absence of a significant association in this study may be due to the low levels of diesel exposure, the lack of a specific marker for diesel exhaust exposure, and/or urinary mutagenicity levels from diesel exposure below the limit of sensitivity for the mutagenicity assay. © 1992AcademicPress, Inc.
INTRODUCTION In the late 1970s, the projected increase in diesel engine use in the United States and the known mutagenicity of diesel exhaust raised concerns about increased cancer from diesel exhaust exposure (National Research Council, 1981; Pepelko et al., 1979). Epidemiologic studies of lung cancer among workers with occupational exposure to diesel exhaust had previously given conflicting results, but these studies were often limited by small sample sizes, inadequate duration of exposure, limited follow-up, or confounding exposures (Schenker and Speizer, 1979; Schenker, 1980). Several new epidemiologic studies were begun to evaluate the effects of long-term occupational exposure to diesel exhaust, and these studies have generally shown a consistent, small increase in lung cancer attributable to diesel (Garshick et al., 1987, 1988; Howe et al., 1983; Schenker et al., 1984; Schenker, 1989). These epidemiologic study findings are consistent with long1 To whom requests for reprints should be addressed. 133 0013-9351/92 $3.00 Copyright© 1992by AcademicPress. Inc. All rightsof reproductionin any formreserved.
134
SCHENKER ET A L .
term animal exposure studies which have confirmed the carcinogenicity of diesel exhaust (Ishinishi et al., 1986; Mauderly et al., 1987). The International Agency for Research on Cancer has determined that diesel engine exhaust is a probable human carcinogen (IARC, 1989). A major weakness in epidemiologic studies of occupational carcinogens, including the early diesel studies, is limited or inaccurate estimation of exposure or dose. Even accurate measurement of exposure may not be an accurate estimate of the dose received because of individual differences in the handling and response to similar exposures. Biologic markers of exposure are a potential method to characterize more accurately exposure in epidemiologic studies, and thereby increase study sensitivity (Hulka and Wilcowsky, 1988; Griffith et al., 1989). Urinary mutagenicity is convenient for epidemiologic studies because of its low cost and ease of sample handling. Increases in urinary mutagenicity have been observed with some occupational exposures and other enumerated toxins such as cigarette smoke (Everson, 1986; Falck et al., 1980; Kado et al., 1983; Pasquini et al., 1989; De Meo et al., 1987; Scarlet et al., 1990). Mutagenicity is particularly useful for complex mixtures such as diesel exhaust because there are no specific markers for diesel exhaust. We therefore measured urinary mutagenicity in a population of railroad workers who were exposed occupationally to diesel exhaust. Potential confounding exposures such as cigarette smoke and dietary factors were estimated by other biologic markers or by questionnaires. Diesel exhaust and environmental tobacco smoke exposures also were measured by personal sampling. METHODS Study Population
All subjects worked for a single railroad in the northeast United States. Sampling was done during six field visits in the winter because our previous work had shown that diesel exhaust exposure was highest during that season (Woskie et al., 1988a; Hammond et al., 1988). Study subjects were selected, based on job category and work location, from five job categories known to provide a range of diesel exhaust exposures among employees of similar socioeconomic status (Table 1). Employees in management and administrative positions were not studied. TABLE 1 SUBJECTS PARTICIPATING IN FIELD STUDY BY JOB CATEGORY
Subjects,
Collection days,
Age,
Non-tobacco users,
Job category
n
n
E ± SD
n (%)
Brakers Carmen Clerks Engineers Shop workers Total
17 18 21 8 23 87
37 35 60 16 158 306
45.3 45.3 51.0 55.7 43.2 47.2
+ + + + + +
9.7 9.9 7.6 5.2 10.2 9.7
11 (65) 5 (28) 15 (71) 4 (50) 17 (74) 52 (60)
U R I N A R Y M U T A G E N I C I T Y A N D DIESEL E X H A U S T
135
All subjects were studied on at least two consecutive workdays. Because of the limited population size and to maximize study power by oversampling the highest diesel exposure groups, several subjects provided more than one 2-day sample set (Table 1). Subjects were recruited after informed consent. Personal sampling pumps with filter cassettes were attached at the beginning of the work shift, and individual logs were kept of work activities, locations, and field conditions on the study days. Pumps and samples were collected at the end of the work shift, cigarettes smoked were counted, and a spot urine was obtained. The same procedure was repeated on the second day, and the subjects also completed a questionnaire on their medical history, nonworkplace exposures, diet, cigarette smoking, and other lifestyle factors. Several nonworkplace exposure variables were tested as predictors of urinary mutagenicity, including clinical conditions, dietary intake in the previous 2 days, and nonoccupational chemical exposures. A specific dietary variable of "protective food" consisted of individuals who reported ingesting cabbage, brussel sprouts, or fish in the previous 2 days. Diesel Exhaust and Environmental Tobacco Smoke Exposure
Personal exposure to diesel exhaust and other air contaminants was measured by constant-flow personal sampling over full work shifts, which ranged from 7 to 10 hr. The sampling method has been described in detail elsewhere (Woskie et al., 1988a; Hammond et al., 1987, 1988). Briefly, respirable particles were collected on 37-mm Teflon-coated fiber filters (Pallflex Corp, Putnam, CT) preceded by a 10-mm nylon cyclone. Particle mass was determined gravimetrically in a room maintained at controlled temperature (70 -+ 5°F) and humidity (50 -+ 10%). Air flow through the sampling train was drawn at 1.7 liters per minute (lpm), and field calibrated (+5%) with a rotameter. A second filter downstream from the particle filter was treated with an acid, sodium bisulfate, to bind with nicotine vapor, which is alkaline. Nicotine vapor loss by this collecting system was measured by studies of environmental tobacco smoke (ETS) exposure in an environmental chamber, and was less than 1% of the nicotine collected on the treated filters (Hammond et al., 1987). Nicotine was measured by aqueous desorption, pH adjustment, extraction into heptane and gas chromatography with nitrogen phosphorus detection. Recovery of 0.5 ~g nicotine spikes on clean filters was 98 --+ 2%. The limit of detection for field sampling at 1.7 lpm for 8 hr is an average nicotine concentration of 0.2 p~g/m3, based on GC limits of quantitation and linearity. Exposure to diesel exhaust among diesel-exposed workers was estimated from respirable particle concentration (RSP) adjusted for respirable particles from ETS as estimated by the nicotine concentration on each sample (Woskie et al., 1988a; Hammond et al., 1987). The correction factor for ETS particles from nicotine concentrations was derived from environmental chamber studies and confirmed by regressing RSP on nicotine among workers in this study who were non-diesel exposed. Both methods yielded a factor of 8.6 as the coefficient of nicotine concentration, so ETS particle concentration was determined by multiplying nicotine
136
SCHENKER ET AL.
concentration by 8.6, This value was subtracted from individual filter RSP results to yield an "adjusted" RSP (ARP). Mutagenicity Testing We used a microsuspension (preincubation) procedure, previously developed to detect mutagens in the urine of cigarette smokers and non-tobacco users (Kado et al., 1983, 1986). The method is at least l0 times more sensitive than the standard plate incorporation test based on absolute amounts of compound added per plate (Ames et al., 1975). Briefly, Salmonella were grown overnight in oxoid nutrient broth, harvested by centrifugation, and resuspended in iced phosphate-buffered saline (PBS, 0.15 M, pH 7.4) to a concentration of 1 x l0 l° cells/ml. For the microsuspension assay the following ingredients were added in order to a 12 x 75-ram glass culture tube placed in an ice bath: 0.1 ml $9 (Ames et al., 1975), 0.005 ml urine extract in dimethyl sulfoxide (DMSO), and 0.1 ml bacteria in PBS (1 × 109 bacteria per tube). The mixture was incubated at 37°C for 90 min with shaking. After incubation, the tubes were placed in an ice water bath and removed singly from the ice bath, and 2 ml of molten top agar containing 90 nmole of both histidine and biotin was added. The molten suspensions were mixed immediately with a Vortex mixer and poured onto minimal glucose plates. The plates were incubated at 37°C in the dark for 48 hr and counted using an automatic colony counter. Genetic markers for the strains were routinely verified. All extraction and testing were carried out in a room fitted with yellow fluorescent lights to minimize potential photooxidation. Urine samples were collected in washed, sterile amber-colored glass or polyethylene bottles and immediately placed on ice and stored at -213°C. Urines were extracted with XAD-2, following the methods of Yamasaki and Ames (1977) and Putzrath and co-workers (1981). Urine extracts were tested for mutagenic activity in batches with their own PBS controls. All urine samples or PBS control samples were tested in duplicate at a minimum of three urine equivalent doses, 2.5, 5, and l0 ml, with and without $9. The mutagens 2-aminofluorene and benzo[a]pyrene served as positive controls for + $9, and 2-nitrofluorene and 4-nitroquinolone-Noxide served as positive controls for - $ 9 assays. All positive control mutagens were tested at three doses in triplicate. Four or five plates were routinely used for the DMSO standard which provided the 0.0 concentration. Only single values are presented for all replicate samples. Thus, the sample size for all samples reflects the number of specimens collected. Mutagenicity (revertants per milliliter equivalent of urine) was determined from a straight-line least-squares regression, followed by subtraction of the slope from the extraction control PBS sample. Points starting from the highest dose were dropped until the t test for quadratic curvature was "nonsignificant" at p > 0.15. Urinary concentration of cotinine was determined by P. Jacob, L. Yu, and N. Benowitz, using a multistep extraction procedure followed by gas chromatographic analyses (Jacob et al., 1981). Concentrations of urine markers were standardized for urine creatinine concentration. Statistics Diesel exhaust exposure was evaluated as a predictor of urinary mutagenicity in
URINARY MUTAGENICITYAND DIESEL EXHAUST
137
three ways: (1) by evaluation of mean urinary mutagenicity among samples from workers classified by a priori levels of diesel exposure as estimated by job categories; (2) by evaluation of urinary mutagenicity levels among ordered strata of diesel-exposed workers as estimated by ARP; and (3) by a full multiple regression model in which the effect of "diesel exposure" was expressed as an interaction between the amount of respirable particles and nicotine in the air and work at jobs with known potential for diesel exposure. Results were adjusted for other predictors of mutagenicity as determined by multiple linear regressions. To accommodate the multiple observations per person that were taken in the study, an estimated generalized least-squares program (BMDP3V) was used, which allowed for within- and between-person variance components. All analyses were done by SAS 5.16 (SAS Institute, Inc., 1985) or BMDP (BMDP Statistical Software, 1983) procedures on a VAX 11/750 computer. RESULTS
The 87 subjects studied had a mean age of 47.2 -+ 9.7 (x -+ SD) and mean years of railroad work of 16.1 + 10.8 (Table 1). Engineers tended to be older than subjects in other job categories, which is consistent with railroad career paths. Thirty-six percent of subjects (n = 31) were cigarette smokers. Four nonsmokers and two smokers used other forms of tobacco. This yielded 95 urine samples for cigarette smokers (32%), 10 samples for cigarette smokers who used other tobacco products (3%), 189 non-tobacco user samples (62%), and 12 samples from subjects who used other tobacco products (4%). Results for cigarette smokers who used other tobacco products were similar to those for cigarette smokers and, therefore, were analyzed together. Results for urine analyses of non-cigarettesmoking subjects who used other tobacco products were excluded or considered separately. The highest rate of tobacco use was among the carmen studied (72% of subjects), with lower rates of smoking present for the other job categories (2%50% of subjects) (Table 1). Among smokers, the average number of cigarettes smoked during the day prior to urine sample collection was 14.4 -+ 8.1. For most analyses the number of cigarettes smoked during the study day up to the time of urine collection was used, since it was felt to better reflect markers present in the urine at the end of the shift than the number of cigarettes smoked only during the shift. Some samples were missing or not analyzed from each source in this large, multidisciplinary field investigation. The reasons were different for the specific variables. The number of samples not analyzed ranged from 4% for respirable particle concentrations to 12% for urinary cotinine. The high number of samples not analyzed for cotinine occurred because it was measured after all other urine assays, and there was often an insufficient quantity of urine remaining. Mutagenicity was not measured with $9 in 6% of samples and without $9 in 10% of samples, in most cases because of an inadequate quantity of urine. Diesel Exhaust and ETS Exposure A total of 303 (99% of collection days) personal exposure samples were collected. However, 8 samples were missing particulate mass values due to filter
138
S C H E N K E R ET A L.
damage and 23 were missing nicotine values. Use of medians for exposure variables (Table 2) adjusts for the skewness commonly seen in the distribution of airborne exposures, which was also present in our data. The carmen had the highest exposures to RSP, even though they have no diesel exhaust exposure. This resulted in part because some of them were exposed to particulate matter generated by welding or by using acetylene torch for cutting and burning chassis of train cars during repairs, an activity that could occupy 1/2 to 4 hr per day. A large difference in RSP was seen in the average exposures of the carmen welders and nonwelders (Table 2), confirming the contribution of weldingassociated exposures to the RSP exposures among carmen. Despite the expected difference in diesel exhaust exposure between the shop workers and the clerks, the RSP concentrations of these job groups (Table 2) were similar, which is consistent with our previous findings from U.S. railroads (Woskie et al., 1988a; Hammond et al., 1988). This similarity of RSP concentrations between job groups was largely due to the RSP mass contributed by ETS for the clerks. Median concentrations of ETS were highest among the carmen, who had the highest prevalence of smoking, with high concentrations also present among the clerks and the engineers. The relative ranking of the jobs by ARP (Table 2), excluding carmen, was close to that expected by the relative contact these groups have with operating diesel locomotives, assuming that all workers are exposed to a background air concentration of about 30 p,g/m 3. Specifically, highest ARPs were seen for the shop workers, who have the highest exposure to diesel exhaust, and lower ARP levels were obtained for the other job categories, which have little or no diesel exposure. This ranking and ARP levels are similar to those obtained in our previous study of railroad workers (Woskie et al., 1988a; Hammond et al., 1988). However, this cannot explain all of the differences seen between the job groups. Daily exposures for individual study days were combined into strata of ARP for regression analyses of urinary mutagenicity and diesel exposure. Workers in groups known a priori not to be diesel exposed (clerks, carmen) were placed in the unexposed strata (Woskie et al., 1988a; Hammond et al., 1988). Cutoff values for the strata of ARP for smokers and non-tobacco users were 77 and 155 ixg/m3, selected to create approximately equal-size strata. The median levels of ARP for the low, intermediate, and high strata of nonsmoking diesel-exposed workers analyzed by regression were 50, 110, and 181 Ixg/m3, respectively. A similar relationship between ARP and diesel exposure was seen for the smoking workers, with median levels of ARP in the three strata of 10, 115, and 222 txg/m3. Urinary Mutagenicity and Diesel Exhaust Exposure Urinary mutagenicity was first evaluated by job group among nonsmokers (Table 3). No association was present for job grouping and urinary mutagenicity +$9 according to a priori classification of diesel exhaust exposure. Highest levels of urinary mutagenicity among the nonsmokers were seen in the carmen welders, probably due to their welding exposures. Similarly, no association was present among cigarette smokers for mutagenicity +-$9 and diesel exposure as estimated by job category (Table 3). Regression models were run separately for smokers and for nonsmokers, for
URINARY
MUTAGENICITY
AND
DIESEL
139
EXHAUST
r-=_-
0
o
2
0
u.l
? o
0.05). None of the differences among the three strata of diesel-exposed workers or comparisons with the unexposed workers was statistically significant. Regressions were also run excluding all carmen, with results similar to those presented (results not shown). Mutagenicity - $9 also failed to demonstrate a dose-response relationship with diesel exposure strata. The mean level of adjusted mutagenicity ( - $ 9 ) in the non-diesel-exposed and three strata of ARP (low to high) among cigarette smokers were 0.48, 0.20, 0.20, and 0.34, respectively (Table 4). As with the results with $9, TABLE 4 CRUDE AND ADJUSTED MUTAGENICITY FROM MULTIVARIATEREGRESSION MODELS BY STRATA OF ADJUSTED RESPIRABLE PARTICLE (ARP) CONCENTRATIONS
Mean
2SE interval of adjusted
Crude ~
Adjustedb
0.33
0.46
0.26, 0.66
0.37 0.37 0.34
0.37 0.34 0.12
0.17, 0.56 0.19, 0.49 - 0 . 1 5 , 0.38
0.14
0.16
- 0.01, 0.33
0.06 0.11 0.10
0.10 0.12 0.03
- 0 . 0 6 , 0.26 - 0.00, 0.25 - 0.18, 0.24
4.16
3.18
2.37, 4. l0
3.78 3.91 4.30
3.12 3.93 3.54
2.06, 4.39 2.85, 5.17 2.09, 5.33
0.46
0.48
0.23, 0.72
0.28 0.18 0.30
0.20 0.20 0.34
- 0 . 1 0 , 0.50 - 0.06, 0.46 - 0 . 0 4 , 0.71
Non-tobacco users +$9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP - $9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP
Smokers +$9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP -$9
Diesel unexposed Diesel exposed Low ARP Medium ARP High ARP
a Average observed values. b Non-tobacco user values adjusted for urinary cotinine, ambient nicotine, and positive batch con-
trol 2-aminofluorene. Smoker values adjusted for urinary cotinine and cigarettes smoked on study day and positive batch control 2-aminofluorene.
142
SCHENKER ET AL.
there was no suggestion of an association between diesel exposure and mutagenicity without $9 among cigarette smokers. Among non-tobacco users, multiple regressions using stratification of ARP also revealed no evidence of a dose response for mutagenicity with or without $9 (Table 4). Mean adjusted urinary mutagenicities (+$9), for the non-dieselexposed and low, medium, and high strata of ARP (diesel) were 0.46, 0.37, 0.34, and 0.12, respectively (Table 4). A similar absence of dose response was seen among non-tobacco users for mutagenicity - $ 9 . Similarly, in the full regression models (not shown), no additional contribution was made to mutagenicity ___$9 by dichotomous diesel exposure status (yes/no) or by the interaction of diesel exposure status (yes/no) with respirable nicotine or with total respirable particles. Among non-tobacco users the estimated between-person variances were approximately one-fourth the magnitude of the within-person (between-sample) variances, indicating a definite lack of independence in observations taken from the same worker. To examine more closely the possible effects of diesel exposure independent of individual differences and the effects of ETS, we sought to identify a subset of workers who (1) had jobs with potentially high diesel exposure, (2) were nonsmokers, (3) were exposed to little or no ETS (based on nicotine on personal samplers), and (4) had been sampled repeatedly. Three repair shop workers met these criteria, each of whom had been sampled on 8 to l0 different days. Plots for the repeated samples on these three workers of urinary mutagenicity +-$9 (revertants per micromole of creatinine) versus the marker of diesel exposure (ARP) showed no suggestion of a positive association (Figs. 1 and 2). This provides additional support for the regression results showing no association between diesel exposure and urinary mutagenicity among nonsmokers. 2-
O
E i []i
• m&
!
•
1(3 170
•
IO 206
[]
ID 285
~gD 0
.0
0
100
1
-100
200
300
400
500
6OO
Adjusted RSP (pg/m 3) FIG. 1. Urinary mutagenicity +$9 versus respirable particle concentration adjusted for environmental tobacco smoke (ARP, see test) for three nonsmoking shop workers with repeated measurements and low environmental tobacco smoke exposures.
143
URINARY MUTAGENICITY AND DIESEL EXHAUST
== L~ ®
E ::3.. t3
=& • •
o
o~
t
-l-
=
I
¢=
a
0
100
•
1D 1 7 0
•
ID206
a
ID285
C
-1 -100
200
300
400
500
600
Adjusted RSP ( p g / m 3 ) FIG. 2. Urinary mutagenicity - $ 9 versus respirable particle concentration adjusted for environmental tobacco smoke (ARP, see text) for three nonsmoking shop workers with repeated measurements and low environmental tobacco smoke exposures.
The "protective food" variable (ingestion of cabbage, brussels sprouts, or fish in the prior 2 days) was associated with lower urinary mutagenicity only with $9 among smokers and non-tobacco users, but inclusion of this variable did not alter the absence of an association of diesel exposure category and mutagenicity (data not shown). DISCUSSION Based on the use of exposure measurements to estimate diesel dose and markers of exposure to potentially confounding exposures (active and passive smoking), there was no association between postshift urinary mutagenicity and diesel exhaust exposure in this population. This absence of an association does not mean that diesel-exposed workers did not excrete diesel-associated mutagens. Factors which are important in evaluating this result include the actual dose of dieselassociated mutagens, limits of the mutagenicity assay sensitivity, pharmacokinetics of mutagen handling including absorption and excretion, interindividual variability, and uncontrolled sources of urinary mutagenicity such as diet. The median level of RSP among nonsmoking subjects in our highest exposed job group (shop workers) was only 132 txg/m3, and 75% of the samples were below 200 txg/m 3. These levels are lower than many historical measurements of diesel exhaust exposure among railroad workers, and probably reflect the decline in U.S. railroad activity with associated reductions in diesel engine use (Woskie et al., 1988b). The use of personal measurements for RSP concentrations adjusted for ETS improved the accuracy of our diesel exposure estimates, but some error in exposure assessment likely remained (Hammond et al., 1988). Another problem with
144
S C H E N K E R ET A L.
using ARP as a measure of diesel exhaust exposure is that RSP has sources other than diesel and ETS, and particles from those sources are included in ARP. This is illustrated by the high ARP levels among carmen, who are known not to be exposed to diesel exhaust. It is likely that this imprecision would reduce the ability to observe an association of diesel exhaust with urinary mutagenicity if one were present, but it is not possible to quantify the magnitude of this effect. The development of improved markers of diesel exhaust exposure such as elemental carbon or mutagenic activity of respirable particles (Kado et al., 1987) should increase the ability to detect an association of exposure and biologic markers of dose. However, accurate markers for complex mixtures such as diesel exhaust remain a difficult challenge (Hammond et al., 1988; Hulka and Wilcowsky, 1988). It is unlikely that a unique marker will be identified for diesel exhaust, similar to the relationship of nicotine to tobacco smoke, but improved estimates of diesel dose will improve the sensitivity of epidemiologic studies. Recently, Zaebst and co-workers (1991) demonstrated that the elemental carbon content of particles is a promising marker for exposure to diesel exhaust. Diesel exhaust-associated mutagens may have been present in the urine, but below the level of detection in the afternoon urines. For example, a worker exposed to a mutagen concentration of 1000 revertant equivalents per cubic meter of air who breathed 7-10 m 3 during the 7-hr exposure period received a total mutagen dose of approximately 7000 to I0,000 revertant equivalents (1000 rev/m 3 x 7 to 10 m3). The 1000 revertant equivalents per cubic meter of air was estimated from a diesel shop worker's personal filters (Kado et al., 1987). If we assume a small fraction of mutagens are excreted in the urine in mutagenic form, then the measurable revertant equivalents can be approximated. This may not be an unreasonable assumption since the mutagens absorbed on diesel particulate matter will have to be desorbed in physiologic fluids. The bioavailability of mutagens on diesel particulate matter has been reported to be a relatively slow desorption process (King et al., 1981). Belisario and co-workers (1984), who injected rats with whole diesel particles and measured urinary mutagenic activity, found that only a small fraction of measurable mutagens (in this case, less than 5%, approximately) are excreted in a 24-hr period. For illustrative purposes, if 1% of the total revertant equivalents are excreted in the afternoon urine sample collected, 70 to 100 revertant equivalents would be in 100 to 200 ml of urine (0.01 x 7000 to 10,000 rev eq). The concentration would therefore be at the most 0.7 to 1 rev eq/ml urine (70 to 100 rev eq/100 ml urine). This is the level of activity generally determined in the urine of nonsmokers (Kado et al., 1983). Another important consideration in evaluating these results is the unknown kinetics of diesel exhaust absorption and excretion in humans and the limited ability to evaluate them in the field setting. For example, it was feasible in this field study to collect only a single spot urine at the end of the work shift. While the spot urine samples showed a good dose-response relationship with smoking dose on the workday, if diesel mutagens were excreted more slowly than mutagens from cigarette smoke, the assay might not have been sensitive enough to detect the diesel exhaust mutagens. It is also possible that the diesel mutagens are in part
URINARY MUTAGENICITYAND DIESEL EXHAUST
145
excreted by another route, such as the feces, or are excreted in a form which is not mutagenic in the urine. The diesel-associated mutagens may be conjugated to polar molecules such as glucuronides or sulfates, which may not be mutagenic and may not be efficiently extracted by XAD-2. Finally, the study power may not have been great enough to observe an association between diesel exhaust and urinary mutagenicity, if one were present, but we do not believe this explains our negative findings. Study power was evaluated for analyses -+$9 in both smokers and nonsmokers. Multiple regression models with four diesel exposure groups were employed. Power was calculated using the three-degrees-of-freedom partial F test of differences in group means, after adjustment for other factors. For non-tobacco users, the mean mutagenic activity (+ $9) was about 0.3 rev/~mole creatinine with a standard deviation of 0.4. The adjusted + $9 means (Table 4) had a maximum difference of about 0.1 rev/p~mole creatinine. If the unexposed group is considered as baseline, assuming true mean differences from baseline of 0.1, 0.2, and 0.3 corrected counts in the three ordered diesel groups, these differences would have been detected with 93% probability in this study. For smokers the maximum difference between adjusted means (Table 4) was about 0.4 rev/wmole creatinine, and 75% of individual observations fell between 1.9 and 5.4 rev/p~mole creatinine. If the true differences from the baseline mean were + 0.2, + 1.5, a n d + 2.0 (lowest to highest exposure), statistical significance at the 0.05 level would have been declared 84% of the time. The above estimates of power are liberal for nonsmokers because of the independence assumption. They are conditional on the observed values of the covariates in our data and on the estimated coefficients for the covariates. Thus, we do not think that a lack of statistical power can explain our findings, but we cannot exclude that possibility. Our findings are consistent with those of Willems et al. (1989) who reported on the urinary and fecal mutagenicity of car mechanics exposed to diesel exhaust. They used the standard plate incorporation test (Ames et al., 1975), and found that no exposure to diesel exhaust (based on job category) resulted in increased mutagenicity in the urine or feces. The net result of the factors discussed above is that detection of elevated urine mutagenicity resulting specifically from diesel exhaust exposure was not possible despite adequate statistical power in this large field study. However, some important findings on urinary mutagenicity in working populations derived from the scope of this study. First, measuring detectable mutagenicity "baseline" was possible for most of the nonsmoking subjects. Second, we were able to estimate personal exposure to diesel exhaust in a field study by adjustment of respirable particle exposure for particulate mass from environmental tobacco smoke. The use of a biologic marker such as urinary mutagenicity to measure diesel exhaust represents a common difficulty in epidemiologic studies of exposure to complex mutagen and carcinogen mixtures of this type. More specific and sensitive assays including molecular markers could increase the chances of success. This study provides no basis upon which to draw conclusions regarding cancer risk associated with diesel exhaust exposure. Other epidemiologic and animal studies have directly addressed this issue (Schenker, 1989; Ishinishi et al., 1986;
146
SCHENKER ET AL.
M a u d e r l y e t al., 1987; G a r s h i c k e t al., 1987, 1988), a n d a r e m o r e a p p r o p r i a t e t o evaluate such an association.
ACKNOWLEDGMENTS Research described in this article was conducted under contract to the Health Effects Institute (HEI), an organization that supports the conduct of independent research and is jointly funded by the U.S. Environmental Protection Agency (EPA) and automotive manufacturers. Publication here implies nothing about the view of the contents by HEI or its research sponsors. Although the work described in this document was funded in part by the U.S. Environmental Protection Agency under Assistance Agreement X808859 with HEI, the contents do not necessarily reflect the views and policies of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
REFERENCES Ames, B. N., McCann, J., and Yamasaki, E. (1975). Methods for detecting carcinogens and mutagens with the Salmonella/mammalian-microsome mutagenicity test. Mutat. Res. 31, 347-364. Atkinson, A. C. (1985). "Plots, Transformations and Regressions." Oxford Univ. Press (Clarendon), London. Belisario, M. A., Buonocore, V., DeMarinis, E., and DeLorenzo, F. (1984). Biological availability of mutagenic compounds absorbed onto diesel exhaust particulate. Mutat. Res. 135, 1-9. BMDP Statistical Software (1983). "Printing with Additions." University of California Press, Berkeley. De Meo, M. P., Dumenil, G., Botta, A. H., Laget, M., Zabaloueff, V., and Mathias, A. (1987). Urine mutagenicity of steel workers exposed to coke oven emissions. Carcinogenesis 8, 363-367. Everson, R. B. (1986). Detection of occupational and environmental exposures by bacterial mutagenesis assays of human body fluids. J. Occup. Med. 28, 647-655. Falck, K., Sousa, M., and Vainio, H. (1980). Mutagenicity in urine of workers in rubber industry. Mutat. Res. 79, 45-52. Garshick, E., Schenker, M. B., Munoz, A., Segal, M., Smith, T. J., Woskie, S. R., Hammond, S. K., and Speizer, F. E. (1987). A case-control study of lung cancer and diesel exhaust exposure in railroad workers. Am. Rev. Respir. Dis. 135, 1242-1248. Garshick, E., Schenker, M. B., Munoz, A., Segal, M., Smith, T. J., Woskie, S. R., Hammond, K., and Speizer, F. E. (1988). A retrospective cohort study of lung cancer and diesel exhaust exposure in railroad workers. Am. Rev. Respir. Dis. 137, 820-825. Griffith, J., Duncan, R. C., and Hulka, B. S. (1989). Biochemical and biological markers: Implications for epidemiologic studies. Arch. Environ. Health 44, 375-381. Hammond, S. K., Leaderer, B. P., Roche, A. C., and Schenker, M. (1987). Collection and analysis of nicotine as a marker for environmental tobacco smoke. Atmos. Environ. 21, 457--462. Hammond, S. K., Smith, T. J., Woskie, S. R., Leaderer, B. P., and Bettinger, N. (1988). Markers of exposure to diesel exhaust and cigarette smoke in railroad workers. Am. Ind. Hyg. Assoc. J. 49, 516-522. Howe, G. R., Fraser, D., Lindsay, J., Presnal, B., and Yu, S. (1983). Cancer mortality (1965-1977) in relation to diesel fume and coal exposure in a cohort of retired railroad workers. J. Natl. Cancer Inst. 70, 1015-1019. Hulka, B. W., and Wilcowsky, T. (1988). Biologic markers in epidemiologic research. Arch. Environ. Health 43, 83-89. International Agency for Research on Cancer (IARC) (t989). "IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Diesel and Gasoline Engine Exhausts and Some Nitroarenes," Vol. 46. IARC, Lyon. Ishinishi, N., Koizumi, A., McClellan, R. O., and Stober, W. (Eds.) (1986). "Carcinogenicity and Mutagenicity of Diesel Engine Exhaust." Elsevier Science, New York. Jacob, P., Wilson, M., and Benowitz, N. L. (1981). Improved gas chromatographic method for the determination of nicotine and cotinine in biologic fluids. J. Chromatogr. 222, 61-70.
URINARY MUTAGENICITY AND DIESEL EXHAUST
147
Kado, N. Y., Guirguis, G. N., Flessel, C. P., Chan, R. C., Chang, K., and Wesolowski, J. J. (1986). Mutagenicity of fine (
