Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jul 1.
Published in final edited form as: Regul Toxicol Pharmacol. 2022 Apr 22;132:105171. doi: 10.1016/j.yrtph.2022.105171

Characterization of population variability of 1,3-butadiene derived protein adducts in humans and mice

Gunnar Boysen 1,*, Ivan Rusyn 2,*, Weihsueh A Chiu 2, Fred A Wright 3
PMCID: PMC9575152  NIHMSID: NIHMS1840429  PMID: 35469930

Abstract

1,3-butadiene is a known human carcinogen and a chemical to which humans are exposed occupationally and through environmental pollution. Inhalation risk assessment of 1,3-butadiene was completed several decades ago before data on molecular biomarkers of exposure and effect have been reported from both human studies of workers and experimental studies in mice. To improve risk assessment of 1,3-butadiene, the quantitative characterization of uncertainty in estimations of inter-individual variability in cancer-related effects is needed. For this, we ought to take advantage of the availability of the data on 1,3-butadiene hemoglobin adducts, well established biomarkers of the internal dose of the reactive epoxides, from several large-scale human studies and from a study in a Collaborative Cross mouse population. We found that in humans, toxicokinetic uncertainty factor for 99th percentile of the population ranged from 3.27 to 7.9, depending on the hemoglobin adduct. For mice, these values ranged from less than 2 to 7.51, depending on the dose and the adduct. Quantitative estimated from this study can be used to reduce uncertainties in the parameter estimates used in the models to derive the inhalation unit risk, as well as to address possible differences in variability in 1,3-butadiene metabolism that may be dose-related.

Introduction

1,3-Butadiene is a ubiquitous environmental and occupational pollutant that is typically released into the environment during manufacturing of 1,3-butadiene-based rubbers; it is also a side product of incomplete combustion and a major tobacco smoke constituent (IARC, 2008; Soeteman-Hernandez et al., 2013). This chemical has been classified as a human carcinogen by the regulatory agencies worldwide and a genotoxic mode of action has been well established as a key mechanism (IARC, 2008; US EPA, 2002). It is metabolized to three epoxides: 1,2-epoxybutene (EB), 1,2;3,4-diepoxybutane (DEB) and 3,4-epoxy-1,2-butanediol (EB-diol); these are detoxified by glutathione S-transferases and epoxide hydrolases to generate mercapturic acids that are excreted in urine and used as biomarkers of butadiene exposure (van Sittert et al., 2000). The 1,3-butadiene-derived epoxides are highly reactive and can alkylate and crosslink DNA bases to form a range of nucleotide adducts which may interfere with accurate DNA replication (Carmical et al., 2000; Kotapati et al., 2015). In addition, adducts of amino acids in proteins, such as hemoglobin, are also formed by these nucleophilic epoxides of 1,3-butadiene (Boysen et al., 2007). These adducts are considered to be even more useful biomarkers of exposure than urinary mercapturic acid metabolites (Ogawa et al., 2006). Several hemoglobin adducts of 1,3-butadiene, such as N-(2-hydroxy-3-butenyl)-valine (HB-Val) formed by EB, N,N-(2,3-dihydroxy-1,4-butadiyl)-valine (pyr-Val) formed by DEB, and N-(2,3,4-trihydroxybutyl)-valine (THB-Val) formed by EB-diol, have been studied most extensively (Boogaard, 2002; Elfarra et al., 2001; Swenberg et al., 2001).

Even though hemoglobin adducts are not causally linked to mutagenesis, they have been used extensively to characterize species differences in metabolism and toxicity of a number of genotoxic industrial chemicals, including 1,3-butadiene (Boogaard, 2002; Ogawa et al., 2006). Blood samples are easier to obtain than tissue specimens in epidemiological and animal studies. Hemoglobin adducts persist in red blood cells over weeks to months and are not removed by enzymatic repair systems like DNA adducts. Therefore, hemoglobin adducts have been evaluated as biomarkers for internal formation of the individual 1,3-butadiene metabolites in both humans and in animal studies (Boysen et al., 2012; Boysen et al., 2007; Georgieva et al., 2010).

While there is no shortage of studies quantifying 1,3-butadiene hemoglobin adducts following exposures, these data are yet to be used in quantitative cancer risk assessment of 1,3-butadiene. In fact, 1,3-butadiene cancer risk assessment was last completed by the United States Environmental Protection Agency almost 20 years ago (US EPA, 2002). At that time, classification of 1,3-butadiene as “carcinogenic to humans by inhalation” was based on the total weight-of-evidence consisting of sufficient evidence from epidemiologic studies of U.S. workers who were occupationally exposed, sufficient evidence in laboratory animal studies by inhalation, and consistent evidence demonstrating that 1,3-butadiene is metabolized into genotoxic metabolites in both experimental animals and humans. The importance of hemoglobin adducts as an effective measure of exposure to 1,3-butadiene was acknowledged in 2008 cancer hazard evaluation by the International Agency for Research on Cancer, but these data were not used explicitly in the working group’s final decision (IARC, 2008). A number of publications since 2002 have provided additional evidence on 1,3-butadiene metabolism, including data on hemoglobin adducts in human (Albertini et al., 2003; Albertini et al., 2007; Boysen et al., 2012) and mouse (Nellis et al., 2021) populations. Therefore, in this study we sought to take advantage of the availability of the data on 1,3-butadiene hemoglobin adducts as a measure of exposure to reduce uncertainties in the parameter estimates used in the models to derive the inhalation unit risk, as well as to address possible differences in variability in 1,3-butadiene metabolism that may be dose-related (US EPA, 2002).

Methods

Study cohorts

The hemoglobin adduct data were extracted from previously published reports. Blood samples had been collected from the individuals exposed to 1,3-butadiene in the workplace as described in detail by Albertini et al (Albertini et al., 2003; Albertini et al., 2007). The cohort from 1998 (Albertini et al., 2003) consisted of 83 male subjects in two production facilities of which 24 were 1,3-butadiene monomer production facility workers, 34 were 1,3-butadiene polymerization facility workers, and 25 were administrative staff who had no known direct exposures to 1,3-butadiene. For exposed individuals, personal monitors were used to evaluate individual exposures on approximately 10 separate occasions for 8-hour work shifts over a 60-day exposure assessment period. For administrative workers, a series of 28 random 1,3-butadiene measurements were taken during the exposure assessment period. Group average exposures to 1,3-butadiene, calculated as 8 hr time-weighted average), were 0.642 mg/m3 (0.290 ppm) for the monomer group, 1.794 mg/m3 (0.812 ppm) for the polymer group, and 0.023 mg/m3 (0.010 ppm) for the administrative staff group.

The cohort from 2003 (Albertini et al., 2007) included female and male workers and administrative staff from the same 1,3-butadiene polymerization facility as in the study detailed above (Albertini et al., 2003). There were 49 individuals exposed to 1,3-butadiene in the workplace with mean 8 hr time-weighted average of 1,3-butadiene exposure of 0.397 mg/m3 (0.180 ppm) for females, and 0.808 mg/m3 (0.370 ppm) for males. Exposure assessment was conducted using personal monitoring of each worker over ten full (8 hrs) shift over a 4-month period. There were 55 administrative staff with 8 hr time-weighted average of 1,3-butadiene exposure of 0.008 mg/m3 (0.0035 ppm) for females and 0.397 mg/m3 (0.180 ppm) for males.

The data from the Collaborative Cross mouse population were from a study of (Nellis et al., 2021) who conducted full body inhalation exposure to 1,3-butadiene at 2, 20 or 200 ppm for 5 days/week for 2 weeks (10-day exposure). Female mice (6–8 weeks old) were from 60 Collaborative Cross strains and one mouse per strain was used for each 1,3-butadiene exposure condition.

Hemoglobin adduct data

HB-Val, pyr-Val and THB-Val in human blood samples were previously quantitated by tandem mass spectrometry as detailed elsewhere (Boysen et al., 2012; Boysen et al., 2004; Swenberg et al., 2000). Briefly, red blood cells were washed in phosphor buffered saline prior to globin isolation according to the procedure by Mowrer et al. (Mowrer et al., 1985). Then, adducted N-terminal valine was selectively cleaved by Edman degradation using pentafluorophenyl isothiocyanate and quantitated by gas chromatography mass spectrometry (Swenberg et al., 2000). For detection of pyr-Val, globin samples were digested with a trypsin-biotin agarose suspension, concentrated using immunoaffinity, and analyzed by liquid chromatography mass spectrometry (Boysen et al., 2012; Boysen et al., 2004).

For the analysis of globin in the mouse study, all three protein adducts were quantified as N-terminal peptide adducts after tryptic digestion as detailed in (Boysen et al., 2019; Nellis et al., 2021). Briefly, blood was obtained by cardiac puncture and globin was isolated and digested with trypsin-agarose beads as described above. The trypsin digests, containing adducted and non-adducted N-terminal peptides, were analyzed by liquid chromatography mass spectrometry. For accurate quantitation of the protein adducts, all methods utilized synthetic stable isotope internal standards, added at the beginning of the analysis.

Regression analyses for exposure and covariates

Hemoglobin adduct values were highly positively skewed; therefore, the statistical analyses were conducted on the natural logarithmic scale, with the minimum positive value substituted for adduct values of zero. Regression analyses for the effect of exposure used log10(dose in ppm) as a predictor, and for the mouse data, a control dose value was chosen so that the log-scale difference between control and the minimum dose was the same as that between the minimum and next higher dose. For the human data, linear regression was performed for adduct values with dose and additional covariates as predictors. For the mouse data, mouse strain effects were naturally considered as random effects, and we fit mixed models using the lme4 package in R 3.5.3 for adduct~strain+dose+covariates, with dose and covariates fit as fixed effects. P-values for the fixed effects used t-statistics computed as coefficient/(standard error) with approximate residual degrees of freedom from the ordinary linear model.

Variability Analyses and Calculation of Uncertainty Factors

Calculation of the estimates of inter-individual or within-strain variability followed statistical analysis procedures detailed elsewhere (Erber et al., 2021; Lewis et al., 2019). For the adducts on the natural log scale, geometric means and standard deviations were deemed appropriate to summarize adducts on the original scale, with confidence intervals exponentiated after applying normality assumptions on the log scale (Sullivan, 2010). In the human data (Albertini et al., 2003; Albertini et al., 2007; Boysen et al., 2012) the covariates were (i) measured exposure (continuous variable) and (ii) smoking history (binary variable). Unobserved sources of variation or covariate measurement errors would tend to produce conservative uncertainty factors. As the human data did not contain within-individual replication, an intraclass correlation estimate from (Jokipii Krueger et al., 2020) for 1,3-butadiene EB-GII DNA adducts was substituted as detailed in (Erber et al., 2021), which when multiplied by the total variance estimate provides an estimate of within-individual variance. For the mouse data, variability estimates were computed within each dose value in order to be sensitive to potential changes with dose. Thus, we deemed regression modeling as an appropriate choice, with age and pre-exposure body weight as covariates, and total variability was used as a conservative upper bound estimate for across-strain variability.

As detailed previously in (Erber et al., 2021), we computed σ as the square root of the across-individual estimated residual variance, and uncertainty factor high (UFH) values using exp(zσ) values for 95% and 99% standard normal z-quantiles. Standard χ2 distributional assumptions on the log-scale data and degrees of freedom from the regression model provided confidence intervals for the conservative uncertainty factor.

Results

Overall, the populations used in this study comprised 177 humans with long-term 1,3-butadiene exposure ranging from 0.002 ppm (effectively below detection limit of 0.004 ppm) to 9.2 ppm, and 60 genetically diverse mouse strains where exposures were controlled at 2, 20 or 200 ppm over 10 days. Three valine adducts in hemoglobin are commonly used as effective measures of exposure to reactive intermediates of 1,3-butadiene metabolism and the data on these adducts were used in this study to quantify toxicokinetic variability in humans exposed to varying levels of 1,3-butadiene occupationally, or in mice that were exposed in a dose-response study design to levels higher of 1,3-butadiene that were comparable to those in humans or were an order of magnitude greater.

We conducted descriptive analyses of the data to determine the concordance between measured exposure dose and blood levels of valine 1,3-butadiene adducts. In data from humans (Figure 1), all three adducts to N-terminal valine in hemoglobin were significantly positively correlated with 1,3-butadiene exposure levels; the strongest correlation was observed for HB-Val adduct and the weakest for pyr-Val adduct. The correlation among adducts was also significant, but the strongest correlation was between HB-Val and THB-Val adducts, with pyr-Val measurements being only weakly, albeit significantly, correlated with THB-Val. In mice (Figure 2), concentration-dependent increases in valine adducts were observed; however, considerable variability was evident among strains, consistent with the prior reports of inter-strain variability in BD-induced DNA adducts (Chappell et al., 2014; Erber et al., 2021; Koturbash et al., 2011). Similar to the data from humans, significant correlations were observed among all three adducts in the mouse study data, but the strongest correlation was found for THB-Val and pyr-Val.

Figure 1.

Figure 1.

Open in a new tab

Correlation analyses of the protein adducts formed from 1,3-butadiene metabolites that were measured in blood of humans exposed occupationally. Pearson correlation (R2 values) and a corresponding two-tailed p-value are shown on each plot. Red dashed line is the best fit linear regression fit to each dataset.

Figure 2.

Figure 2.

Open in a new tab

Left panel, a quantitative analysis of the protein adducts formed from 1,3-butadiene metabolites that were measured in blood of Collaborative Cross mice exposed to 2–200 ppm of butadiene for 10 days. Violin plots are shown separately for each exposure group. Horizontal red line is the median value and the horizontal black lines are 25th and 75th quartiles. Right panel, correlation analysis of the protein adducts formed from 1,3-butadiene metabolites that were measured in blood of Collaborative Cross mice exposed to 2–200 ppm of butadiene for 10 days. Pearson correlation (R2 values) and a corresponding two-tailed p-value are shown on each plot.

The relationships between the exposure to 1,3-butadiene (i.e., the concentrations measured in each study) and other covariates on the level of hemoglobin adducts in human and mouse samples is shown in Table 1. For all of the adducts, exposure was significantly positively associated with measured adduct levels. For both humans and mice, HB-Val showed the largest regression coefficient, although with widely differing standard errors. Coefficient estimates were generally higher in the human data as compared to that in the mouse study, possibly due to nonlinearity of dose-response relationships over large ranges of exposure. The highest exposure in the mouse study was about 2 orders of magnitude greater than that for humans. For humans, smoking status was not significant after including exposure as a predictor. For mice, age was significantly positively associated with adduct levels for HB-Val only, and pre-exposure body weight was not a significant predictor of adduct levels.

Table 1.

Regression modeling for ln(adduct levels) following 1,3-butadiene exposure using fixed effect modeling and mixed effects modeling (mice), with fixed effect predictor factors as shown.

Species Adduct Factor Estimate Std. Error P-value
Humans HB-Val 1,3-butadiene exposure (log10 dose) 0.86455 0.11235 3.41e-11**
Smoking status −0.05311 0.21421 n.s.
THB-Val 1,3-butadiene exposure (log10 dose) 0.49332 0.04858 <2e-16**
Smoking status 0.05099 0.03827 n.s.
pyr-Val 1,3-butadiene exposure (log10 dose) 0.41643 0.0818 9.44e-07**
Smoking status −0.02527 0.06692 n.s.
Mice HB-Val 1,3-butadiene exposure (log10 dose) 0.71574 0.10128 1.92e-11**
body weight pre-exposure −0.0069 0.0199 n.s.
Age 0.31376 0.06453 2.16e-06**
THB-Val 1,3-butadiene exposure (log10 dose) 0.13382 0.03715 3.88e-04*
body weight pre-exposure −0.00321 0.008232 n.s.
Age 0.017968 0.024287 n.s.
pyr-Val 1,3-butadiene exposure (log10 dose) 0.073513 0.021135 6.05e-04**
body weight pre-exposure −0.00379 0.004828 n.s.
Age −0.0015 0.013901 n.s.
Open in a new tab
**

P<0.001 and

*

P<0.01.

Next, we quantified the population variability of HB-Val, THB-Val, and pyr-Val adducts following 1,3-butadiene exposure in humans and mice (Table 2). These data enable translation to implied chemical-specific uncertainty factors for human variability (UFH) that can be directly applied in quantitative risk assessment of 1,3-butadiene. Because the hemoglobin adducts evaluated herein serve as biomarkers of metabolism, observed inter-individual variability in this case represents only the toxicokinetic (TK) portion of the overall uncertainty, UFH,TK, the default value of which is 10½ = 3.16 (WHO/IPCS, 2018). For data from the Collaborative Cross mouse population, the values represent total (combined inter- and intra-individual) variability because there was only one mouse per strain in each exposure group, so these values may over-estimate exact inter-strain variability. Nonetheless, the mouse values were generally smaller than the default UFH,TK, with the exception of the values for HB-Val at 2 and 20 ppm exposure levels, for which the confidence intervals also overlapped with the human values. For humans, these values represent the inter-individual variability, corrected for intra-individual variability using the intraclass correlation as described above. The observed human inter-individual variation was generally greater than the default UFH,TK, being up to 7.9 for 99% coverage of population variability. Only in case of THB-Val at 95% population coverage, the UFH,TK was lower (=2.31) than the default assumption. These values are consistent with the analysis of human toxicokinetic population variability data by the World Health Organization’s International Programme on Chemical Safety (WHO/IPCS, 2018), that found that at 95% protection, 90% of chemicals would be expected to have UFH,TK between 1.3 and 4.7, with the corresponding range at 99% protection being 1.4 to 8.8. Thus, the uncertainty factor derived here for all adducts from human exposures to 1,3-butadiene are not only based on the human data, but also they are generally concordant with the data from a mouse population, and are also consistent with human data for other chemicals.

Table 2.

Quantitation of variability in protein adduct levels following 1,3-butadiene exposure in humans and Collaborative Cross mice.

Humans (controlled for exposure, sex, and smoking) Mice (controlled for age and pre-exposure body weight)
1,3-Butadiene exposure, ppm 0.002–9.2 2 20 200
HB-Val
GM uncorrected (95% CI), pmol/g 0.56 (0.42–0.73) 7e-04
(6e-04–9e-04)		 0.010
(0.007–0.012)		 0.017
(0.016–0.018)
GSD uncorrected (95% CI), pmol/g 3.49 (2.95–4.37) 2.21 (1.95–2.66) 2.44 (2.12–3) 1.2 (1.17–1.25)
σ2 (total) 0.93 0.54 0.75 0.03
σ2 (within) 0.3 - - -
σ2 (across) 0.63 - - -
UFH,TK (95%, CI) 3.69 (3.26–4.30) 3.35 (2.74–4.55) 4.16 (3.3–5.88) 1.36 (1.3–1.46)
UFH,TK (99%, CI) 6.34 (5.32–7.86) 5.53 (4.15–8.52) 7.51 (5.4–12.25) 1.55 (1.44–1.71)
THB-Val
GM uncorrected (95% CI), pmol/g 211 (187–238) 1.04 (0.96–1.12) 1.19 (1.08–1.31) 1.76 (1.63–1.90)
GSD uncorrected (95% CI), pmol/g 2.19 (2.03–2.41) 1.32 (1.26–1.41) 1.42 (1.34–1.54) 1.33 (1.27–1.42)
σ2 (total) 0.38 0.07 0.11 0.0
σ2 (within) 0.12 - - -
σ2 (across) 0.26 - - -
UFH,TK (95%, CI) 2.31 (2.13–2.55) 1.57 (1.45–1.76) 1.74 (1.59–1.99) 1.61 (1.49–1.79)
UFH,TK (99%, CI) 3.27 (2.93–3.76) 1.89 (1.7–2.22) 2.19 (1.92–2.64) 1.95 (1.76–2.28)
pyr-Val
GM uncorrected (95% CI), pmol/g 0.073 (0.061–0.087) 0.032 (0.03–0.033) 0.034 (0.032–0.037) 0.04 (0.038–0.041)
GSD uncorrected (95% CI), pmol/g 3.44 (3.06–3.97) 1.2 (1.16–1.25) 1.26 (1.21–1.33) 1.23 (1.19–1.28)
σ2 (total) 1.16 0.03 0.05 0.04
σ2 (within) 0.37 - - -
σ2 (across) 0.79 - - -
UFH,TK (95%, CI) 4.31 (3.76–5.12) 1.35 (1.28–1.46) 1.44 (1.36–1.57) 1.4 (1.33–1.51)
UFH,TK (99%, CI) 7.90 (6.49–10.07) 1.53 (1.42–1.7) 1.67 (1.54–1.9) 1.61 (1.49–1.8)
Open in a new tab

Abbreviations: CI: confidence interval; GM, geometric mean; GSD, geometric standard deviation; σ2 (total), variance of log-transformed adduct levels from this study; σ2 (within), variance of log-transformed adduct levels within subject (only estimated for humans); σ2 (across), variance of log-transformed adduct levels across subjects (only estimated for humans); UFH,TK (95%), human toxicokinetic variability factor for the 95th percentile relative to the median; UFH,TK (99%), human toxicokinetic variability factor for the 99th percentile relative to the median.

Discussion

There are many biological factors that underlie individual susceptibility to chemical exposures, both occupationally (Christiani et al., 2008) and in the general population (Yeatts et al., 2006). The need for assessing interindividual variability when conducting human health risk assessments of chemicals is acute and a number of improvements have been proposed to the use of default assumptions when data are unavailable (WHO/IPCS, 2018). While extrapolations from human data available on a small number of chemicals are useful paths forward to estimate the range of variability and quantify the extent of uncertainty, data from experimental population models provide chemical-specific information and derivation of the uncertainty factors (Chiu and Rusyn, 2018; Harrill and McAllister, 2017; Zeise et al., 2013). The experimental population-based approaches include studies in genetically-defined human cell lines (Abdo et al., 2015), genetically-diverse rodent models (Harrill and McAllister, 2017), and the mechanistic and genome-wide association studies in exposed humans (Ritz et al., 2017). While there are many possible avenues for characterizing population variability experimentally, few options exist in reality when improved estimates of total uncertainty are needed for hazardous chemicals. In this respect, mouse population studies offer the opportunity to address population variability in many key areas that can be immediately translatable to advance the human health assessments of chemicals (Chiu and Rusyn, 2018). These include exposure assessment to characterize population toxicokinetic variability in internal dose, hazard identification to potentially reduce both false positive and false negative signals, dose-response assessment to quantitatively estimate the degree of toxicokinetic and/or toxicodynamic variability. Collectively, such data should increase confidence in setting chemical-specific health-protective exposure limits, and improve our understanding of the mechanisms of toxicity to identify key biological pathways that may be underpinning susceptibility.

A number of case studies have been published using Collaborative Cross (Threadgill and Churchill, 2012) and Diversity Outbred (Churchill et al., 2012) mouse populations. Such studies demonstrated the opportunities for improving risk assessment and decision-making, many even offered quantitative estimates of uncertainty factors that can be integrated into risk assessments and decision-making. For example, studies of chlorinated solvents tri- and tetra-chloroethylene have been used to refine toxicokinetic parameters (Chiu et al., 2014; Dalaijamts et al., 2020; Dalaijamts et al., 2021), characterize variability in toxicodynamics (Cichocki et al., 2017; Luo et al., 2019; Luo et al., 2018), propose population-based points-of-departure (Venkatratnam et al., 2018), and show species similarities in hazard traits at the population level (Yoo et al., 2015).

Other high production volume chemicals of public health concern for which the value of mouse population-wide studies has been demonstrated are benzene (French et al., 2015) and 1,3-butadiene (Chappell et al., 2017; Lewis et al., 2019). The study of benzene in the Diversity Outbred population (French et al., 2015) provided data that are immediately applicable to human risk assessment, such as demonstration of a dose-dependent increase in benzene-induced chromosomal damage and estimation of a benchmark concentration limit of 0.205 ppm benzene. Previous studies of 1,3-butadiene estimated variability for tissue-specific epigenetic and genotoxic phenotypes, data that are informative but not directly comparable to human exposures. More recently, a direct comparison of urinary levels of 1,3-butadiene DNA adducts in Collaborative Cross mice with those in occupationally-exposed humans was conducted (Erber et al., 2021). It showed that the degree of variability in urinary EB-GII adducts in mice, when expressed as an uncertainty factor for the inter-individual variability (UFH), was relatively modest (<3-fold), possibly due to metabolic saturation. By contrast, the variability in urinary EB-GII (when adjusted for exposure) for humans was larger than that in mice or the default value.

The findings reported in Erber et al. (2021) prompted us to examine what other data may be available to provide further confidence in UFH for 1,3-butadiene. In this regard, protein adducts have been an important component in risk assessment of 1,3-butadiene as they allow estimation of the internal dose of its reactive epoxides (Swenberg et al., 2011). To date, no previous attempt was made to quantify uncertainty in toxicokinetics of 1,3-butadiene using published human data on hemoglobin adducts (Albertini et al., 2003; Albertini et al., 2007; Boysen et al., 2012). In addition, a study in a Collaborative Cross mouse population (Nellis et al., 2021) provided another dataset for this chemical that enables species comparisons in UFH,TK. Together, these data allowed us to estimate uncertainty in each species and to offer comparisons to other data for 1,3-butadiene.

When comparing with a previous study that used pro-mutagenic DNA adducts of 1,3-butadiene (Erber et al., 2021), which included a toxicodynamic component in addition to toxicokinetics, several observations are noteworthy. First, in a study of Collaborative Cross mouse population (Nellis et al., 2021), the HB-Val hemoglobin adducts showed marked decrease in variability at the highest exposure of 200 ppm. This is consistent with a hypothesis that metabolic saturation would reduce overall variability at high exposures, a suggestion made by Erber et al. (2021) to explain the relatively small UFH values estimated for the urinary DNA adducts in mice. Additionally, except for the lower two doses of HB-Val, the UFH,TK estimated here (Table 2) is similar to or slightly smaller than the overall UFH reported for DNA adducts by Erber et al. (2021), consistent with a small toxicodynamic component to variability in mice at highest (600 ppm) exposures. For data from humans, it should be noted that the toxicokinetic variability observed here is also less than the combined toxicokinetic and toxicodynamic variability reported for DNA adducts in mouse studies (Erber et al., 2021). This observation is consistent with the notion that the majority of overall variability being due to toxicodynamics rather than toxicokinetics, because the protein adducts σ2(across) values are less than half the σ2(across) values for DNA adducts. This observation is also consistent with previous analyses by WHO/IPCS (WHO/IPCS, 2018), which found that toxicodynamic variability has a larger contribution to overall variability than toxicokinetic does. Finally, the much lower variability observed in Collaborative Cross mice, both for toxicokinetics in this study and combined variability in Erber et al. (2021), suggests either that mice are not an optimal model for human variability studies of 1,3-butadiene (Swenberg et al., 2011), or that saturation effects in both metabolism and toxicodynamics may be misrepresenting the observed variability.

Overall, there is now a compendium of studies of 1,3-butadiene exposure in the Collaborative Cross mouse population that examined population variation in several biomarkers ranging from protein adducts (this study), DNA adducts (Erber et al., 2021; Lewis et al., 2019), and epigenetic markers (Lewis et al., 2019). Collectively, these studies provide quantitative empirical information on population variability, and again it is noteworthy that the degree of quantified variability is highly dependent on the chosen biomarker, tissue, and dose. For instance, at high exposures (≥200 ppm), the experimentally derived UFH were generally no more than 3-fold for most biomarkers (Lewis et al., 2019). Only the current study examined lower exposures in mice, and only for blood HB-Val was the derived UFH greater than 3. By contrast, analyses of human data found generally greater variability, but again this was highly dependent on the chosen biomarker – from more than 10-fold for EB-GII (Erber et al., 2021) to no more than 3-fold for blood THB-Val (current study).

Unfortunately, we did not have data on genetic polymorphisms or activities of key enzymes in the metabolic pathways to test hypotheses as to the mechanistic basis of the observed variation. Previous analyses of human population variability in other substrates for CYP2E1 and CYP2A6, key enzymes in 1,3-butadiene metabolism, suggest that their contributions to variability would be less than the 3-fold default (Dorne et al., 2004), so it is likely that other factors are also influential for biomarkers with greater than default variation. Detailed mechanistic studies using a small number of inbred strains (Chappell et al., 2017; Israel et al., 2018) provide additional molecular insights into the observed variability in strain- and organ-specific effects of 1,3-butadiene, suggesting that pre-exposure differences among individuals in transcriptional and epigenetic states may influence responses to exposure. Moreover, these mechanistic studies suggest that genetics alone may not be the dominant driver of variability in susceptibility. Indeed, the apparent discordance in the estimated degree of variability between the human and Collaborative Cross mouse populations may be consistent with this observation, because while genetically distinct, mouse strains are still treated under otherwise nearly identical conditions. By contrast, human populations include not only genetic variation, but variation across a wide range of various “background” conditions and lifestyles which may be highly influential in variation in susceptibility (Zeise et al., 2013). “Bottom-up” approaches have been developed with some success at capturing population variation in toxicokinetics through use of physiologically-based models and population-based data on components of variation (Ring et al., 2017), but this has to be done with care because individual sources of variation are not additive when put into a physiological context. Hence, such an approach has not yet been attempted for toxicodynamics in part due to incomplete knowledge as to the sources of variation to measures such as adduct levels and their interaction in a whole organism. Rather, most of the approaches thus far have been, like those reported here, are empirically-based.

In terms of application to regulatory decision-making, the findings reported herein and those published in Erber et al. (2021) may be considered in the process of addressing human variability as part of the development of an oral slope factor for cancer risk of 1,3-butadiene, a key gap in the practice of cancer risk assessment identified by the National Academies Science and Decisions report (National Research Council, 2009). Specifically, because the current inhalation slope factor for 1,3-butadiene is based on epidemiological data, it represents the arithmetic mean risk over the human population, so for population-based risk contexts, such as economic benefit-cost analysis, variability is already averaged over across the population. However, in most risk assessment contexts, the goal is protection against individual risk, in particular ensuring the protection of more sensitive members of the population (National Research Council, 2009). In such cases, the population-level risk is not the same as the individual-level risk, and under the usual assumption of a lognormal distribution of variation in susceptibility, the relationship between the two depends strongly on the geometric standard deviation of the distribution. Therefore, unlike for non-cancer effects, where the uncertainty factor is the ratio between the sensitive individual and the median, for population cancer risk derived from epidemiologic studies, the adjustment for individual sensitivity is the ratio between the sensitive individual and the mean (which is >median). The key issue for the regulatory agencies, as noted above, will be deciding on the appropriate biomarker(s) to be used.

Overall, this study characterized variability in protein adducts formed as a result of 1,3-butadiene inhalational exposure in both mouse and human populations. The degree of variability was found to be both dose- and adduct-dependent, it appeared to be greater in human populations than in the mouse population. Taken in the context of other recent studies of population variation using a wide range of exposure and effect biomarkers of 1,3-butadiene, several conclusions can be drawn. First, measured population variability is highly dependent on choice of biomarker (e.g., adducts or other molecular events), and thus either mechanistic or empirical data are needed to select biomarkers, individually or in combination, that are most predictive of the ultimate outcome (in this case, cancer). Additionally, for complex outcomes such as cancer, genetic variability among individuals may not be the main driver of susceptibility, so experimental studies in genetically diverse mouse populations such as the Collaborative Cross may only provide a lower bound as to the degree of human variability. Thus, despite their limitations, human biomarker studies can provide a useful complement to experiment studies in characterizing human variability. In conclusion, these studies of 1,3-butadiene provide a useful case study for implementing the National Academies Science and Decisions report’s (National Research Council, 2009) recommendation to formally include human variability into cancer dose-response modeling and risk characterization.

References

  1. Abdo N, et al. , 2015. Population-based in vitro hazard and concentration-response assessment of chemicals: the 1000 genomes high-throughput screening study. Environ Health Perspect. 123, 458–466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Albertini RJ, et al. , 2003. Biomarkers in Czech workers exposed to 1,3-butadiene: a transitional epidemiologic study. Res Rep Health Eff Inst. 1–141; discussion 143–62. [PubMed]
  3. Albertini RJ, et al. , 2007. Molecular epidemiological studies in 1,3-butadiene exposed Czech workers: female-male comparisons. Chem Biol Interact. 166, 63–77. [DOI] [PubMed] [Google Scholar]
  4. Boogaard PJ, 2002. Use of haemoglobin adducts in exposure monitoring and risk assessment. J Chromatogr B Analyt Technol Biomed Life Sci. 778, 309–22. [DOI] [PubMed] [Google Scholar]
  5. Boysen G, et al. , 2012. Formation of 1,2:3,4-diepoxybutane-specific hemoglobin adducts in 1,3-butadiene exposed workers. Toxicol Sci. 125, 30–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Boysen G, et al. , 2004. Analysis of diepoxide-specific cyclic N-terminal globin adducts in mice and rats after inhalation exposure to 1,3-butadiene. Cancer Res. 64, 8517–20. [DOI] [PubMed] [Google Scholar]
  7. Boysen G, et al. , 2007. N-terminal globin adducts as biomarkers for formation of butadiene derived epoxides. Chem Biol Interact. 166, 84–92. [DOI] [PubMed] [Google Scholar]
  8. Boysen G, et al. , 2019. A simplified method for detection of N-terminal valine adducts in patients receiving treosulfan. Rapid Commun Mass Spectrom. 33, 1635–1642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Carmical JR, et al. , 2000. Mutagenic potential of adenine N(6) adducts of monoepoxide and diolepoxide derivatives of butadiene. Environ Mol Mutagen. 35, 48–56. [PubMed] [Google Scholar]
  10. Chappell G, et al. , 2014. Epigenetic events determine tissue-specific toxicity of inhalational exposure to the genotoxic chemical 1,3-butadiene in male C57BL/6J mice. Toxicol Sci. 142, 375–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chappell GA, et al. , 2017. Variation in DNA-Damage Responses to an Inhalational Carcinogen (1,3-Butadiene) in Relation to Strain-Specific Differences in Chromatin Accessibility and Gene Transcription Profiles in C57BL/6J and CAST/EiJ Mice. Environ Health Perspect. 125, 107006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chiu WA, et al. , 2014. Physiologically-Based Pharmacokinetic (PBPK) Modeling of Inter-strain Variability in Trichloroethylene Metabolism in the Mouse. Environ Health Perspect. 122, 456–463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Chiu WA, Rusyn I, 2018. Advancing chemical risk assessment decision-making with population variability data: challenges and opportunities. Mamm Genome. 29, 182–189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Christiani DC, et al. , 2008. Genetic susceptibility to occupational exposures. Occup Environ Med. 65, 430–6; quiz 436, 397. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Churchill GA, et al. , 2012. The Diversity Outbred mouse population. Mamm Genome. 23, 713–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Cichocki JA, et al. , 2017. Characterization of Variability in Toxicokinetics and Toxicodynamics of Tetrachloroethylene Using the Collaborative Cross Mouse Population. Environ Health Perspect. 125, 057006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Dalaijamts C, et al. , 2020. PBPK modeling of impact of nonalcoholic fatty liver disease on toxicokinetics of perchloroethylene in mice. Toxicol Appl Pharmacol. 400, 115069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Dalaijamts C, et al. , 2021. Quantitative Characterization of Population-Wide Tissue- and Metabolite-Specific Variability in Perchloroethylene Toxicokinetics in Male Mice. Toxicol Sci. 182, 168–182. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Dorne JL, et al. , 2004. Human variability for metabolic pathways with limited data (CYP2A6, CYP2C9, CYP2E1, ADH, esterases, glycine and sulphate conjugation). Food Chem Toxicol. 42, 397–421. [DOI] [PubMed] [Google Scholar]
  20. Elfarra AA, et al. , 2001. Reactive metabolites of 1,3-butadiene: DNA and hemoglobin adduct formation and potential roles in carcinogenicity. Adv Exp Med Biol. 500, 93–103. [DOI] [PubMed] [Google Scholar]
  21. Erber L, et al. , 2021. Intra- and Inter-Species Variability in Urinary N7-(1-Hydroxy-3-buten-2-yl)guanine Adducts Following Inhalation Exposure to 1,3-Butadiene. Chem Res Toxicol. [DOI] [PMC free article] [PubMed]
  22. French JE, et al. , 2015. Diversity Outbred Mice Identify Population-Based Exposure Thresholds and Genetic Factors that Influence Benzene-Induced Genotoxicity. Environ Health Perspect. 123, 237–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Georgieva NI, et al. , 2010. Exposure-response of 1,2:3,4-diepoxybutane-specific N-terminal valine adducts in mice and rats after inhalation exposure to 1,3-butadiene. Toxicol Sci. 115, 322–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Harrill AH, McAllister KA, 2017. New Rodent Population Models May Inform Human Health Risk Assessment and Identification of Genetic Susceptibility to Environmental Exposures. Environ Health Perspect. 125, 086002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. IARC, 2008. IARC monographs on the evaluation of carcinogenic risks to humans. Volume 97. 1,3-butadiene, ethylene oxide and vinyl halides (vinyl fluoride, vinyl chloride and vinyl bromide). IARC Monogr Eval Carcinog Risks Hum. 97, 3–471. [PMC free article] [PubMed] [Google Scholar]
  26. Israel JW, et al. , 2018. Tissue- and strain-specific effects of a genotoxic carcinogen 1,3-butadiene on chromatin and transcription. Mamm Genome. 29, 153–167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Jokipii Krueger CC, et al. , 2020. Urinary N7-(1-hydroxy-3-buten-2-yl) guanine adducts in humans: temporal stability and association with smoking. Mutagenesis. 35, 19–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kotapati S, et al. , 2015. High throughput HPLC-ESI(−)-MS/MS methodology for mercapturic acid metabolites of 1,3-butadiene: Biomarkers of exposure and bioactivation. Chem Biol Interact. 241, 23–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Koturbash I, et al. , 2011. Epigenetic mechanisms of mouse interstrain variability in genotoxicity of the environmental toxicant 1,3-butadiene. Toxicol Sci. 122, 448–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lewis L, et al. , 2019. Population-Based Analysis of DNA Damage and Epigenetic Effects of 1,3-Butadiene in the Mouse. Chem Res Toxicol. 32, 887–898. [DOI] [PubMed] [Google Scholar]
  31. Luo YS, et al. , 2019. Using Collaborative Cross Mouse Population to Fill Data Gaps in Risk Assessment: A Case Study of Population-Based Analysis of Toxicokinetics and Kidney Toxicodynamics of Tetrachloroethylene. Environ Health Perspect. 127, 67011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Luo YS, et al. , 2018. Characterization of inter-tissue and inter-strain variability of TCE glutathione conjugation metabolites DCVG, DCVC, and NAcDCVC in the mouse. J Toxicol Environ Health A. 81, 37–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mowrer J, et al. , 1985. Modified Edman degradation applied to hemoglobin for monitoring occupational exposure to alkylating agents. Toxicol.Environm.Chem. 11, 215–231. [Google Scholar]
  34. National Research Council, 2009. Science and Decisions: Advancing Risk Assessment. National Academies Press, Washington, DC. [PubMed] [Google Scholar]
  35. Nellis M, et al. , 2021. Lung metabolome of 1,3-butadiene exposed Collaborative Cross mice reflects metabolic phenotype of human lung cancer. Toxicology. 463, 152987. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Ogawa M, et al. , 2006. Hemoglobin adducts as a marker of exposure to chemical substances, especially PRTR class I designated chemical substances. J Occup Health. 48, 314–28. [DOI] [PubMed] [Google Scholar]
  37. Ring CL, et al. , 2017. Identifying populations sensitive to environmental chemicals by simulating toxicokinetic variability. Environ Int. 106, 105–118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Ritz BR, et al. , 2017. Lessons Learned From Past Gene-Environment Interaction Successes. Am J Epidemiol. 186, 778–786. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Soeteman-Hernandez LG, et al. , 2013. Tobacco smoke-related health effects induced by 1,3-butadiene and strategies for risk reduction. Toxicol Sci. 136, 566–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Sullivan M, 2010. Statistics : informed decisions using data. Prentice Hall is an imprint of Pearson, Upper Saddle River, N.J. [Google Scholar]
  41. Swenberg JA, et al. , 2011. 1,3-Butadiene: Biomarkers and application to risk assessment. Chem Biol Interact. 24, 809–817. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Swenberg JA, et al. , 1,3-Butadiene: cancer, mutations, and adducts. Part V. Hemoglobin adducts as biomarkers of butadiene exposure and metabolism. Vol. 92. Health Effects Institute, Cambridge,MA, 2000, pp. 191–209. [PubMed] [Google Scholar]
  43. Swenberg JA, et al. , 2001. Using DNA and hemoglobin adducts to improve the risk assessment of butadiene. Chem Biol Interact. 135–136, 387–403. [DOI] [PubMed]
  44. Threadgill DW, Churchill GA, 2012. Ten years of the Collaborative Cross. Genetics. 190, 291–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. US EPA, 1,3-Butadiene [CASRN 106–99-0]: Quantitative Estimate of Carcinogenic Risk from Inhalation Exposure. In: National Center for Environmental Assessment, (Ed.). Integrated Risk Information System (IRIS), US Environmental Protection Agency, Washington, DC, 2002. [Google Scholar]
  46. van Sittert NJ, et al. , 2000. Biomarkers of exposure to 1,3-butadiene as a basis for cancer risk assessment. Toxicol Sci. 56, 189–202. [DOI] [PubMed] [Google Scholar]
  47. Venkatratnam A, et al. , 2018. Population-based dose-response analysis of liver transcriptional response to trichloroethylene in mouse. Mamm Genome. 29, 168–181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. WHO/IPCS, Guidance document on evaluating and expressing uncertainty in hazard characterization. World Health Organization & International Programme on Chemical Safety, Geneva, Switzerland, 2018, pp. 159 p. [Google Scholar]
  49. Yeatts K, et al. , 2006. A brief targeted review of susceptibility factors, environmental exposures, asthma incidence, and recommendations for future asthma incidence research. Environ Health Perspect. 114, 634–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Yoo HS, et al. , 2015. Comparative analysis of the relationship between trichloroethylene metabolism and tissue-specific toxicity among inbred mouse strains: kidney effects. J Toxicol Environ Health A. 78, 32–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Zeise L, et al. , 2013. Addressing human variability in next-generation human health risk assessments of environmental chemicals. Environ Health Perspect. 121, 23–31. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES