ETS Exposure and PAH Body Burden in Nonsmoking Italian Adults

Active smoking is associated with increased body burden of polycyclic aromatic hydrocarbons (PAHs); the aim of this study was to assess whether environmental tobacco smoking (ETS) increases the internal dose of PAHs. In 344 nonsmoking Italian adults, out of 497 individuals selected as representative of the population of the town of Modena, ETS exposure was evaluated by a self-administered questionnaire and by the measurement of urinary cotinine (COT-U). PAH exposure was assessed by the measurement of urinary 1-hydroxypyrene (1-OHPYR) and of ten urinary PAHs. In all subjects, median (5th–95th percentile) COT-U was 0.47 (<0.1–3.91) µg/L. While 58 subjects reported to be ETS exposed (ETSQUEST), 38 individuals were identified as ETS exposed on the basis of a COT-U value of 1.78 (90% confidence interval 1.75–1.80) µg/L, previously derived as an upper reference value in not ETS exposed Italian adults (ETSCOT). Median COT-U levels were 1.38 (<0.1–9.06) and 3.63 (1.80–17.39) µg/L in ETSQUEST and in ETSCOT subjects, respectively. Significant correlations between COT-U and 1-OHPYR, and urinary anthracene, fluoranthene, pyrene, and chrysene were found among all subjects. Significantly higher levels of 1-OHPYR, and urinary fluorene, anthracene, and pyrene were found in ETSCOT individuals. The results of multiple linear regression analyses, taking into consideration diet and other sources of PAHs exposures such as the residence area/characteristics and traffic, confirmed that 1-OHPYR and urinary fluorene were affected by ETS exposure, even if ETS played a minor role.


Introduction
Environmental tobacco smoke (ETS), also called passive smoke or involuntary smoking, is made of several components: secondhand smoke, that is the combination of the mainstream smoke exhaled by the smoker and of side-stream smoke released by the burning of the tobacco product, and thirdhand smoke, that is residual tobacco smoke pollutants remaining on surfaces and dust after tobacco has

Study Population and Sample Collection
The data presented in this paper have been collected through a cross-sectional biomonitoring study on exposure to emissions from a local urban-waste incineration plant [19]. Recruitment, interviewing, and sampling took place between November 2012 and April 2013. The study population consisted of volunteer adult participants (18-70 years) from the general population of Modena, a medium-sized town in Northern Italy (Emilia-Romagna region, Italy). Eligible subjects were randomly selected from the population base, which comprises approximately 40% of the town population. Sampling method implied stratification by gender, age group (18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46)(47)(48)(49) and 50-69 years), and exposure. The study sample was similar to the source population (the town population) in terms of sex, age, and citizenship based on a comparison with data of the population register and of the health surveillance system. The number of subjects to be included was calculated to detect a significant difference of at least 20% in biomarker levels comparing the most exposed to the least exposed subject; this results in about 500 individuals. Sampling comprised the selection of three replacements for each subject, belonging to the same sampling stratum. Detailed information on the recruitment criteria and procedure was described previously [17,18].
Invitations to participate in the study were sent out by post. Individuals were supplied with a study pack containing the invitation letter, the questionnaire, a disposable polyethylene bottle, and the instruction to collect a spot urine sample from the first void of the day. Subjects were telephonically contacted about one week after the dispatch of the invitation letter and those answering positively were invited to the Local Health Unit to provide the biological sample and to complete the questionnaire on personal and lifestyle characteristics. Urine samples were immediately refrigerated at 4 • C and delivered to the laboratory, where they were kept at −20 • C in the dark. Not-respondents and refusals were substituted in an appropriate way to maintain the stratification homogeneity.
All participants were informed about the aims of the research and signed an informed consent form. The study was approved by the ethics committee of Modena. The study has been carried out in accordance with the Declaration of the World Medical Association (Declaration of Helsinki) for experiments involving humans (http://www.wma.net/e/policy/b3.htm).

Questionnaire for Assessment of Active or Environmental Exposure to Tobacco Smoke
Participants completed a detailed questionnaire about current active and passive smoke exposure [18]. The used items were adapted from available questionnaires used in large population surveys and had been verified in our previous studies [18,20]. The questionnaire was reviewed by a trained interviewer at the moment of urine sample collection. In particular, to classify current active exposure to tobacco smoke, the following questions were asked: current active tobacco smoking (yes/no), smoking product (cigarette/cigar/pipe/e-cigarette/other), product commercial name, weekly and daily smoking intensities, and smoking environment (only open places/only closed places/both open and closed places). To classify current ETS exposure, the following questions were asked: living with smokers (yes/no), cohabitants smoked in the house (yes/no), working with smokers (yes/no), coworkers smoked in the same room (yes/no), and daily ETS exposure within the last week (yes/no). If a participant answered "yes" to the last question, then information on ETS duration (how many days/week; how many hours/week; how many hours/day), smoking type (cigarette/cigar/pipe smoke), and environment (home/work/leisure time/car; open/closed places) was collected. Moreover, the time elapsed from the last smoke exposure to urine collection was ascertained at the moment of urine collection.

Urinary PAHs and 1-OHPYR Analysis
Prior to analysis, urinary samples were left at room temperature until completely thawed. After shaking, two aliquots for the determination of urinary PAHs and 1-OHPYR were transferred to the analysis vials and underwent the respective analytical procedures, as described below.

Urinary Creatinine
Urinary creatinine was determined using Jaffe's colorimetric method [22]. The creatinine value was used to assure sample validity, excluding samples with excessive physiologic dilution or concentration according to the 0.3 g/L ≤ creatinine ≤3.0 g/L range [23]. Moreover, creatinine was introduced in the regression models as an independent variable [24].

Statistical Analysis
Statistical analysis was performed by using the IBM SPSS (ver. 24.0 for Windows; SPSS Statistics, IBM Italia, Segrate, Italy) and by the STATA 11 software packages (Stata Corp LP, College Station, TX, USA). The median and percentiles of COT-U and PAH biomarkers were used to describe the non-parametric distributions in all subjects and in subgroups stratified by ETS exposure. Active smoking was defined as self-reported current smoking or COT-U levels ≥ 30 µg/L. Among correctly identified nonsmokers (individuals self-classified as nonsmokers and with COT-U < 30 µg/L), subjects were classified as ETS exposed alternatively based on the questionnaire, when daily ETS exposure within the last week was reported in the questionnaire (ETS QUEST ), or based on a COT-U value of 1.78 (90% confidence interval 1.75-1.80) µg/L, previously derived as upper reference value in not ETS exposed Italian adults (ETS COT ) [18].
Pearson's correlations were used to test the associations between variables. The Student's t test for independent samples was used on decimal log-transformed data for the univariate evaluation of differences between groups. The chi-square test was used to compare the percentage distributions among groups.
For each biomarker, three multivariate linear regression models were estimated to study the influence of ETS exposure on PAH body burden. ETS exposure was introduced in the regression models alternatively as a continuous variable (not transformed COT-U, µg/L, Model 1) or as a dichotomous variable (ETS COT in Model 2 or ETS QUEST in Model 3). Considering the large number of confounding factors that may influence PAH exposure, an extensive pool of complementary information including diet, environmental sources of exposure, occupational exposure, characteristic of the residence, atmospheric parameters, and personal habits, was collected by a detailed questionnaire, as described previously [17]. Covariates included in the finals models were those found to have a significant effect in preliminary regression analysis or known to affect biomarkers based on the pertinent literature. Some covariates were included in models as fixed variables, regardless of the statistical significance: gender, age, education level (low/high), citizenship (Italian/others), body mass index, creatinine, residence zone (rural, industrial, urban, mixed), daily temperature, precipitation, outdoor traffic exposure (low, medium, high), heating exposure (NOx), time spent at home (h/day), residential distance from major roads, and occupational exposure to PAHs (yes/no). The sampling day was introduced as a spline variable (with up to three knots) to control for exposure variability.
To account for the impact of a potential selection bias, the Inverse Probability Weighting (IPW) was calculated; the weighting allows to correct for the likelihood of adherence estimated by socio-demographic variables (gender, age, citizenship), which were the only markers available in the administrative database both for respondents and non-respondents. Unfortunately, the education level, which is known to be related to compliance, as well as the other used socio-demographic variables, was not available in the administrative database.
In the regression models, PAH biomarkers were transformed according to the distribution characteristics by Box-Cox transformation. A logarithmic transformation was applied if the lambda of the Box-Cox transformation was equal to zero. For biomarker values below the LOQ the actual values were used. Outliers, defined as samples with biomarker levels above the 99th percentile, were excluded from statistical analysis. Samples with creatinine concentrations above 3 g/L or below 0.3 g/L were excluded as well. Figure 1 shows the participant flow-chart. Among 497 individuals recruited to the biomonitoring study (response rate 53.5%, non-availability rate 19.2%, refusal rate: 27.3%), two participants were excluded from the analysis because they did not complete the questionnaire, while 151 were identified as active smokers and were excluded as well. Finally, 344 individuals were identified as nonsmokers and were included in this study. Their mean age (minimum-maximum) was 45 (18-69) years, 183 (53%) participants were female, 324 (94%) were white individuals from Italy or other European countries (Table 1). level, which is known to be related to compliance, as well as the other used socio-demographic variables, was not available in the administrative database.

Study Population
In the regression models, PAH biomarkers were transformed according to the distribution characteristics by Box-Cox transformation. A logarithmic transformation was applied if the lambda of the Box-Cox transformation was equal to zero. For biomarker values below the LOQ the actual values were used. Outliers, defined as samples with biomarker levels above the 99th percentile, were excluded from statistical analysis. Samples with creatinine concentrations above 3 g/L or below 0.3 g/L were excluded as well. Figure 1 shows the participant flow-chart. Among 497 individuals recruited to the biomonitoring study (response rate 53.5%, non-availability rate 19.2%, refusal rate: 27.3%), two participants were excluded from the analysis because they did not complete the questionnaire, while 151 were identified as active smokers and were excluded as well. Finally, 344 individuals were identified as nonsmokers and were included in this study. Their mean age (minimum-maximum) was 45 (18-69) years, 183 (53%) participants were female, 324 (94%) were white individuals from Italy or other European countries (Table 1).   Table 1. Characteristics of participants (n = 344), stratified by environmental tobacco smoke (ETS) exposure based on COT-U excretion (ETS COT : COT-U > 1.78 µg/L = ETS exposed) and on self-classification by questionnaire (ETS QUEST ).
Twenty-three subjects reported to be ETS exposed and had COT-U levels higher than 1.78 µg/L. Table 2 reports the results of urinary PAH and 1-OHPYR levels in all subjects and in subjects stratified by ETS exposure based on self-classification by questionnaire and on COT-U excretion. Considering all subjects together, urinary PAHs were detected at least in 65% of samples, beside U-BaA, that was detected only in 18% of samples. U-NAP, U-PHE, and U-ANT were detected in all samples. 1-OHPYR was quantified only in 33% of subjects.

PAH Exposure and ETS
In subjects stratified by ETS exposure based on self-reporting, the percentage of analytes above the LOQ was similar between ETS QUEST and not-exposed individuals. Urinary PAH or 1-OHPYR levels were not different comparing ETS QUEST vs. not-exposed individuals; only for U-ANT, slightly higher levels were observed in ETS QUEST than in not-exposed subjects (2.0 vs. 2.2 µg/L, p = 0.102).
In subjects stratified by ETS exposure based on COT-U excretion, the percentage of analytes above the LOQ was similar between ETS COT and not-exposed individuals; only for U-ACE and 1-OHPYR a slightly higher percentage of samples above the LOQ was found in ETS COT than in not-exposed subjects (89 vs. 78%, p = 0.102; and 42 vs. 32% p = 0.213, respectively). Significantly higher levels of 1-OHPYR (p = 0.037), U-FLU (p = 0.028), U-ANT (p = 0.028), and U-PYR (p = 0.052) were found in ETS COT than in not-exposed individuals.  s t test comparing ETS-exposed vs. non-ETS exposed subjects.

Contribution of ETS to PAH Exposure
In all subjects, COT-U levels were significantly, or marginally significantly, correlated with 1-OHPYR (Pearson's r = 0.121, p = 0.024), U-ANT (r = 0.128, p = 0.017), U-FLT (r = 0.094, p = 0.083), U-PYR (r = 0.106, p = 0.051), and U-CHR (r = 0.114, p = 0.035). Table 3 shows the results of the adjusted linear regression models in detail. Only beta-coefficients (β) of ETS exposure, together with the 95% confidence intervals, are shown regardless of the significance level. Due to the outlier exclusion, the sample size was different for each PAH biomarker and ranged from 287 (1-OHPYR) to 313 (U-ACY, U-FLU, and U-FLT). The complete regression model, showing beta-coefficients for all the independent variables, is shown in Supplementary Table S1 for Model 1. Multivariate regression models were performed for each analyte, except for U-BaA, given the high proportion of samples with levels below the LOQ for this biomarker.
When ETS exposure was introduced in the regression model as a dichotomous variable based on self-classification by questionnaire (ETS QUEST , Model 3), none of the PAH biomarker resulted associated to ETS.
For both Model 2 and Model 3, the results regarding other variables affecting PAH biomarkers were very similar to what is shown for Model 1 in Supplementary Table S1. Table 3. Results of multiple linear regression analyses for predicting the levels of urinary PAHs and 1-OHPYR in study subjects. ETS exposure was introduced in the regression models alternatively as a continuous variable (unadjusted COT-U levels, Model 1), or as a dichotomous variable (ETS COT in Model 2, and ETS QUEST in Model 3). Beta-coefficients (β) of ETS exposure, together with the 95% confidence intervals, are shown regardless of the significance level. The complete regression model is shown in Supplementary Table S1 for Model 1.

Biomarkers of ETS and PAH Exposure
In this study, urinary cotinine was measured in samples from 344 individuals from a representative group of the general population. Among the investigated subjects, 58 (17%) reported being daily ETS exposed (ETS QUEST ) ( Table 1). Median COT-U levels in these subjects were higher than those in subjects not reporting ETS exposure (1.38 vs. 0.39 µg/L, p < 0.001) and consistent with those found in studies conducted in countries where a smoking-ban in public places has been implemented [25,26]. As several variables may affect ETS exposure (i.e., source, duration, intensity) and certain individuals may be particularly prone to report it, the use of the upper reference value of CUT-U (1.78 µg/L) as a cut-off value allowed us to identify those individuals who were exposed to high ETS level (ETS COT ). Median COT-U levels (3.63 µg/L) in these subjects were similar to those previously reported in non-smoking adults before the smoking-ban was implemented [12,[27][28][29].
The exposure to PAHs was evaluated by measuring urinary 1-OHPYR and ten unmetabolized PAHs (Table 2). In all subjects, 1-OHPYR was quantifiable only in 33% of the samples, with median levels <0.05 µg/L and well below the reference value for the Italian population (<0.3 µg/L in non-smokers and <0.7 µg/L in smokers) [30]. Urinary PAHs, except U-BaA, were quantifiable in the large majority of samples, even though median levels of all analytes were very low (ng/L order of magnitude). Not so many studies reported the measurement of urinary PAHs in the general population: the levels here found were similar or lower than those reported for adults in the town of Modena [31] and in a small group of adolescents participating in the Flemish Environment and Health Study [32], and much lower than those of non-smoking adults living in an industrial polluted area in Poland [33]. Altogether, PAH biomarkers measured in this study are indicative of very low PAH exposure.

ETS and PAH Exposure
The possible role of ETS in determining PAH intake has been seldom investigated. Higher levels of PAH metabolites in ETS exposed adults than in not-exposed were reported for 1-OHPYR with a linear relationship between 1-OHPYR excretion and duration of ETS exposure [34], for hydroxylated metabolites of phenanthrene and again 1-OHPYR in highly ETS exposed subjects [12], and for hydroxylated metabolites of phenanthrene, fluorene, and pyrene in the U.S. general population participating in the 1999-2002 National Health and Nutrition Examination Survey [11]. More recently, 1-OHPYR was found higher in adults exposed to ETS during the weekend or more than 4 h/day along the week than in not-exposed adults participating in a human biomonitoring program on a national scale in Spain [35]. Among several PAH biomarkers investigated in this study, the levels of 1-OHPYR, U-FLU, U-ANT, and U-PYR were significantly higher in subjects stratified as ETS-exposed based on COT-U excretion (ETS COT subjects) (Table 2). Thus, consistently with previous reports, higher exposure to PAHs was evident only in highly ETS-exposed subjects.
As regards the exposure to specific PAHs, a very interesting result of this study is that U-FLU was higher in ETS COT subjects than in not-ETS exposed subjects (Table 2). Fluorene, which is a low-molecular weight PAH with three aromatic rings like anthracene, resulted among the compounds more contributing to mainstream smoke composition (as ng/cigarette) and among the most abundant in the gas phase of tobacco smoke [4]. Previous studies indicated fluorene hydroxylated metabolites as more specific and selective biomarkers than 1-OHPYR to discriminate PAH exposure in smokers and nonsmokers [10,36]. Our present results suggest that U-FLU may be indicative of exposure to low-molecular weight compounds in ETS-exposed subjects too.
Positive correlations were found between COT-U and some of the measured PAH biomarkers, including 1-OHPYR, U-ANT, U-FLT, U-PYR, and U-CHR. It is interesting that 4-ring PAHs, such as fluoranthene, pyrene and chrysene, are among the most important contributors to side-stream smoke [3]. This result is consistent with a previous study where a correlation between environmental chrysene exposure and urinary cotinine was shown [12] and it highlights the existence of a relation between ETS exposure and particle-bound PAHs.

Role of ETS in Determining PAH Exposure
Given that PAHs are originated from multiple sources and co-exposure is very common, especially in not-occupationally exposed individuals, the regression analysis enabled us to evaluate the role of ETS exposure in determining PAH exposure (Table 3 and Supplementary Table S1). The regression models suggested a weak contribution of ETS to PAH intake and only when ETS was defined by COT-U excretion (both as continuous variable in Model 1, and as a dichotomous variable in Model 2). This underlines that the questionnaire may be not an adequate tool to define ETS exposure for the wide variability of ETS exposure occurrence and the consequent individual's difficulty in recognizing and/or recalling it [18]. Among the measured biomarkers, significant or marginally significant associations with COT-U were found only for 1-OHPYR (with a percentage increment of 4% for each 10-fold increase in cotinine excretion) and for U-FLU, partly confirming the results of the univariate analysis.
Moreover, the regression analysis highlighted that ETS is a minor source of PAH exposure among other sources: other contributors to 1-OHPYR excretion were creatinine, the education level, and the intake of coffee, while other contributors to U-FLU excretion were the exposure to the waste incinerator emissions, the residence area, the exposure to traffic, and the presence of mold on the residence wall (Supplementary Table S1). While some associations may be expected, such as those found with diet, traffic exposure, and the residence area, others, among which the presence of mold on the residence wall, the education level, and the citizenship may be not so obvious and may be proxies of other factors (i.e., different diet, cooking habits, lifestyle, and residence characteristics).

Limitations and Strengths
Some limitations may be identified in this study, such as the lack of PAH personal air measurements that would have allowed making associations between PAH exposure from ETS and biomarkers. Moreover, even if the original sample population consisted of about 500 individuals, with a sample size adequate to observe a significant difference in the biomarkers levels, the number of ETS exposed subjects was actually small, which may have influenced the robustness of the present findings. The strength of this study is that several PAH biomarkers were investigated simultaneously on a representative group of the general population, giving the opportunity to better characterize the exposure profile. Few studies reported the association between ETS and PAH exposure, and only hydroxylated metabolites were investigated [11,12], whereas this is the first time that unmetabolized PAHs have been studied to this aim. Moreover, other potential sources of possible PAH exposure were carefully accounted for by regression models allowing us to investigate the role of ETS in determining PAH body burden.

Conclusions
In conclusion, this study highlighted a weak, but significant, contribution of ETS exposure to the PAH body burden, as shown by 1-OHPYR and fluorene increases in ETS exposed individuals. On the other hand, our results suggest that ETS exposure experienced by subjects in this study represents a minor source of exposure to PAHs.
Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/15/6/1156/ s1, Table S1: Estimates of regression models using urinary PAHs as dependent variables. Beta-coefficients (β) of COT-U are shown regardless of the significance level, whereas only β values below the 0.10 significance level are illustrated for the other covariates.