A field method for measuring solvent vapors in exhaled air--application to styrene exposure.

A field method for measuring solvent vapors in exhaled air - application to styrene exposure. Scand J Work Environ Health 1991: 17:195-204. A method is described for measuring solvent vapors in mixed-exhaled air. The subject ex hales through a carbon-containing tube connectedto a Wright respirometer. Adsorbed vapors are subse quently eluted by carbon disulfide and analyzed by gas chromatography. Twenty-minute exposures to styrene and the corresponding concentrations of styrene in the breath and venous blood were repeatedly measured for two subjects. Regression analysesindicated that the breath measurements were highly cor related with both the exposures and the blood concentrations of styrene. In another study, styrene was measured simultaneously in the mixed-exhaled air by this technique and in the end-exhaled air by a portable gas chromatograph. The mixed-exhaled air obtained with this method contained about half alveolar air. Analysisof the components of the variance obtained from all the data indicated that the error in measurement by this method was about one-fourth of the total variance.

Biolo gical monitoring of workers expo sed to or ganic solvent s is desirable because it allows the uptake o f the substances to be estimated regardless of the route o f expos ure. Of the currently used techniques, mea sur ement o f the vapor in exhal ed air is particularly attra ctive because it is noninvasi ve. [See recent reviews by Dro z & Guillemin (1) and Wilson (2)]. Breath sampling in field studies has relied upon the collect ion of mixedor end -exhaled air in either glass tubes (eg, references 3 and 4) or in bags made of inert polymers or metal foils (eg, reference 5). After the collection of exhal ed air, the analyte can be measured directly in the field if a suita ble instrument (eg, that presented in refer ence 6) or analytical method (5) is available. However , the receptacle containing the analyte is generall y transported to a laboratory for anal ysis. In some cases, the analyte is pre con centrated from the sampling device by the withdrawing of air from the receptacle through an appropriate sorbent (as, eg, in references 7 and 8).
In contemplating a large field investigation of styrene exposure in th e reinforced plastics industry, we evaluated several of the existin g techniques for measuring styrene vapor in exh aled air and found them un suit abl e for two reasons. First, we required a technique which would allow between 100 and 200 breath samples to be gathered in a single day. This necessity made it impractica l to collect breath in bag s or other bulky containers or to preconcentrate styrene from such receptacles. Second, we requi red a method which would allow samples to be easily transported and stored for several days prior to analysis. Our experience with laboratory trials ind icated that styrene vapor was ab sorb ed by even " inert" polymers such as Teflon and was irreversibly lost in glass containers if not analyzed within hours of collection.
In seekin g an alternative field procedure, we were intrigued by the description of a method in which solvent vapor was concentrated directly from the breath as the subject exhaled through a tube containing activated coconut carbon (personal communication, 1986, J Fajen , National Institute for Occupational Safet y and Health, Cin cinnati, Ohio, Un ited States). Since coconut carbon is used routinely for the air monitoring of styren e and other or ganic solvents in the workplace (9), and because styrene vapor has been sho wn to be stable following adsorption by coconut carbon (10), we thought that such a technique would be am enable to the collection of large numbers of samples and would allow samples to be stored for several days prior to an alysis. In what follow s we report the application of thi s technique to measure styrene repeatedly in mixed-exhaled air obtained from two subjects in the workplace. Preliminary results are also pre sented from a larger investigation in which mixedexhaled air and end-exhaled air were measured simultaneously for 27 workers who were exposed to styrene.
Th e studies were designed to determine whether coconut carbon could be used in the field to determine the levels of styrene in the breath accurately. If this were the case, we expected to observe that the exhaled air concentration obtained by this technique would be proportional to the concentration of styrene measured simultaneously in venous blood and in end-exhaled (alveolar) air . We also anticipated that, because styrene is rapidly cleared from the blood (11), the concentration of styrene in exhaled air would be related to the corresponding exposure during the 20 min or so preceding the breath measurement.

Experimental design
Two studies were performed. In the first, serial samples of environmental air, exhaled air, and blood were collected from two subjects in a factory where styrene was used in the manufacture of reinforced fiber glass products (bathtubs and showers) . The subjects were exposed to styrene for 4 h as they observed the work practices of employees engaged with spraying styrenecontaining resins into mold s and laminating the products by hand. The factory was naturally ventilated with air supplied through large doors and other openings. On the day of the investigation, the ambient temperature was very warm, about 40°C . The subjects' exposures to styrene were measured at intervals of about 20 min with the use of pumps and carbon tubes . Mixed-exhaled air samples were collected in an area free of styrene to ensure that the analyte had been released from the blood and was not merely unabsorbed styrene from the anatomic dead space . Thus at the end of each exposure the subjects exited the workplace to an outdoor location where the ambient concentration of styrene was negligible . There, the blood and breath samples were collected and the carbon tubes were changed for the next determination. The mean interval that the subjects were absent from the workplace between exposures was 3.8 (SD 1.3) min. After 4 h of exposure, additional samples of blood and breath were obtained at 20-min intervals for 2-3 h so that the elimination rate of styrene could be determined. During this period the subjects were more-orless sedentary.
The second study was designed to determine the relationship between the concentration of styrene in the mixed-exhaled air , as measured with the new method, and the concentration of styrene in the end-exhaled air (alveolar air), measured with a portable gas chromatograph. Samples were obtained from 27 workers who were employed in a factory where fiber glass boats were manufactured. The subjects were taken to an area free of styrene where mixed-and end-exhaled air samples were obtained within a few seconds of each other.

Subjects
In the first study, two healthy subjects volunteered with informed consent to participate. One subject was male, 29 years of age, and 170 ern tall, and weighed 63.9 kg (l6070 fat). The other subject was female, 26 years of age, and 152 em tall, and weighed 52.0 kg (l6 % fat). Neither subject had been significantly exposed to sty-196 rene in the month preceding the investigation. In the second study, samples of mixed -and end -exhaled air were obtained with informed consent from 27 workers of both sexes who were exposed routinely to styrene .

Styrene in environmental air
The subje cts' exposures to styrene were measured by personal monitoring with battery-operated pumps and tubes containing 150 mg of coconut carbon (number 226-35, lot 120, SKC West, Fullerton, California, United States). The air flow rate was 0.6 l/min. After the collection, the carbon tubes were sealed with polyethylene caps and stored at 4°C in an area free of styrene. Thirteen days after the collection, the carbon sections (l00 mg primary section, 50 mg backup section) were removed from the tubes and placed in 4-ml glass vials. Three milliliters of carbon disulfide (Omnisolve, liquid chromatographic grade) was added to the carbon, and the vials were capped with Teflon-lined rubber septa. After 1 h of gentle agitation at room temperature , the solutions were analyzed by gas chromatography . One-microliter aliquots of the solutions were injected into a Varian model 3700 gas chromatograph equipped with a flame ionization detector. The glass column was 2 mm (inner diameter) x 2 m, packed with 10 % SP-2100 on 801100 Supelcoport (Supelco, Bellefonte, Pennsylvania, United States). The carrier gas was nitrogen at a flow rate of 38 ml/min. The column , injector, and detector temperatures were 95°, 210°, and 240°C, respectively . The peak areas of the samples were measured with a Varian Vista 401 data system and compared against those obtained from analytical standards prepared by the injection of known amounts of styrene (Aldrich , 99 % + Gold Label) into carbon disulfide. The samples were corrected for the desorption efficiency of styrene from carbon, as will be described. The analysis of the backup sections of thc carbon tubes indicated that there was no breakthrough of styrene.

Styrene in blood
Blood was collected from the subjects through an indwelling catheter inserted into the brachial vein at the beginning of the experiment. At the end of each exposure between 8 and 10 ml of blood was obtained in heparinized containers. Approximately 1 h after the cessation of exposure, blood was transferred to tared glass vials (20-ml capacity). One milliliter of hexane containing 8.86 ug/rnl of chlorobenzene (internal standard) was added, and the vials were sealed with Teflonlined caps, shaken vigorously and immediately frozen in dr y ice. The samples were maintained at -20°C for 14 d prior to the analysis.
The analytical method was adopted from Karbowski & Braun (12). After thawing, the samples were weighed and centrifuged. The hexane layer was transferred to a 4-ml glass vial with a Teflon -lined septum whereupon Collection ofsamples. Prior to the collection of a sample, the ends of the glass tube containing the carbon were broken with a special tool and smoothed with a piece of carborundum. Then the tube was inserted into the apparatus and a disposable mouthpiece was affixed to the inlet. It was the most convenient for the subject and the investigator to sit facing each other across a small table . During the collection of the sample, the subject was instructed to maintain a comfortable but constant pressure as he or she exhaled through the device. This situation was easily accomplished by allowing the subject to ob serve the pressure reading on the gauge as the sample was obtained. A total of 3.0 1(uncorrected) was obtained for each sample in 0.5-1 increments. The subject was coached to take a breath and then to exhale through the device until instructed to 1~I was injected into a gas chromatograph. The chromatographic conditions were the same as those given for styrene in air except that the column temperature was 75°C . The ratios of the peak areas of styrene to those of the internal standard were compared with a calibration curve obtained from standard solutions containing between 1.8 and 18 ug/rnl of styrene in hexane and the same concentration (8.86Ilg/ml) of the internal standard. The mean extraction efficiency of styrene from blood was 95 (SD 10) 0/0. The quantitation limit of the method was estimated to be 0.05 ug/rnl blood.

Styrene in mixed-exhaled air
Apparatus. Exhaled air samples were obtained with the apparatus illustrated in figure I. The subject forcibly exhaled through a disposable mouthpiece (cardboard number 1021-250, Vacumed, Inc, Ventura, California, United States) which was connected to a I-em (outer diameter) X 8.5-cm glass tube containing 200 mg of 20/40 mesh coconut carbon (tubes were obtained as a special order from SKC West and contained lot number 120 carbon). The outlet from the carbon tube was connected to a Wright respirometer (Haloscale 00-301 , Ferraris Development and Engineering Ltd, London), which measured the volum e of air passing through the apparatus. Connections between the carbon tube and the mouthpiece and the respirometer were made with special parts machined from aluminum and connected by Swagelok (I .O-cm inner diameter) brass fittings and Teflon ferrules.  Correction of air volume. The resistance to exhalation imposed by the carbon tube reduced the volumetric flow rate through the Wright respirometer below the linear range of operation. However, we found that by monitoring and recording the pressure upstream from the carbon tube during exhalation, we were able to correct the volume since the obser ved and true volumes were found to be proportional at a given pressure . Before and after each field experiment, we calibrated the apparatus by connecting the inlet to a source of compressed air and the outlet to a dr y gas meter. Th e air pressure was then increased in small increments over the working range of 21 to 72 mm Hg (19.2-67.0 Pa); at each pre ssure, the observed and true volume s which passed through the respirometer were noted. A typical calibration curve is shown in figure 2. We corrected the volumes by dividing the measured volume by the correction factor, which was the ratio of the obser ved to the true air volume obtained [rom calibration. stop by the investigator, who monitored the volume recorded by the respirometer. In practice about 5-10 s were required for each 0.5-1 exhalation. The subject repeated the process until six exhalations had been performed. The final volume and the pressure maintained during the collection of the sample were recorded by the investigator. The tubes were sealed with polyethylene caps and stored in an area free of styrene prior to the analysis.
Analysis ofsamples. The samples were analyzed 12 d after the collection. Carbon was removed from the tubes, desorbed with 1.5 ml of carbon disulfide, and analyzed by gas chromatography as described for the environmental air samples. The samples were corrected for the desorption of styrene from the carbon as will be described. The quantitation limit was estimated to be 0.8 Ilg of styrene per sample (O.4llg styrene/lOO mg carbon), or about 0.2 ug/I in an air sample with a corrected volume of 4 1. The gas chromatograph, as received from the manufacturer, had a significant memory for styrene which carried over between injections. This memory was traced to the absorption of styrene in the inlet and transfer lines, which were constructed of Teflon tubing, 198 and to a Teflon switching valve which was an integral part of the instrument. By replacing the tubes with stainless steel, we were able to reduce the memory to about 15 070 of the value from the previous injection.

Styrene in end-exhaled air
The sample concentrations were adjusted for residual styrene by injecting clean air (room air drawn through 1.2 g of 20/40 mesh activated coconut carbon) between samples so that the background levels could be determined. The precision of the method was estimated to be between 5 and 10 % as a coefficient of variation (six determinations with between four and six measurements/determination). The quantitation limit of the method was 1.7 ug/l, Determination of desorption efficiency Addition ofsolutions containing styrene. Known quantities of styrene (2-300 ug) were added in triplicate to 200 mg portions of SKC carbon (Lot 120, SKC West) in sealed 4-ml glass vials. The styrene was administered in 5 III of carbon disulfide, and the vials were capped with Teflon-lined septa. After 20 h the carbon was desorbed and analyzed by gas chromatography as has already been described.
Collection of styrene vapor. Groups of three exhaled air tubes were exposed to styrene atmospheres produced dynamically. Air was generated at a flow rate of 30 lImin and at 50 % relative humidity and 25°C by a special system (model HCS 202, Miller Nelson Research, Carmel Valley, California, United States). Liquid styrene was injected continuously into the air stream by a syringe pump (Sage Instrument Co, model 355, Cambridge, Massachusetts, United States), containing either a 10-or 100-111 syringe, the needle of which had been inserted through a septum into the air stream. The vapor-air mixture entered a Teflon chamber [3.2 em (inner diameter) x 76 em] from which samples of the atmosphere were withdrawn through ports arranged radially 25 em from the exit. The carbon tubes were connected to the ports with minimal lengths of Teflon tubing. The airflow rate through the tubes was maintained at 0.17 lImin with critical orifices. Five groups of samples were collected at styrene concentrations of either 4 or 25 ug/I for periods ranging between 5 and 42 min. After the collection, the samples were immediately desorbed with carbon disulfide and analyzed by gas chromatography as has already been described.

Desorption efficiency
At low loadings of styrene on the coconut carbon (< 30 Ilg/lOO mg carbon) we observed that relatively large amounts of the adsorbate were not released by carbon disulfide. Thus, while only minor corrections for recovery were required for samples of environmental air C(llg/ ml)

Styrene in air, breath and blood
The results of measurements of serial exposures to styrene for two subjects and the corresponding concentrations of styrene in the mixed-exhaled air and in the blood are shown in figure 4. It is apparent that the breath concentrations measured with the new technique were closely related to the concentrations of styrene in blood samples collected at the same time. The figure also shows that the concentrations of styrene in both the blood and the breath reflected the exposure received during the preceding 20 min.
When the mixed-exhaled air concentrations (MIXED) obtained from each subject were regressed first upon the blood concentrations (BLOOD) and then separately upon the preceding exposures (EXPO-SURE), as shown in figure 5, the relationships revealed (where the loadings were high), significant corrections were needed for samples of exhaled air, for which most loadings of adsorbate were between I and 10 ug of styrene/100 mg of carbon.
The results of the desorption tests are shown in figure 3 as a log-log plot of the data after linearization to the Freundlich adsorption isotherm according to the method described by Rudling (13). The x-axis is the liquid concentration of styrene [C (ug/rnlj], and the y-axis is the residual amount of styrene adsorbed by the carbon [X (/lg/100 mg carbon)]. [Note: X = V (Co-C) [lOO/(weight carbon, mg)], where V is the volume of carbon disulfide and Co is the hypothetical liquid concentration with the assumption of no adsorption]. Figure 3 indicates that, because the fit of the data to the isotherm was good, the underlying relationship can be used to correct samples for recovery.   (table  1). Analysis of the residual errors from the regressions showed no evidence of nonlinearity and indicated that the residuals were normally distributed with constant variances over the ranges of BLOOD and EXPOSURE which were encountered. Since none of the intercepts from the simple regressions shown in table I were significant, we repeated analyses by forcing the regressions through the origins. Then the slopes, obtained from the two data sets, were averaged to yield the following relationships: where the units of the three variables are all equivalent. Although these equations were based upon data obt ained from only two subjects, and should therefore be considered as preliminary, they are used to explore other relationships in subsequent sections of this paper.

Elimination oj styrene in breath
Dur ing the 2-3 h immediately following exposure to styrene, the concentrations of vapor in the breath should be governed by first-order kinetics with a decay curve which is dominated by the release of styrene from a large central compartment (II ). In figure 6 the logarithms of the concentrations of styrene in the mixed-exhaled air were plotted against time after exposure. In both cases, the fit of the data to the singlecompartment model was good with r 2 = 0.90 for sub-

Mixed-versus end-exhaled air
It is generally assumed that the mixed-exhaled air of a subject breathing normally is comprised of 70 070 alveolar air and 30 % inhaled air (14). Since the method of breath collection described in this report imposes a restriction on normal breathing which results in a protracted exhalation of fixed volume (0.5 I uncorrected for pressure), we wished to determine empirically the proportion of the mixed-exhaled air which was alveolar. Thus we investigated pairs of samples of mixed and end-exhaled air which had been collected from 27 workers exposed to styrene.

Partitioning the variability of breath measurements
The presented data indicate that the variability of the breath concentrations of styrene was relatively large. For a better perspective of the results, it is necessary to partition the total variation into several parts. This partitioning can be accomplished with an analysis-ofvariance model in which it is assumed that there are three major components of variability, those associated with changes in breath concentration within individuals over time, those associated with consistent differences between individuals, and those associated with errors in measurement. If it is assumed that each of these sources of variability is lognormally distributed, then the total variation can be accounted for conveniently as follows (16): where S2 L , T ' S2 L,w, S\,B' and S\,M represent the estimated variances of log-transformed data for the total distribution, the intra-and interindividual distributions, and the distribution of measurement errors, respectively.
The presented data allow the three components of the total variance to be estimated. The error in measurement can be estimated from the data regarding the elimination of styrene in the breath, shown in figure  6, if it is assumed that the residual error arises entirely from the collection and analysis of breath samples . The estimated intraindividual variance can be derived from the serial breath measurements obtained from two subjects (figure 4), since the variance of these data represents the sum of s\, w + S2 L , M' Taking logarithms of the data and estimating sample variances, we obtained values of [S\,W+S2L,M]=0.221 and 0.252 for subjects 1 and 2, respectively. These values can also be pooled to yield a single estimate of [S2 L. w + S2 L • M] = 0.236. Subtracting the measurement error of S\,M = 0.091, we obtained the value of S2 L •w = 0.145. Since the interindividual variance reflects the variation in exposure which occurs over short periods, it is likely that different workplaces and jobs will result in different variances (17). Thus the estimate of s\,w= 0.145 should only be considered representative of continuous exposure in a naturally-ventilated workplace. Fortunately, much of the heaviest exposure to styrene occurs in the manufacture of reinforced plastics in such workplaces; therefore it seems appropriate to use this value to represent the intraindividual variance in the model.
The interindividual variance can be estimated from the regression of In(MIXED) on In(END) shown in figure 7 B if it is assumed that the residual error arises from measurement error and from consistent differences in the exhalation of styrene between people. In this case the residual error is equivalent to [s\ B+ s\ M] = 0.195. Therefore, by subtracting the err~r in m~asurement of S\'M =0.091, we obtained a value of s\'B=0.104. Finally, by taking the sum of the three components, we estimated the total variance to be s\'T=0.091 +0.146+0.104=0.340. The individual variances can now be expressed as proportions of the total, where S2 L,M/s\,T=O.266, s\.W/S2L,T= 0.428, and s\. B/ s\ .T= 0.305 . Thus the error of measurement should contribute only about one-fourth of the total variation observed in the breath samples when s\, w reflects the variability in breath measurements within individuals, which is associated with short-term variation in exposure. If this method were applied to breath concentrations at the beginning of work, following 16 h of nonexposure, then s\.w would result from the presumably smaller variation in exposure which occurs over longer periods (1). Then s\.w would contribute between about one-fourth and onehalf of the total variance.

Concluding remarks
The method described in this paper for the measurement of mixed-exhaled air should be suitable for predicting either the uptake of or exposure to styrene over the ranges of concentrations typically encountered in the workplace. This conjecture is supported by a longitudinal study of 48 styrene-exposed workers in which mixed-exhaled air measurements were found to be significantly correlated with long-term exposure and with sister chromatid exchanges in peripherallymphocytes (18).
The precision of this method, estimated by S2L, M = 0.091, represents the combined random errors from all sources related to the collection and analysis of the samples, including use of the Wright respirometer in the field and desorption of the carbon and gas chromatography. The precision can also be given in terms of the coefficient of variation (CV) by employing the relationship CV = [exp(s2 L , M -1) v,] =0.31. Although a coefficient of variation of 31 070 is relatively large when compared with that associated with the measurement of styrene in a single sample of environmental air (5.8 %) (9), it should be repeated that this error contributed no more than one-fourth of the total variability in the field measurements. The other threefourths of thetotalvariability arose from factors which were not related to the methodology but rather to temporal changes in exposure and to differences between individuals. Since these latter variables ultimately dominate the behavior of a breath monitoring program, it is clear that many measurements are required before occupational exposures can be assessed properly. Thus it can be seen that the real utility of a simple field monitor such as the one described in this paper lies not in its precision, but rather in its convenience and low cost.
Although the application described in this paper involved exposure to styrene, it should be clear that this method for measuring concentrations of vapor in the breath can easily be extended to other organic solvents which are stable on activated carbon. We have recently applied this methodology to workers exposed to tetrachloroethylene in the dry-cleaning industry (unpublished data).