The Influence of Physicochemical Properties on the Internal Dose Of

Home , Lipophilicity

ORIGINAL ARTICLE The inﬂuence of physicochemical properties on the internal dose of trihalomethanes in humans following a controlled showering exposure

Lalith K. Silva1, Lorraine C. Backer2, David L. Ashley1, Sydney M. Gordon3, Marielle C. Brinkman3, John R. Nuckols4, Charles R. Wilkes5 and Benjamin C. Blount1

Although disinfection of domestic water supply is crucial for protecting public health from waterborne diseases, this process forms potentially harmful by-products, such as trihalomethanes (THMs). We evaluated the influence of physicochemical properties of four THMs (chloroform, bromodichloromethane, dibromochloromethane, and bromoform) on the internal dose after showering. One hundred volunteers showered for 10 min in a controlled setting with fixed water flow, air flow, and temperature. We measured THMs in shower water, shower air, bathroom air, and blood samples collected at various time intervals. The geometric mean (GM) for total THM concentration in shower water was 96.2 mg/l. The GM of total THM in air increased from 5.8 mg/m3 pre shower to 351 mg/m3 during showering. Similarly, the GM of total-blood THM concentration increased from 16.5 ng/l pre shower to 299 ng/l at 10 min post shower. THM levels were significantly correlated between different matrices (e.g. dibromochloromethane levels) in water and air (r ¼ 0.941); blood and water (r ¼ 0.845); and blood and air (r ¼ 0.831). The slopes of best-fit lines for THM levels in water vs air and blood vs air increased with increasing partition coefficient of water/air and blood/air. The slope of the correlation plot of THM levels in water vs air decreased in a linear (r ¼ 0.995) fashion with increasing Henry’s law constant. The physicochemical properties (volatility, partition coefficients, and Henry’s law constant) are useful parameters for predicting THM movement between matrices and understanding THM exposure during showering.

Journal of Exposure Science and Environmental Epidemiology (2013) 23, 39--45; doi:10.1038/jes.2012.80; published online 25 July 2012 Keywords: trihalomethane exposure; water disinfection by-products; human blood; showering; partition coefﬁcients; Henry’s law constants

INTRODUCTION by which DBPs might cause adverse health effects are not well Water disinfection is important to control infectious disease understood, a more accurate assessment of exposure will improve outbreaks such as typhoid, hepatitis, Giardia infection, and cholera the precision of epidemiological studies by reducing exposure arising from the public water supply. However, potentially harmful misclassiﬁcation. chemicals, such as trihalomethanes (THMs), can be formed as The presence of THMs in the domestic water supply leads to 8--13 by-products. THMs are halogenated organic compounds formed exposure to THMs from daily water-related activities, such as during disinfection by the reaction of chlorine with naturally drinking and bathing/showering. THMs in tap water are inhaled occurring organic matter such as humic and fulvic acids. Chloro- following aerosolization or vaporization during showering. In form, bromodichloromethane (BDCM), dibromochloromethane addition, THMs are absorbed dermally during showering and (DBCM), and bromoform are the primary THMs found in tap bathing. Because THMs are lipophilic and metabolized rapidly, the water in the United States.1 levels in blood may vary depending on the time of sample col- 14 Epidemiological studies suggest that exposure to THMs increases lection after exposure. the risk of various adverse health outcomes in humans. Long-term Because of prevalent exposure to THMs from the use of tap 15 exposure to THMs may lead to increased risk of cancers in the water, the US Environmental Protection Agency (US EPA) bladder, stomach, pancreas, kidney, and rectum.2--4 THM expo- established regulatory limits for THM levels in tap water. These sure may also increase the risk of Hodgkin’s and non-Hodgkin’s regulatory changes appear to have resulted in decreased THM lymphoma.2--5 Furthermore, exposure to disinfection by-products levels in both domestic water supply and in US residents using (DBP), including THMs, may be associated with adverse reproduc- that water.16 Estimating human exposure to THMs is complicated tive outcomes such as reduced gestational age, intrauterine owing to variation in routes of exposure (e.g. ingestion, inhalation, growth retardation, and birth defects.6,7 Although the mechanisms and dermal), absorption, distribution, metabolism, and excretion.

1Division of Laboratory Sciences, Centers for Disease Control and Prevention, Atlanta, Georgia, USA; 2Division of Environmental Hazards and Health Effects National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, Georgia, USA; 3Battelle Memorial Institute, Columbus, Ohio, USA; 4Department of Environmental & Radiological Health Sciences, Colorado State University, Fort Collins, Colorado, USA and 5Wilkes Technologies, 10126 Parkwood Terrace, Bethesda, Maryland, USA. Correspondence to: Dr. Lalith K. Silva, Division of Laboratory Sciences, National Center for Environmental Health, CDC, 4770 Buford Highway, NE, Mail Stop F47, Atlanta, GA 30341, USA. Tel.: þ 770 488 3559. Fax: þ 770 488 0181. E-mail: [email protected] The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention. Received 17 October 2011; accepted 8 May 2012; published online 25 July 2012 Blood and air trihalomethane levels after showering Silva et al 40 Measuring metrics of internal dose (e.g. THM concentrations in 25 min after the end of the total exposure period, which is the same as 10 blood) is therefore a useful approach for assessing recent human and 30 min, respectively, after showering ended. It is important to note exposure to THMs.17 Internal dose of THMs is related to many that both the 10- and 30-min post-shower samples were collected during factors, including tap water THM levels; duration and frequency of the elimination phase for THMs, and thus the samples do not represent the water use activity; route of exposure; genetics; metabolic factors; peak blood concentrations but rather occur in the downslope of the blood and THM physicochemical properties (e.g. volatility, partition concentration vs time curve.20 Samples were analyzed for THM levels using coefficient, half-life in blood). headspace solid-phase micro-extraction (SPME) coupled with gas chroma- To better understand how physicochemical properties affect tography (GC) and high-resolution mass spectrometry (MS). Analyte THM uptake during showering, we evaluated THM levels in shower quantification was based on stable isotope dilution.21 water, shower air, and blood of 100 study participants who showered under controlled conditions (e.g. 10-min shower, shower temperature, shower water flow). A summary of our findings Air Samples has been published.10 In this paper, we examine the relationships We collected three air samples near the subject’s breathing zone: a pre- between THM physicochemical properties (Henry’s law constant, exposure instantaneous sample collected in the bathroom (hereafter volatility, and partition coefficients) and post-showering blood referred to as the ‘‘pre-shower’’ air samples), a 10-min time-integrated THM levels. sample collected in the shower stall during the entire 10-min showering period (hereafter referred to as the ‘‘during shower’’ air samples), and a 5-min time-integrated post-shower sample collected in the unventilated METHODS bathroom, starting immediately after the shower ended (hereafter referred Human Subjects to as the ‘‘post-shower’’ air samples). Air samples were collected remotely The institutional review boards of the Centers for Disease Control and using evacuated stainless steel canisters and a continuously flowing 100 Prevention (CDC), the National Institutes of Health, the General Clinical sampling line (copper tubing, 4 outer diameter). Filled canisters were Research Center (GCRC) at the University of Pittsburgh, and Battelle Memorial sealed and shipped to Battelle (Columbus, OH, USA) for analysis. We Institute approved this study protocol. We complied with all applicable analyzed the samples for THMs by automated GC/MS using a modified 22 requirements for protection of human subjects. Study participants gave version of US Environmental Protection Agency method TO-14. Full 23 written informed consent before the study. details of air sampling and analysis procedures are available elsewhere.

Study Design and Participants Water Samples Pittsburgh was chosen for this study, because the tap water there typically The shower head was modified to allow remote water sampling. Duplicate contains measurable levels of the four DBPs of interest. Brominated DBPs samples were collected 5 min after each shower began. Participants were can form during disinfection of water containing natural organic material instructed to set the shower water temperature between 40 and 41 1C. and bromide. Significant levels of bromide are typically found in the Shower water temperature was monitored by the participant via a digital Allegheny River in Western Pennsylvania, possibly because of a combina- thermometer in the shower stall, and remotely by the study staff via 18 tion of natural and anthropogenic sources. The Allegheny River was the wireless transmission to a display outside the bathroom. Water samples source for water for the utility that serviced the GCRC at the University of were collected in borosilicate glass vials containing sodium thiosulfate to Pittsburgh, and thus initial tap water samples contained measurable levels quench further THM formation and phosphate buffer to standardize pH of all four target THMs. between 6.0 and 6.5. We analyzed water samples for THM levels using We measured THM levels in shower water, shower air, bathroom air, and headspace SPME--GC/MS with quantification based on stable isotope blood of 100 study participants who showered under controlled conditions dilution.24 (10-min shower, 40 1C shower temperature, 5.6--6.7 l/min shower water flow). Briefly, we recruited adults from a panel who volunteered to participate in a research project at the GCRC and conducted the study in Statistical Analysis July--September 2004. Eligible subjects (18--45 years of age) had a normal Unweighted estimates (geometric mean, mean, median, SD, SE, and the blood screen, were not pregnant, and were willing to provide blood by range) were calculated using JMP (John’s Macintosh Project) statistical venipuncture. Questionnaire data were collected by direct face-to-face software, version 8.0 (SAS Institute, Cary, NC, USA). Correlation analyses for interview on the day of the showering event. Occupation and data on distribution of THMs in water vs shower air, blood vs shower air, and blood recent THM exposure activities (bathing, swimming, using hot tub or vs water were performed using JMP. The slope and the Pearson’s sauna, washing clothes or dishes) were collected for each participant just correlation coefficient (r) for linear regression for all THMs for 100 study before the exposure activities began. To limit the exposure to THMs participants were obtained using JMP. unrelated to the showering, participants were asked not to flush the toilet or run tap water while in the study area, to dry off and dress within 5 min, and to stay in a separate room, away from the shower room, while waiting RESULTS for the 30 min post-shower blood sample to be drawn. We provided study We measured four THMs in water, air, and blood samples collected participants with THM-free bottled drinking water to be consumed while at before, during, and after controlled showers for 100 participants. the study site to minimize/eliminate oral THM exposure. Further details of 10 Distribution of THM levels (geometric mean and 95% confidence this study design were reported by Backer et al. interval) in shower water collected 5 min after the shower began is presented in Table 1. The shower water THM geometric means Biological Specimen across the full study period were as follows: chloroform (GM Whole-blood samples for THM analysis were collected by a certified 64.1 mg/l); BDCM (GM 19.9 mg/l); DBCM (GM 8.3 mg/l); and bromo- phlebotomist using vacutainer tubes processed to remove THM contam- form (GM 0.760 mg/l) (Table 1). The geometric mean for total THMs ination.19 A baseline blood sample was collected shortly before shower was 96.2 mg/l (Table 1). We observed a decrease in brominated activity began. Blood samples were collected at 10 and 30 min after the THM levels over the study period (July to September 2004) end of the 10-min shower. Participants stayed in the shower room for (Supplementary Figure S7). Relative percent differences between 5 min after the shower while they were getting dressed. Therefore, study duplicate water samples for a subset (30 out of 100 participants) participants had additional THM exposure by inhalation during the 5 min of showers were o0.65% for all THMs (Supplementary Table S1), post shower in the bathroom, and thus the total exposure period was indicating uniform sampling and precise THM measurement actually 15 min. One can note that the blood samples were collected 5 and (Supplementary Figure S1).

Journal of Exposure Science and Environmental Epidemiology (2013), 39--45 & 2013 Nature America, Inc. Blood and air trihalomethane levels after showering Silva et al 41 Table 1. Geometric mean (±95% conﬁdence interval (CI)) of THMs for Table 2. Geometric mean (±95% conﬁdence interval) of THMs for tap water samples collected 5 min after shower began and air samples blood samples collected before shower, 10 min after shower, and collected 10-min time-integrated during shower and 5-min time- 30 min after shower for study participants. integrated post-shower. Analyte Time of sample GM (±95% CI), Analyte Water (5 min after the shower began) GM collection ng/l (±95% CI), mg/l Before shower 11.0 (9.87--12.2) Chloroform 64.1 (63.1--65.2) Chloroform 10 min after 197 (196--198) Bromodichloromethane 19.9 (18.9--21.0) 30 min after 99.3 (98.2--100) Dibromochloromethane 8.30 (7.18--9.42) Bromoform 0.760 (0.500--1.98) Before shower 2.25 (1.11--3.40) Total THMs 96.2 (95.2--97.2) Bromodichloromethane 10 min after 64.8 (63.7--65.8) 30 min after 32.6 (31.6--33.7) Analyte Air (10 min time-integrated sample during shower) GM (±95% CI), mg/m3 Before shower 1.30 (0.150--2.45) Dibromochloromethane 10 min after 27.9 (26.7--29.0) Chloroform 243 (242--244) 30 min after 15.0 (13.8--16.1) Bromodichloromethane 70.9 (69.8--72.0) Dibromochloromethane 25.0 (23.9--26.1) Before shower 0.960 (0.140--2.06) Bromoform 2.67 (1.56--3.81) Bromoform 10 min after 3.42 (2.26--4.58) Total THMs 351 (350--352) 30 min after 2.36 (1.22--3.50)

Analyte Air (5 min time-integrated post-shower Before shower 16.5 (15.3--17.6) sample) GM (±95% CI), mg/m3 Total THMs 10 min after 299 (298--300) 30 min after 152 (151--154) Chloroform 255 (254--256) Bromodichloromethane 74.3 (72.3--75.3) Dibromochloromethane 25.8 (24.8--27.0) Bromoform 2.69 (1.53--3.83) and during-shower water samples (r40.583, Table 3). Scatter plots Total THMs 367 (367--369) for DBCM in blood, shower air, and water are presented in Figure 1 and other THMs in Supplementary Figures S4--S6 for blood samples collected 10 min after shower, and air and water samples collected during shower. Regression parameters (slope, SE for Results from the analysis of the during-shower and post-shower slope, and Pearson’s correlation coefficient) for the simple linear samples for THMs in air are presented in Table 1. The geometric 3 regression analysis of all THMs in water vs shower air, blood vs air, mean of pre-shower air levels of total THMs was 5.8 mg/m . and blood vs water are presented in Table 3. Similar to water, chloroform was the most abundant THM The partitioning of each THM between water and shower air (GM ¼ 243 mg/m3), followed by BDCM (GM ¼ 70.9 mg/m3), DBCM 3 3 (regression slope of water vs air concentrations) remained nearly (GM ¼ 25.0 mg/m ), and bromoform (GM ¼ 2.67 mg/m ) for the constant between samples collected during and after the shower samples collected during the shower. Post-shower air samples had (Table 3). Figure 2 demonstrates the relationship between Henry’s higher total-THM levels (GM ¼ 367 mg/m3) than the during-shower 3 law constants and the regression slopes for the water/air partition air samples (GM ¼ 351 mg/m ). for each THM during the shower. Regression slopes decreased For some participants, duplicate air samples were collected. linearly with increasing Henry’s law constants. Relative percent differences for post-shower duplicate air samples Regression slopes of THMs in water/air and blood/air during the (8 out of 100) were o0.91%, for all THMs (Supplementary Table shower correlated well with the partition coefficient (K) of THMs S1), indicating uniform sampling and precise THM measurement (Figure 3). Regression slopes increased linearly with increasing (Supplementary Figure S2). THM levels in air samples collected partition coefficients in water/air (K ) and blood/air (K ). during and post showering are well correlated (Supplementary water/air blood/air Figure S3). Relative percent differences for the duplicate integrated air samples (8 out of 100) during and post (7 out of 100) showering were o1.31% for all THMs (Supplementary Table S1). DISCUSSION Table 2 presents the distributions of blood THM levels for We observed significant exposure to THMs resulting from samples collected before shower, 10 min after shower, and 30 min showering in chlorinated tap water. Volatility and partition after shower. The relative quantities of THMs in baseline blood coefficients of THMs influence the amount of THM transferred followed the same pattern found in water and air: chloroform into the air during showering, and thus the amount of THM that (GM ¼ 11.0 ng/l); BDCM (GM ¼ 2.25 ng/l); DBCM (GM ¼ 1.30 ng/l); can be inhaled by the person showering. THM levels in shower air and bromoform (GM ¼ 0.960 ng/l, Table 3). In the samples collected decreased with decreasing volatility of THM and increasing water/ 10 min after shower, GM of total THM concentrations in blood air partition coefficient (Kwater/air) (Tables 1 and 4, and Figure 3). Of was increased 18-fold (GM ¼ 299 ng/l) from GM of baseline levels. the four THMs we measured, chloroform has the lowest boiling Blood total-THM levels subsequently decreased in samples collected point. The boiling point and Kwater/air of the THMs increases with 30 min after shower (GM ¼ 152 ng/l, Table 2). Similarly, compared increasing substitution of bromines in the THM molecule (Table 4), with baseline, the GM of blood chloroform collected 10 min after leading to decreased partitioning from shower water into shower the shower were increased 18-fold from 11.0 to 197 ng/l, GM of air. Therefore, the ratio of geometric mean THM levels in shower BDCM increased 29-fold from 2.25 to 64.8 ng/l, GM of DBCM water (mg/l) to shower air (mg/m3) increases from chloroform (0.26) increased 21-fold from 1.3 to 27.9 ng/l, and GM of bromoform to BDCM (0.28) to DBCM (0.33). Furthermore, the slope of the best- increased 3.5-fold from 0.96 to 3.42 ng/l. The GM of blood levels of fit regression line for the paired water vs post-shower air THM level all THMs 30 min after shower were lower than in blood samples increases with increasing bromine content (chloroform, slope ¼ collected 10 min after the shower (Table 2). 0.174; BDCM, slope ¼ 0.219; DBCM, slope ¼ 0.321; bromoform, Post-shower blood levels of each of four THMs correlated well slope ¼ 0.376 (Table 3)). Thus, as THM volatility increases and with THM levels in during-shower air samples (r40.574, Table 3) Kwater/air decreases, partitioning into air during showering increases

& 2013 Nature America, Inc. Journal of Exposure Science and Environmental Epidemiology (2013), 39--45 Blood and air trihalomethane levels after showering Silva et al 42 Table 3. Simple linear regression parameters (slope and Pearson’s correlation coefﬁcient) for THM distribution between water/air, blood/air, and blood/water.

Analyte Water/air Blood/air Blood/water

During shower Post shower During shower Post shower During shower Post shower

Chloroform Slope 0.159 0.174 0.616 0.320 3.42 1.52 (SE) (0.010) (0.009) (0.063) (0.042) (0.437) (0.220) Pearson’s correlation coefﬁcient 0.846 0.879 0.574 0.597 0.605 0.556

Bromodichloromethane Slope 0.188 0.219 0.873 0.526 3.45 1.75 (SE) (0.016) (0.016) (0.116) (0.062) (0.477) (0.243) Pearson’s correlation coefﬁcient 0.761 0.808 0.593 0.636 0.583 0.574

Dibromochloromethane Slope 0.315 0.321 1.34 0.640 3.92 1.92 (SE) (0.011) (0.009) (0.085) (0.042) (0.239) (0.125) Pearson’s correlation coefﬁcient 0.941 0.963 0.838 0.831 0.847 0.831

Bromoform Slope 0.378 0.376 1.71 0.829 3.90 2.17 (SE) (0.030) (0.008) (0.224) (0.165) (0.361) (0.273) Pearson correlation coefﬁcient 0.968 0.980 0.908 0.868 0.913 0.909 Blood samples were collected 10 min after the shower ended (post) and water and air samples were collected during the shower (10-min time-integrated).

3 Y = 0.315 X + 0.437 2

g/m R = 0.886

0 120 Y = 3.92 X – 3.18 Y = 1.32 X – 3.45 R2 = 0.719 R2 = 0.691

30 DBCM levels in blood after shower, ng/L DBCM levels in air during shower, 0 0 5 10 15 20 25 0 20 40 60 DBCM levels in water, g/L DBCM levels in air during shower, g/m3 Figure 1. Scatter plot for dibromochloromethane (DBCM) in whole blood, shower water, and shower air samples. Blood samples were collected 10 min after the shower, water samples were collected 5 min after shower began, and air samples were collected during the shower (10-min time-integrated).

and likely leads to larger relative contributions of inhalation We observed a substantial increase in all THM blood levels at compared with dermal exposure. Dermal exposure likely remains 10 min post shower compared with pre-shower levels (Table 2). an important exposure route for all four THMs,25 with skin- This signiﬁcant increase in blood THMs after showering is con- permeability coefﬁcient increasing26 with increasing bromine sistent with the published literature; showering in THM-containing content (Table 3). water leads to a substantial increase in THM levels in blood.8--10,27

Journal of Exposure Science and Environmental Epidemiology (2013), 39--45 & 2013 Nature America, Inc. Blood and air trihalomethane levels after showering Silva et al 43 0.5 and empirically measured THM levels in shower air and water. This Y = -0.0889 X+ 0.414 ﬁnding indicates that our measurements were well controlled. As R2 = 0.991 an additional test of this relationship, predicted Henry’s law 1 0.4 CHBr3 constants at 40 C were also linearly related with the regression slopes for THM water/air partitioning (Supplementary Figure S9). DBCM For all THMs studied, the levels correlated well among water, air, 0.3 and blood. Scatter plots of THMs (Figure 1 and Supplementary Figures S4--S6) graphically display correlations of each THM in BDCM water, shower air, and post-shower blood. For all four THMs, the 0.2 post-shower air levels increased linearly with the water levels (all CHCl3 Pearson’s correlation coefﬁcients 40.808, Table 3). The tight correlations observed likely result from tightly controlling variables that 0.1 impact transfer of THMs from water to air (e.g. water temperature,

Regression slope of water/air shower spray parameters, showering time, and shower stall ventilation), leading to reproducible partitioning of THMs from 0.0 water to air. The consistency of the air THM data also indicates 0123complete mixing of air between shower stall and bathroom Henry's Law Constant, atm-m3/mol x 103 over the course of exposure period. The minimal variability in the experimental system is also supported by the high degree of Figure 2. Relationship between partition of trihalomethanes (THMs) correlation between during- and post-shower air samples in shower water/shower air (regression slope of scatter plot) and (Supplementary Figure S3). In Table 3, THM slopes increased for Henry’s law constant (at 251C) in water for chloroform (CHCl3), bromodichloromethane (BDCM), dibromochloromethane (DBCM), water/air, blood/air, and blood/water plots with increasing bromine content (i.e., chloroform BDCM DBCM bromoform); bromoform (CHBr3). o o o the same trend was observed for the Pearson’s correlation, with the exception of the BDCM water--air correlation. These trends are likely inﬂuenced by the physical and chemical properties of the 2.0 THMs: increasing the bromine content leads to decreased volatility

CHBr3 and increased lipophilicity, which result in increased skin permeability and greater dermal absorption, and thus increased water/ air, blood/air, and water/blood slopes and Pearson’s correlations. 1.5 DBCM Although blood THM concentrations increased linearly with THM levels in water and shower air, correlations between THM levels in water and air were stronger than correlations of THM 1.0 levels between blood and either shower air or water. The stronger BDCM correlation between THM levels in water and air matrices likely CHCl 3 results from a variety of factors. Shower water and air in this study were tightly controlled to minimize variability in temperature and 0.5 CHBr3 DBCM Water/Air ventilation. Conversely, THM levels in blood were affected by BDCM Blood/Air Y = 0.0124 X +0.513, R2 = 0.967 numerous factors that vary between study participants, including CHCl 3 Y = 0.0105 X + 0.135, R2 = 0.930 genetic and physiological variables (e.g. THM exhalation rate, lung

Regression slope of water/air and blood/air 0.0 capacity, metabolism, blood lipids, BMI, and cardiac parameters). 0 20406080100120 Correlations were weaker for chloroform and BDCM compared Partition Coefficient of trihalomethanes with DBCM and bromoform, likely due to differences in volatility, Figure 3. Relationship between distribution of trihalomethanes partitioning characteristics, and exposure pathways. Specifically, (THMs) in water/air and blood/air (regression slope of scatter plot) dermal absorption likely becomes increasingly more important and partition coefficient in water/air and blood/air for chloroform with increasing bromine substitution, and may be a less variable (CHCl3), bromodichloromethane (BDCM), dibromochloromethane exposure route than inhalation. In summary, this study shows that (DBCM), and bromoform (CHBr3) during showering. exposure to chloroform, BDCM, DBCM, and bromoform does occur during showering, but the extent of exposure is influenced by physicochemical and physiological parameters. Although each participant was exposed to chlorinated tap THM uptake from tap water into blood is affected by shower water water under controlled conditions, the reported individual level temperature.28 For this reason shower water temperature was of water, air, and blood THM levels varied widely. Total water THM held at 40.7 1C±0.9 1C, thus allowing us to focus on the levels during the study period ranged from 75.0 to 130 mg/l most contribution of physicochemical properties in modulation of likely owing to variation of THM concentration in the water THM exposure. Study participants showered in 40 1C water; distribution system. As a result, THM levels in shower air ranged however, we found no published work on Henry’s law constants from 184 to 542 mg/m3 (10-min time-integrated sample during measured at that temperature. Therefore, we related THM parti- shower), and 256--542 mg/m3 (5-min time-integrated sample post tioning in our study to Henry’s law constants that were previously shower). Total blood THMs (10 min post shower) ranged from 47.5 measured at 25 1C (Figure 2). These empirically determined to 615 ng/l in the study participants, due at least in part to Henry’s law constants were used rather than constants predicted differences in water and shower air THM levels during and after from lower temperature measurements,29 because of uncertain- showering. An additional source of variability of blood THM levels ties introduced by extrapolating from data collected at much is most likely due to interindividual differences in metabolism and lower temperatures. As shown in Figure 2, Henry’s law constants physiology.10 were linearly related with the regression slopes for the water/air The physicochemical properties of the THMs impacted the rate partition for each THM; regression slopes decreased linearly with at which THMs were eliminated from the blood. From 10 min post increasing Henry’s law constants. Thus, we found the expected shower to 30 min post shower, blood THM levels decreased. The relationship between published work on Henry’s law constants percent decrease of THMs in blood from 10 min post shower

& 2013 Nature America, Inc. Journal of Exposure Science and Environmental Epidemiology (2013), 39--45 Blood and air trihalomethane levels after showering Silva et al 44 Table 4. Relevant physicochemical properties (boiling point, Henry’s law constant, and partition coefﬁcient) of trihalomethanes in water, air, and human blood.

3 Analyte Kwater/air Kblood/air Kblood/water Henry’s law constant atm-m /mol Boiling point (measured)a (measured)a (calculated)a (103) (1C)

Meauredb at Predictedc at 25 1C 40 1C

Chloroform 3.66±0.31 10.7±0.70 2.98 3.00 8.97 62 Bromodichloromethane 7.43±0.35 26.6±1.4 3.58 2.41 4.77 90 Dibromochloromethane 11.8±0.71 49.2±2.3 4.16 0.990 2.62 119 Bromoform 24.7±4.0 103±8.8 4.14 0.560 1.48 149

aBatterman et al.,32 partition coefﬁcients measured at 37 1C. bHandbook of Physical Properties of Organic Chemicals, 1997;33 Henry’s law constants reported at 25 1C. cNicholson et al.;29 Henry’s law constants predicted at 40 1C.

Table 5. Comparison of ratios of THMs in 10 min post-shower blood to shower water for participants in four controlled showering studies.

Trihalomethane Ratio of median THMs in post-shower blood (ng/l) to shower water (mg/l)a

Backer et al.9b Lynberg et al.27b Nuckols et al.12b Present studyb

High site Low site High site Low site

Chloroform 4.29 3.29 7.13 1.89 2.25 3.03 Bromodichloromethane 3.50 2.71 3.58 2.82 2.55 3.14 Dibromochloromethane 5.00 2.93 3.00 2.17 3.00 3.19 Bromoformc NA NA NA NA NA 3.76

aBlood samples collected B10 min after shower ended. bN ¼ 11 in the Backer et al. study; N ¼ 25 at each site of the Lynberg et al. study; N ¼ 4 and 3 at the Nuckols et al. high and low sites, N ¼ 100 in our study, respectively. cBromoform was below or near detection limit in water source for above studies except our study.

to 30 min post shower was 49.6%, 49.7%, 46.2%, and 31.0% 30 min after shower than for baseline samples (r ¼ 0.733), similar (calculated from GMs listed in Table 2) for chloroform, BDCM, to the findings of Miles et al.31 DBCM, and bromoform, respectively. Compared with the other We compared the ratio of median THM levels in blood to water THMs, bromoform levels in blood decreased at a slower rate. THM in this study with results reported in other studies (Table 5), elimination is predicted by physiological half-life (t1/2). The t1/2 because THMs in water drive post-showering blood levels. Table 5 of THMs is dependent on physical, chemical, and metabolic presents the ratio of 10-min post-shower blood levels to water characteristics. Because of their high volatility, THMs may be levels for our study and three other studies. Previous studies eliminated via exhalation from the lungs as a function of Kblood/air. evaluated showering exposures in a smaller number of subjects The greater lipid solubility of bromoform may lead to retention in (Nr25) compared with the current study (N ¼ 100). Study param- poorly perfused adipose tissue and thus slower elimination than eters such as water temperature, duration of shower, and other the less lipophilic THMs.30 activities related to exposure were controlled in all of these We chose the study site (Pittsburgh, PA) because bromide studies. The ratio of the chloroform levels in blood to water in the present in the raw source water would lead to formation of current study was 3.03, which agreed well with Backer et al.9 brominated THMs during chlorination.31 Therefore, we expected (4.29),27 high THM site (3.29), and Nuckols et al.12 (1.89--2.25). The to find measurable levels of brominated THMs in the shower ratio of blood to water levels of less volatile THMs (e.g. BDCM and water. The shower water of the first eight participants contained DBCM) are in good agreement with Backer et al.9,27 and Nuckols relatively high bromoform levels (46 mg/l) and had a relatively et al.12 (Table 5). In the present study, higher blood to water ratios high ratio of the sum of the molar concentrations of bromide were observed for the brominated THMs compared with chloroform. in each individual THM species to the total concentration of four In conclusion, exposure to THMs occurs during showering with THMs (bromine incorporation factor (BIF)).31 Median BIF for chlorinated tap water. As expected, THM levels in post-shower shower water was 0.649 for these eight study participants. blood are highly correlated with THM levels in water and air However, the BIF dropped substantially (Supplementary Figure during showering. Under these controlled exposure conditions S7) for the subsequent 92 study participants (median BIF ¼ 0.295). (temperature, duration, air circulation) THMs transferred from This reduction in BIF may have been caused by heavy rain in the shower water into showering individuals are consistent with their area that diluted bromide levels in the source water. Blood BIFs in physicochemical properties, most importantly, Henry’s law samples collected 10 and 30 min after the shower are presented constant, volatility, and partition coefficients. Therefore, physi- in Supplementary Figure S8. BIFs of THMs in blood samples cochemical properties of THMs must be considered when are highly correlated with BIF in water samples with a higher predicting THM exposures in epidemiological studies where correlation coefficient (r ¼ 0.978) for samples collected 10 and conditions cannot be controlled.

Journal of Exposure Science and Environmental Epidemiology (2013), 39--45 & 2013 Nature America, Inc. Blood and air trihalomethane levels after showering Silva et al 45 CONFLICT OF INTEREST 16 Lakind J.S., Naiman D.Q., Hays S.M., Aylward L.L., and Blount B.C. Public health The authors declare no conflict of interest. interpretation of trihalomethane blood levels in the United States: NHANES 1999--2004. J Expo Sci Environ Epidemiol 2010: 20(3): 255--262. 17 Pirkle J.L., Needham L.L., and Sexton K. Improving exposure assessment by monitoring human tissues for toxic-chemicals. J Expo Sci Environ Epidemiol 1995: ACKNOWLEDGEMENTS 5(3): 405--424. We acknowledge Mitchell Smith for unconditional laboratory assistance, Stephen 18 Handke P. Trihalomethane Speciation and the Relationship to Elevated Total Stanfill for technical editing, and John Morrow for data analysis. The use of trade Dissolved Solid Concentrations Affecting Drinking Water Quality at Systems names and commercial sources is for identification purposes only and does not imply Utilizing the Monogahela River as a Primary Source During the 3rd and 4th endorsement by the U.S. Department of Health and Human Services or the Centers Quarters of 2008, 2009. Available at: http://files.dep.state.pa.us/Water/Wastewater% for Disease Control and Prevention. 20Management/WastewaterPortalFiles/MarcellusShaleWastewaterPartnership/dbp_ mon_report__dbp_correlation.pdf. 19 Cardinali F.L., Mccraw J.M., Ashley D.L., Bonin M., and Wooten J. Treatment of vacutainers for use in the analysis of volatile organic-compounds in human blood REFERENCES at the low parts-per-trillion level. J Chromatogr Sci 1995: 33(10): 557--560. 1 Minear R.A., and Amy G.I. Disinfection by Products in Water Treatment: The Chemistry 20 Leavens T.L., Blount B.C., DeMarini D.M., Madden M.C., Valentine J.L., Case M.W. of the Formation and Control. Lewis Publishers, Boca Raton, FL, 1996, 339--371. et al. Disposition of bromodichloromethane in humans following oral and dermal 2 Bull R.J., Birnbaum L.S., Cantor K.P., Rose J.B., Butterworth B.E., Pegram R., and exposure. Toxicological Sci 2007: 99(2): 432--445. Tuomisto J. Water chlorination: essential process or cancer hazard? Fundam Appl 21 Bonin M.A., Silva L.K., Smith M.M., Ashley D.L., and Blount B.C. Measurement Toxicol 1995: 28(2): 155--166. of trihalomethanes and methyl tert-butyl ether in whole blood using gas 3 Koivusalo M., Jaakkola J.J.K., Vartiaunen T., Hakulinen T., Karjalainen S., and chromatography with high-resolution mass spectrometry. J Anal Toxicol 2005: Pukkala E. Drinking water mutagenicity and gastrointestinal and urinary tract 29(2): 81--89. cancers: an ecological study in Finland. Am J Public Health 1994: 84(8): 1223--1228. 22 Mcclenny W.A., Pleil J.D., Evans G.F., Oliver K.D., Holdren M.W., and Winberry W.T. 4 Mcgeehin M.A., Reif J.S., Becher J.C., and Mangione E.J. Case-control study of Canister-based method for monitoring toxic VOCs in ambient air. J Air Waste bladder-cancer and water disinfection methods in Colorado. Am J Epidemiol 1993: Manage Assoc 1991: 41(10): 1308--1318. 138(7): 492--501. 23 Gordon S.M., Brinkman M.C., Ashley D.L., Blount B.C., Lyu C., Masters J., and Singer 5 Morris R.D., Audet A.M., Angelillo I.F., Chalmers T.C., and Mosteller F. Chlorination, P.C. Changes in breath trihalomethane levels resulting from household water-use chlorination by-products, and cancer---a metaanalysis. Am J Public Health 1992: activities. Environ Health Perspect 2006: 114(4): 514--521. 82(7): 955--963. 24 Cardinali F.L., Ashley D.L., Morrow J.C., Moll D.M., and Blount B.C. Measurement of 6 Bove F., Shim Y., and Zeitz P. Drinking water contaminants and adverse trihalomethanes and methyl tertiary-butyl ether in tap water using solid-phase pregnancy outcomes: a review. Environ Health Perspect 2002: 110: 61--74. microextraction GC-MS. J Chromatogr Sci 2004: 42(4): 200--206. 7 Nieuwenhuijsen M.J., Smith R., Golfinopoulos S., Best N., Bennett J., and 25 Jo W.K., Weisel C.P., and Lioy P.J. Routes of chloroform exposure and body Aggazzotti G. Health impacts of long-term exposure to disinfection by-products burden from showering with chlorinated tap water. Risk Anal 1990: 10(4): in drinking water in Europe: HIWATE. J Water Health 2009: 7(2): 185--207. 575--580. 8 Ashley D.L., Blount B.C., Singer P.C., Depaz E., Wilkes C., and Gordon S. Changes in 26 Xu X., Mariano T.M., Laskin J.D., and Weisel C.P. Percutaneous absorption of blood trihalomethane concentrations resulting from differences in water quality trihalomethanes, haloacetic acids, and haloketones. Toxicol Appl Pharmacol 2002: and water use activities. Arch Environ Health 2005: 60(1): 7--15. 184(1): 19--26. 9 Backer L.C., Ashley D.L., Bonin M.A., Cardinali F.L., Kieszak S.M., and Wooten J.V. 27 Lynberg M., Nuckols J.R., Langlois P., Ashley D., Singer P., and Mendola P. Household exposures to drinking water disinfection by-products: whole blood Assessing exposure to disinfection by-products in women of reproductive age trihalomethane levels. J Expo Anal Environ Epidemiol 2000: 10(4): 321--326. living in Corpus Christi, Texas, and Cobb County, Georgia: descriptive results and 10 Backer L.C., Lan Q., Blount B.C., Nuckols J.R., Branch R., and Lyu C.W. Exogenous methods. Environ Health Perspect 2001: 109(6): 597--604. and endogenous determinants of blood trihalomethane levels after showering. 28 Wilkes C.R., Nuckols J.R., and Koontz M.D. Evaluating Alternative Data Gathering Environ Health Perspect 2008: 116(1): 57--63. Methods for 1999 Disinfection By-Product Field Study. American Water Works 11 Cammann K., and Hubner K. Trihalomethane concentrations in swimmers and Association, Denver, CO, 2004. Report No.: Project # 2831. bath attendants blood and urine after swimming or working in indoor swimming 29 Nicholson B.C., Maguire B.P., and Bursill D.B. Henry law constants for the pools. Arch Environ Health 1995: 50(1): 61--65. trihalomethanes---effects of water composition and temperature. Environ Sci 12 Nuckols J.R., Ashley D.L., Lyu C., Gordon S.M., Hinckley A.F., and Singer P. Influence Technol 1984: 18(7): 518--521. of tap water quality and household water use activities on indoor air and internal 30 Corley R.A., Gordon S.M., and Wallace L.A. Physiologically based pharmacokinetic dose levels of trihalomethanes. Environ Health Perspect 2005: 113(7): 863--870. modeling of the temperature-dependent dermal absorption of chloroform by 13 Weisel C.P., Kim H., Haltmeier P., and Klotz J.B. Exposure estimates to disinfection humans following bath water exposures. Toxicol Sci 2000: 53(1): 13--23. by-products of chlorinated drinking water. Environ Health Perspect 1999: 107(2): 31 Miles A.M., Singer P.C., Ashley D.L., Lynberg M.C., Mendola P., Langlois P.H., and 103--110. Nuckols J.R. Comparison of trihalomethanes in tap water and blood. Environ Sci 14 Blount B.C., Backer L.C., Aylward L.L., Hays S.M., and Lakind J.S. Human exposure Technol 2002: 36(8): 1692--1698. assessment for DBPs: factors influencing blood trihalomethane levels. Encyclope- 32 Batterman S., Zhang L., Wang S.G., and Franzblau A. Partition coefficients for the dia of Environmental Health, Nriagu JO (ed) 2011: 3: 100--107. trihalomethanes among blood, urine, water, milk and air. Science of the Total 15 USEPA. National Primary Drinking Water Regulations: Stage 2 Disinfectants and Environ 2002: 284(1--3): 237--247. Disinfection Byproducts Rule; Final Rule, 2006. Available at: http://www.epa.gov/ 33 Howard P.H., Meylan W.M. et al. Handbook of Physical Properties of Organic fedrgstr/EPA-WATER/2006/January/Day-04/w03.pdf. Chemicals. Lewis Publishers, Boca Raton, Florida. 1997.

Supplementary Information accompanies the paper on the Journal of Exposure Science and Environmental Epidemiology website (http:// www.nature.com/jes)

& 2013 Nature America, Inc. Journal of Exposure Science and Environmental Epidemiology (2013), 39--45