Benzo[a]pyrene (BaP) metabolites predominant in human plasma following escalating oral micro-dosing with [¹⁴C]-BaP

Abstract

Benzo[a]pyrene (BaP) is formed by incomplete combustion of organic materials (petroleum, coal, tobacco, etc.). BaP is designated by the International Agency for Research on Cancer as a group 1 known human carcinogen; a classification supported by numerous studies in preclinical models and epidemiology studies of exposed populations. Risk assessment relies on toxicokinetic and cancer studies in rodents at doses 5–6 orders of magnitude greater than average human uptake. Using a dose-response design at environmentally relevant concentrations, this study follows uptake, metabolism, and elimination of [¹⁴C]-BaP in human plasma by employing UPLC - accelerator mass spectrometry (UPLC-AMS). Volunteers were administered 25, 50, 100, and 250 ng (2.7–27 nCi) of [¹⁴C]-BaP (with interceding minimum 3-week washout periods) with quantification of parent [¹⁴C]-BaP and metabolites in plasma measured over 48 hours. [¹⁴C]-BaP median T_max was 30 minutes with C_max and area under the curve (AUC) approximating dose-dependency. Marked inter-individual variability in plasma pharmacokinetics following a 250 ng dose was seen with 7 volunteers as measured by the C_max (8.99 ± 7.08 ng × mL⁻¹) and AUC_0–48hr (68.6 ± 64.0 fg × hr⁻¹ × mL⁻¹). Approximately 3–6% of the [¹⁴C] recovered (AUC_{0–48 hr}) was parent compound, demonstrating extensive metabolism following oral dosing. Metabolite profiles showed that, even at the earliest time-point (30 min), a substantial percentage of [¹⁴C] in plasma was polar BaP metabolites. The best fit modeling approach identified non-compartmental apparent volume of distribution of BaP as significantly increasing as a function of dose (p=0.004). Bay region tetrols and dihydrodiols predominated, suggesting not only was there extensive first pass metabolism but also potentially bioactivation. AMS enables the study of environmental carcinogens in humans with de minimus risk, allowing for important testing and validation of physiologically based pharmacokinetic models derived from animal data, risk assessment, and the interpretation of data from high-risk occupationally exposed populations.

Graphical Abstract

1. Introduction

1.1. BaP formation and carcinogenicity

Incomplete combustion of carbon based materials such as petroleum products (Sadiktsis et al., 2012; Titaley et al., 2016), coal (Simoneit et al., 2007), wood (Li et al., 2011), and tobacco (Hecht, 1999; Staretz et al., 1997; Stepanov et al., 2010) produces hundreds of polycyclic aromatic hydrocarbons (PAHs) including benzo[a]pyrene (BaP). The International Agency for Research on Cancer classifies BaP as a Group 1 known carcinogen for humans (IARC, 2021)_. Found at 524 hazardous waste sites on the National Priorities List (NPL), BaP is ranked number 8 out of 275 chemicals on the Priority List of Hazardous Substances (ATSDR, 2020)_. While frequently occurring at NPL sites, BaP is ubiquitous in the environment and often used as an indicator compound to measure PAH exposure in humans and other animals (ATSDR, 2020; Bostrom, et al., 2002; Usenko et al., 2010, 2007). BaP is also included in U.S. EPA’s Persistent Bioaccumulative and Toxic Chemical Program where it is used as an index chemical for deriving Relative Potency Factors (BaP = 1) to estimate carcinogenicity of other individual PAHs or PAH mixtures (U.S. EPA, 2017, 2010, 2013)_. Many types of cancers are linked to BaP exposure along with developmental, immunological, cardiovascular, reproductive, and neurological toxicities (Miller and Ramos, 2001; Nebert, 1989; U.S. EPA, 2017)_.

1.2. BaP metabolism and DNA adduction

Risk following exposure is largely determined by metabolism as BaP is bioactivated or detoxified by a number of phase 1 and phase 2 enzymes in various tissues (Bauer et al., 1995; Cavalieri and Rogan, 1995; Conney et al., 1994; Kim et al., 1998; Penning et al., 1996; Shimada et al., 1999; Shimada and Fujii-Kuriyama, 2004). Cytochromes P-450 in the 1 family (CYP1A1, 1A2, and 1B1) are the most efficient of the CYPs in activation of BaP by oxygenation at the bay region to form BaP-7,8-epoxide (endo or exo) (Bauer et al., 1995; Hecht et al., 2010; Kim et al., 1998; Shimada et al., 1999; Shimada and Fujii-Kuriyama, 2004). After action by epoxide hydrolase to form two trans-BaP-7,8-dihydrodiols ((±)-trans-BaP-DHD), a second epoxygenation by CYP1 produces (±)-trans/cis-BaP-7,8-dihydrodiol-9,10-epoxides (BaP-DHDE, 4 enantiomers with the (+)-anti-BaP-DHDE being the most carcinogenic (Figure 1)). The electrophilic BaP-DHDE covalently binds to deoxyguanine (dG) or deoxyadenine with a preference for the N² position of dG resulting in G to T transversion (Bauer et al., 1995; Cavalieri and Rogan, 1995; Conney et al., 1994; Hecht et al., 2010; Kim et al., 1998; Meinl et al., 2008; Melendez-Colon et al., 1999; Penning et al., 1996; Shimada et al., 1999, 1997; Shimada and Fujii-Kuriyama, 2004). BaP is also an aryl hydrocarbon receptor (AHR) agonist and up-regulates CYP1, increasing the likelihood of CYP1-dependent bioactivation. Further, BaP-mediated AHR activation also stimulates the mitogen activated protein kinase (MAPK) signaling cascade, altering cellular processes including those fundamental to mitigating effects like BaP-induced DNA damage (Vázquez-Gómez et al., 2018).

Figure 1.

Metabolism of benzo[a]pyrene and adduction to DNA (formation of diones not depicted).

1.3. BaP levels in humans

Concentrations of BaP in blood, urine, and tissue, as well as covalent adducts, are used as biomarkers of BaP exposure and risk (Lodovici et al., 1998; Obana et al., 1981; Ross et al., 1995). Human exposure can occur as a result of environmental contamination, like an oil spill or wildfire, through ingestion, smoking, or occupational exposures such as asphalt paving or iron work (Brandt and Watson, 2003; Pastorelli et al., 2000; Tuominen et al., 2002). Roofers are exposed to coal tar dust as well as hot roofing tar. A study by Herbert et al. (1990) reported that BaP exposure (as per personal air monitors and skin wipes) in roofers averaged 0.8 μg/m³ resulting in a 10 fold increase in DNA adducts when compared to controls (1.3/10⁸ nucleotides compared to 0.13/10⁸ nucleotides, respectively). BaP has been identified in drinking water (4–24 ng/L; IPCS, 1998; Mumtaz, 1995), prepared meals (mean of 0.14 and a maximum of 1.15 μg/day; Butler et al., 1993)) and human milk (mean of 6.5 ng/kg, n = 51 in Japan; IPCS, 1998). Dietary intake from packaging, processing/cooking, and deposition from the atmosphere to food and water are the greatest contributors of BaP intake in the general human population (Bansal and Kim, 2015; Domingo and Nadal, 2015; Kazerouni et al., 2001). Human liver and fat samples from autopsied individuals have shown average concentrations of BaP to be 0.020 μg/kg (or 200 ng/kg, wet weight), with no significant differences noted between tissue, age, or sex (n=6; Obana et al., 1981). Rodent PBPK models have previously been published by our group (Crowell et al., 2011) and AMS studies suggest these can be applied for human low dose exposures (Madeen et al., 2015, 2016, 2019).

1.4. Accelerator Mass Spectrometry and dosing humans with carcinogens

Graphite AMS has been employed to determine DNA adduction in target tissues of numerous [¹⁴C]-labeled carcinogens, including aflatoxin B₁ (AFB₁), BaP, dibenzo[def,p]chrysene (DBC), 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx) and 2-amino-1-methyl-6-phenylimidazole[4,5-b]pyridine (PhIP) (Table S1). Preliminary studies employing AMS with [¹⁴C]-BaP in human subjects suggested considerable metabolism as determined by plasma levels of the parent PAH following oral micro-dosing (Madeen et al., 2019). These studies measured only total [¹⁴C], however the coupling of UPLC and AMS by development of a “moving wire” interface now allows for the identification of metabolites (Thomas et al., 2011) and with UPLC–AMS the metabolic profile of [¹⁴C]-BaP can be determined. Further, the sensitivity of this method of detection allows for human studies with minimal risk, providing data on a known carcinogen at environmental levels.

Seven human volunteers were given escalating doses of [¹⁴C]-BaP (25 to 250 ng) to further investigate potential bioactivation and metabolism of the compound. Plasma extracts (0–48 hours) analyzed by UPLC-AMS revealed the extent of [¹⁴C]-BaP metabolism and inter-individual variability, plus identification and quantitation of ten metabolites. With a dose range of 25–250 ng we observed considerable first pass metabolism and metabolic profiles suggested bioactivation over detoxification. These observations offer insights into in situ human metabolism of a key PAH at actual environmental exposure concentrations, providing important information for future risk assessment and subsequent models.

2. Materials and methods

2.1. Human subjects

2.1.1. Recruitment, enrollment, and pre-dosing protocol

Seven subjects (5 males and 2 females, Table S2) ranging from 26 to 65 years old with BMIs of 19.0–33.3 were recruited and enrolled under the guidelines of an FDA-IND (#117175) as well as IRBs at both Oregon State University (OSU; #8233) and Lawrence Livermore National Laboratory (LLNL; Protocol #2017–008). The study details can be found at ClinicalTrials.gov (NCT03318978). Inclusion and exclusion criteria have been previously described in detail (Madeen et al., 2019). This study protocol required abstaining from consumption of charcoal-broiled meats, smoked meats and cheeses, as well as cruciferous vegetables and condiments for two weeks prior to administration of [¹⁴C]-BaP and throughout the 48 hour study duration. Dietary questionnaires (designed to assess total BaP consumption) were collected for the 3 days immediately preceding dosing through the final blood draw. Just prior to ingestion of [¹⁴C]-BaP, blood was drawn for determination of background levels of 62 PAHs by GC-MS/MS (Anderson et al., 2015). Doses were separated with a washout period of three weeks minimum with all four doses administered in less than one year.

2.1.2. Micro-dosing with [¹⁴C]-BaP, timed blood collection, and preparation of plasma

Participants fasted overnight prior to ingesting the morning capsule containing 25, 50, 100, or 250 ng of [¹⁴C]-BaP (2.7, 5.4, 10.8, or 27 nCi, respectively) with 200–250 ml water. Preparation of the food-grade [¹⁴C]-BaP capsule, as well as quality control of the amount and purity of [¹⁴C]-BaP, has been described previously (Hummel et al., 2018). Fasting continued for 2 hours following dosing. Blood (8–10 mL) was drawn into heparinized tubes prior to dosing (time 0) and at 0.25, 0.5, 1, 1.5, 2, 3, and 4 hours post-dose using an in-dwelling catheter placed in a vein in the antecubital area. Later blood collections at 8, 24, and 48 hours post-dose were done by needle stick. Blood was centrifuged (rcf`~1000 for 10 min) immediately following the draw and plasma extracted within 2 hours of collection.

2.2. Extraction of plasma and UPLC-AMS at LLNL

Following centrifugation, samples were held at 4°C until extraction with established methods (Crowell et al., 2011; Hummel et al., 2018). Briefly, plasma aliquots (1.5 mL) were acidified and vortexed with ethyl acetate (3x at 1.5 mL) in the presence of K₂SO₄. The ethyl acetate layers of each sample were pooled and, following evaporation to dryness under nitrogen, reconstituted with 100 μL ethyl acetate and stored at −80°C until shipped on dry ice to the BioAMS Center at LLNL or GC-MS/MS analysis by the Analytical Chemistry Core of the Oregon State University Superfund Research Program.

Just prior to UPLC-AMS analysis, plasma samples stored at −80°C were evaporated to dryness under vacuum and reconstituted in 50 or 100 μL acetonitrile depending on initial dose of [¹⁴C]-BaP (25 and 50 ng into 50 μL with 100 and 250 ng in 100 μL, respectively). Samples were maintained in the autosampler of a Waters Acquity UPLC H-Class System at 20°C, the column at 28°C, and injection volume set at 10 μL. A Waters Acquity UPLC BEH C₁₈ 1.7 μm 2.1 × 50 mm column with a Vanguard HSS PFP 1.8 μm 2.1 × 5 mm guard column was used for separation with 0.3% formic acid in water (A) and PEG-free acetonitrile (B). Elution of [¹⁴C]-BaP and metabolites was achieved over 20 min at 0.25 mL/minute utilizing a gradient (0.0 min 70:30 A:B; 0.1 60:40; 10.0 54:46; 10.1 43:57; 14.3 to 16.3 0:100; 16.4 to 20 min 70:30). [¹⁴C]-BaP metabolites were identified by retention time using analytical standards obtained from Oregon State University Superfund Program PAH repository (http://limsweb.science.oregonstate.edu/).

Elution times of metabolites and parent compound included in the analytical standard are in Table 1.

Table 1.

Elution time of BaP and BaP metabolite standards.

Compounds	Elution Time (min)	Compounds	Elution Time (min)
BaP tetrols (mix)	2.16	6,12 - dione BaP	9.50
9,10 - dihydrodiol BaP	2.68	7,8 - dione BaP	10.23
4,5 - dihydrodiol BaP	4.20	9 - OH BaP	12.59
7,8 - dihydrodiol BaP	4.56	3 - OH BaP	12.95
1,6 - dione BaP	7.40	[14C] - BaP / BaP	15.01
3,6 - dione BaP	8.16

After passing through a PDA detector, the UPLC eluate was split with 0.12 mL/min (48%) deposited onto the moving wire interface for AMS analysis with the rest sent to waste. The eluate was carried by periodically indented nickel wire through a drying oven then a combustion oven, evaporating solvent and combusting analytes to [¹⁴C]-CO_2, allowing transfer to the CO₂-gas-accepting ion source of the 250 kV AMS (Ognibene et al., 2019; Thomas et al., 2011). Measurements of ¹³C and ¹⁴C with known isotope ratios allow for calculations of [¹⁴C]-BaP and [¹⁴C]-metabolite concentrations. Prior to injection of each sample the precision and accuracy of the moving wire interface and AMS were confirmed using an aqueous sucrose reference solution (Keck et al., 2010; Ognibene et al., 2015). Established validation parameters include accuracy (1–3%), precision (expressed as coefficient of variation, 1–6%), and sensitivity expressed as limit of detection (LOD) of 1 attomole of [¹⁴C] (0.58 fg for [¹⁴C]-BaP) (Keck et al., 2010). The lower limit of quantitation (LOQ, 0.60 fg/mL) was established using extracted plasma blanks from each individual and measurements at the known elution times for each compound.

2.2.1. UPLC-AMS separation, detection, and identification

Biological samples naturally contain carbon isotopes that can change background levels of AMS measurements of ¹²C, ¹³C, and ¹⁴C and these vary by source article, individual, and sample type (Keck et al., 2010; Salehpour et al., 2008). AMS counts individual ¹⁴C atoms in relation to ¹²C and/or ¹³C ion currents with extremely high sensitivity and any variation in the background signal of isotopes can change the degree of confidence of LOQ. To increase confidence, a matrix-matched LOQ was determined using the standard deviation of the signal in the blank samples (t = 0 h) from each participant. These extracted blanks have the same matrix effects as study samples and by time-matching the signal to noise ratio from blanks and study samples the LOQ takes into account whether the peaks seen are the compounds of interest or are part of the background noise.

$$LOQ=10\times \text{signal}\text{SD}\text{in}\text{blanks}\left(\text{at}\text{respective}\text{elution}\text{time}\right)$$

Subsequently, the calculated LOQ for each compound showed little variation and values were rounded up to a single method LOQ of 0.60 fg/mL plasma (calculated range 0.55–0.59 fg/mL for all compounds) that was consistent and uniform across all participants. With AMS analysis the ¹⁴C naturally present in the plasma blanks is measured with the same precision as a low-signal sample, supporting this method of determining the LOQ. (FDA, 1995; Keck et al., 2010)

The radiolabeled carbon detected is either part of the naturally occurring [¹⁴C] in humans or from the dosed [¹⁴C]-BaP. Due to the unavailability of radiolabeled BaP metabolite standards, chromatography and metabolite identification cannot be confirmed with AMS directly. In order to use BaP metabolite standards with known retention times to identify metabolites, the UPLC eluate went to a PDA detector and a moving wire interface for AMS analysis, however not all peaks seen using UPLC-AMS could be identified using the analytical standard mix.

Unknown A eluted at the beginning of the chromatography run (0.48 min) and is likely composed of conjugated metabolites. BaP phenols, dihydrodiols, dihydrodiol-epoxides and diones are all subject to conjugation (glucuronides, sulfates, and glutathione) and would be the most polar BaP metabolites in plasma. Further research is in progress to separate and identify conjugates in both the urine and plasma of participants. Based on retention time Unknowns B (3.16 min) and C (4.85 min) are possibly additional dihydrodiols. Unknown D (5.39 min) is either a late eluting dihydrodiol or a dione. The peak identified as the 3-phenol may also contain small amounts of the minor metabolites 7- and 1-BaP phenols as these hydroxylated compounds co-elute if present.

2.3. Pharmacokinetic analysis

Pharmacokinetics of [¹⁴C]-BaP, metabolites, and unknowns were evaluated using non-compartmental analyses. The area under the curve (AUC) of [¹⁴C]-BaP and metabolite concentrations in plasma from time zero to the last measured time point or extrapolated to infinity ((AUC_0−∞) using the last 3 time points) were calculated using the trapezoidal rule (Gibaldi and Perrier, 1982). Mean residence times were calculated as a ratio of AUC under the first moment curve extrapolated to infinity to AUC extrapolated to infinity. Non-compartmental half-lives were calculated as the product of mean residence times with the natural log of two.

Pharmacokinetics of parent [¹⁴C]-BaP was also evaluated using compartmental analysis. A two-compartment pharmacokinetic model was used to evaluate the time course of [¹⁴C]-BaP (Equations 1–3), where “A0”, “A1”, and “A2” represent the amount (fg) of [¹⁴C]-BaP in the absorption, central, and peripheral compartments, respectively; “ka”, “ke”, “k12”, and “k21” are first-order rate constants. The concentration of the central compartment (C1) was calculated by normalizing the amount with the apparent volume of distribution (V1, L, Equation 4). Optimizations of model parameters were obtained using a maximum log likelihood objective, and all data points were weighted equally when fitting. Initial values were set by adjusting parameters visually.

$$\frac{d{A}_0}{dt}=-\left(k_{\text{a}}\times A_0\right)$$ (1)

$$\frac{d{A}_1}{dt}=\left(k_{\text{a}}\times A_0\right)-\left(k_e\times A_1\right)-\left(k_{12}\times A_1\right)+\left(k_{21}\times A_2\right)$$ (2)

$$\frac{d{A}_2}{dt}=\left(k_{12}\times A_1\right)-\left(k_{21}\times A_2\right)$$ (3)

$$C_1=\frac{A_1}{V_1}$$ (4)

Pharmacokinetic parameters were evaluated for linearity using a best fit modeling approach. First, pharmacokinetic parameters were evaluated for significant change as a function of dose using a standard linear regression model including a fit y-intercept. Slopes were compared using a t-test and an alpha value of 0.05. If the parameter changed as a function of dose, we further evaluated the parameter with a linear regression model (Equation 5), where k is the slope through the origin, and a Michaelis-Menten model (Equation 6), which assumes saturation at V_max and an affinity constant K, to individual parameters (P) as a function of external [¹⁴C]-BaP dose. The Bayesian information criterion (BIC) was used to judge the best-fit model and provide evidence of linear or saturable pharmacokinetics (Schwarz, 1978).

$$P=k\times D$$ (5)

$$P=\frac{Vmax\times \text{D}}{K+\text{D}}$$ (6)

Software used to analyze data was “R: A language and environment for statistical computing” Version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria).

3. Results and discussion

3.1. Background levels of PAHs in plasma

Blood samples were drawn prior to each [¹⁴C]-BaP dose to assess background levels of 62 PAHs and alkylated PAHs commonly found in environmental mixtures including food (Anderson et al., 2015). Table 2 shows plasma levels of the 17 PAHs detected above the LOD in all seven participants. Variation was high, with no statistically significant differentiation of overall PAH concentration between participants. No PAHs with assigned RPFs for cancer were detected. Naphthalene and retene were the only PAHs present in all samples (7 subjects at 4 doses, n = 28) and below LOD in all quality control samples.

Table 2.

Plasma concentrations of background PAHs from all participants at the beginning of each cycle (n = 28). Plasma samples were collected between 9:00–10:00 AM following an overnight fast. The GC-MS/MS method employed quantifies 62 parent and alkylated PAHs; 49/62 PAHs were below the LOD (each similar to the ranges shown for detected PAHs). Non-detect compounds and those below the LOD were omitted. Full list of compounds, CAS numbers, and LODs provided in Table S3.

Compound	Range (ng/mL)	Mean (ng/mL)	Number of samples above LOD (n=28)	LOD (ng/mL)	LOQ (ng/mL)
1-methylnaphthalene	0.07 – 0.29	0.13 ± 0.07	8 (28.57%)	0.06	0.3
1-methylpyrene	0.32 – 4.73	0.79 ± 0.97	19 (67.86%)	0.08	0.4
2,6-dimethylnaphthalene	1.00 – 4.01	2.17 ± 1.01	11 (37.29%)	0.2	0.9
2-methylanthracene	1.58	/	1 (3.57%)	0.09	0.5
2-methylnaphthalene	0.16 – 0.99	0.41 ± 0.32	10 (35.71%)	0.1	0.7
2-methylphenanthrene	4.95	/	1 (3.57%)	0.08	0.4
6-methylchrysene	1.26	/	1 (3.57%)	0.2	0.9
7,12-dimethylbenz[a]anthracene	4.00	/	1 (3.57%)	0.2	0.9
9-methylanthracene	0.64	/	1 (3.57%)	0.2	0.9
benzo[c]fluorene	0.53	/	1 (3.57%)	0.06	0.3
benzo[k]fluoranthene	1.54	/	1 (3.57%)	0.1	0.5
fluoranthene	0.9	/	1 (3.57%)	0.1	0.5
fluorene	0.13 – 1.36	0.48 ± 0.32	12 (42.86%)	0.2	0.8
naphthalene	0.16 – 2.29	0.84 ± 0.55	28 (100%)	0.2	1.0
phenanthrene	0.64	0.47 ± 0.16	5 (17.86%)	0.09	0.5
pyrene	0.13 – 6.8	1.94 ± 2.35	7 (25%)	0.08	0.4
perylene	23.9 – 52.2	35.43 ± 12.23	4 (14.29%)	1.0	5.0
retene	1.01 – 57.8	8.81 ± 10.66	28 (100%)	0.2	0.8

PAHs in multiple plasma samples exhibited wide concentration ranges. For example, 1-methylpyrene ranged from 0.32 – 4.73 ng/mL (n = 19; 0.79 ± 0.97 ng/mL). Fluorene, perylene, alkyl-naphthalenes, phenanthrenes, and pyrenes were also present in a number of samples. In vitro studies and animal models have shown that previous exposure and/or co-exposure to other PAHs can alter the metabolism and excretion of BaP (Courter et al., 2007; Jarvis et al., 2014). Assessment of how participants were potentially impacted by previous or co-exposure of PAHs is out of the scope of this study and limited by size of the study pool. No BaP was detected in any background plasma samples (LOD 0.24 ng/mL; n = 28). For comparison, in study subjects dosed with an environmentally relevant dose of 250 ng [¹⁴C]-BaP, the C_max averaged 8.99 ± 7.08 fg/mL (range, 2.16–23.29 fg/mL), or 36 ± 28 fM, 0.0009–0.01% of the LOD for background BaP as assessed by GC-MS/MS.

None of the known or probable carcinogenic PAHs, including BaP were detected perhaps due, in part, to our exclusion criteria and the dietary PAH restrictions required. For comparison, a study by Pleil et al. (2010) measured levels of 22 PAHs in two separate healthy populations. These study groups included non-smoking university students and non-smoking adults from the general population with no recent occupational or other known PAH exposures. The median (and range) of BaP was 0.004 (≤LOD-0.017) and 0.019 (≤LOD-0.195) ng/mL in plasma from the students and adult general population, respectively (Pleil et al., 2010).

3.2. [¹⁴C]-BaP pharmacokinetics

Absorption of BaP was rapid (Figure 2), with a median T_max of 0.5 hours following each dose (Table 3). The BaP C_max (Table 3) ranged from 2.51 ± 2.53 to 13.8 ± 9.52 fg BaP/mL) over the dose range of 25–250 ng. Interestingly, the largest C_max and AUC_0–48h (with AUC_0–48h ranging from 12.2 ± 14.5 to 88.5 ± 14.1 fg BaP/mL x h, respectively in the non-compartmental model) is observed at the 100 ng dose and not the largest dose of 250 ng [¹⁴C]-BaP. In a preliminary study (2 subjects) employing UPLC-AMS, following a 46 ng dose, the T_max and C_max for [¹⁴C]-BaP was 0.75–1.0 h and 2.2–4.1 fg/mL, respectively. The closest comparable dose in the present study was 50 ng which gave a T_max (0.5–1.0 h) and C_max of (5.7 ± 4.7 fg/mL) which is in good agreement. A notable finding from the Madeen et al. (2019) study was, although as in this study a wide inter-individual variability between subjects was apparent, the intra-individual variability was low (even with over a year between doses). The large inter-individual variability is readily apparent at the two highest doses of 100 and 250 ng [¹⁴C]-BaP (Figure 3) where an order of magnitude difference between the highest and lowest C_max and AUC_0–48hr is apparent. As in previous AMS studies with both [¹⁴C]-BaP and [¹⁴C]-DBC, there was no obvious correlation to sex, age, or BMI, plus the sample size is too small for robust analysis to make strong conclusions (Hummel et al., 2018; Madeen et al., 2015, 2016, 2019). Comparison of the AUC (either AUC_{0–48 hr} or AUC_0−∞) values for [¹⁴C]-BaP (Table 3) to previous results for [¹⁴C]-BaP_eq is more tenuous given that the latter encapsulates all metabolites and potential unknown metabolites in addition to parent [¹⁴C]-BaP.

Figure 2.

Plasma levels of [¹⁴C]-BaP (fg/mL) as a function of dose in each participant as determined by UPLC-AMS over 0–48 hours. [¹⁴C]-BaP (fg/mL plasma) as a function of dose (25, 50, 100, and 250 ng) in each participant (023, 021, 024, 028, 026, 022, and 027; left to right, top to bottom) at each time point (0.25, 0.5, 1, 1.5, 2, 3, 4, 8, 24, and 48h).

Table 3.

Summary of [¹⁴C]-BaP pharmacokinetics as a function of dose, estimated using a non-compartmental model. The results are given as mean ± standard deviation except T_max, which is the median T_max for the dose and includes all seven participants and all quantifiable [¹⁴C]-BaP data points (n=170). The number listed under each parameter is the number of [¹⁴C]-BaP data curves used to estimate the parameter across all participants (potential n=7, one data curve for each participant). Curves were excluded due to an inadequate number of data points (less than 3) for a dose/individual.

	Dose (ng)
Parameter	25	50	100	250
Cmax (fg BaP/mL)	2.51 ± 2.53 n = 5	5.68 ± 4.70 n = 7	13.8 ± 9.52 n = 7	8.99 ± 7.08 n = 7
Tmax (h) (Median) (Range)	0.5 0.5 – 4.0	0.5 0.5 – 1.0	0.5 0.5 – 1.0	0.5 0.5 – 3.0
AUC (fg BaP/mL×h)	12.2 ± 14.5 n= 2	19.6 ± 15.7 n = 5	88.5 ± 14.1 n = 6	68.6 ± 64.0 n = 7
AUC0-∞ (fg BaP/mL×h)	30.8 ± NA n = 1	29.1 ± 22.5 n = 5	151 ± 54.0 n = 5	6,600 ± 1,260 n = 5
Half-life (h)	4.31 ± NA n = 1	9.33 ± 12.7 n = 5	42.4 ± 21.0 n = 5	130 ± 243 n = 5
Clearance (L/h)	811 ± NA n = 1	2,970 ± 2,270 n = 5	725 ± 230 n = 5	4,810 ± 6,950 n = 5
Vol. of Distribution (L)	5,040 ± NA n = 1	20,000 ± 14,800 n = 5	39,200 ± 6,810 n = 5	77,000 ± 51,000 n = 5

Figure 3.

Individual variability in plasma levels of [¹⁴C]-BaP from 0.25–48 hours at doses of 100 (top) and 250 ng (bottom) as determined by UPLC-AMS. Inset figures at each dose illustrate the variability of the date with log concentration of ¹⁴C-BaP over time (3–48 hours). Time point 0 h had no [¹⁴C]-BaP in any participant at any dose. Legends are consistent, with each participant represented by the same color and shape in both figures.

The best fit modeling approach identified non-compartmental apparent volume of distribution of BaP as significantly increasing as a function of dose (p=0.004, Table S5). Mean non-compartmental apparent volume of distributions increased 15× from in 25 to 250 ng BaP dose groups. Although not statistically significant (p=0.051), compartmental model apparent volume of distributions also demonstrated an increasing trend with dose (Table S4), where apparent volume of distributions increased 4.6× in 25 to 250 ng BaP dose groups. Since volunteers were the same in each dose group, and volunteer body weights were not significantly altered over the duration of the experiment (6.7 kg maximum weight change, 2.31 ± 2.28 kg), this observation suggests the fraction of BaP absorbed decreased as a function of dose. BaP likely existed as a solid in the administered capsule also containing lactose. As such, the lipophilic BaP would need to be solubilized to absorb, and larger masses of BaP may have required more time for complete dissolution and absorption if the particle surface area did not increase in the same proportion as the BaP mass. Large apparent volumes of distribution (thousands of liters, Tables 3 and 4) calculated in both approaches indicated that BaP was not well absorbed. Another possible explanation for dose-dependent increases in apparent volume of distribution is saturating a plasma protein binding site for BaP, allowing for greater BaP distribution into tissues at higher doses and higher apparent volume of distribution. Due to the low measured concentrations of BaP in blood (fg/mL), this biding site would require high specificity for BaP (Kd <100 fM) and have relatively low abundance in plasma (~250 fmol total binding sites). Due to these unlikely theoretical constraints, we do not think saturation of plasma protein binding sites is plausible. The absorption rate (k_a) of [¹⁴C]-BaP was slightly greater than in two different individuals examined in a previous study (Madeen et al., 2019), which may have limited precision due to the few number of data points defining the constant in this study. Future analyses of urinary BaP metabolites will provide additional evidence on extent of BaP oral absorption and implications of dose on fraction absorbed.

Table 4.

Summary of [¹⁴C]-BaP pharmacokinetics as a function of dose, estimated using a two-compartmental model. Includes all seven participants and all quantifiable [¹⁴C]-BaP data points (n=170). The number listed under each parameter is the number of [¹⁴C]-BaP data curves used to estimate the parameter across all participants (n=5, 6, or 7); curves were excluded due to an inadequate number of data points for a dose/individual. The results are given as the mean ± standard deviation except 25 ng dose (n = 1).

	Dose (ng)
Parameter	25	50	100	250
ka (1/h)	592 ± NA	1.30 × 107 ± 2.90 × 107 n = 5	3.52 ± 4.75 n = 6	43.2 ± 111 n = 7
k12 (1/h)	0.331 ± NA	0.83 ± 1.11 n = 5	2.60 ± 3.45 n = 6	4.27 ± 6.92 n = 7
k21 (1/h)	0.028 ± NA	0.270 ± 0.253 n = 5	0.190 ± 0.180 n = 6	0.290 ± 0.564 n = 7
k1e (1/h)	0.0068 ± NA	0.694 ± 0.412 n = 5	0.558 ± 0.897 n = 6	1.25 ± 1.71 n = 7
Vol. of Distribution (L)	2944 ± NA	4423 ± 2699 n = 4	2419 ± 2498 n = 5	10732 ± 7844 n = 6
α (1/h)	0.366 ± NA	1.69 ± 0.89 n = 5	3.33 ± 4.28 n = 6	5.80 ± 6.64 n = 7
α half-life (h)	1.892 ± NA	0.484 ± 0.187 n = 5	0.400 ± 0.212 n = 6	0.361 ± 0.407 n = 7
β (1/h)	5.03 × 10−4 ± NA	0.105 ± 0.164 n = 5	0.020 ± 0.021 n = 6	0.014 ± 0.019 n = 7
β half-life (h)	1307 ± NA	37.86 ± 43.24 n = 5	84.1 ± 88.0 n = 6	12070 ± 24100 n = 7

All other non-compartmental and compartmental pharmacokinetic parameters of BaP concentrations in plasma did not change significantly as a function of administered BaP dose (p≥0.10 Tables S4 and S5). This observation is counter to an expected increasing AUC and C_max with dose. In contrast, for most of the individual BaP metabolites (Table S6) and a sum of all BaP metabolites (Table S7, a significant linear increase as a function of dose was observed for AUC and C_max. This suggests that increasing BaP metabolism as a function of dose may play a role in the observed trends. Non-significant changes in BaP AUC and C_max may be obscured by high variability of BaP concentration in blood data (especially at the high BaP dose (250 ng)), due to potentially prolonged and highly variable fraction of BaP absorbed (perhaps exacerbated at higher doses), and hypothesized dose-dependent decrease in fraction of BaP absorbed.

3.3. [¹⁴C]-BaP metabolite profiles

Plasma [¹⁴C]-BaP metabolites were measured even at the earliest time point assessed (0.25 hours). A representative plasma metabolite profile from a single individual over 0–48 hours following administration of the 100 ng dose is shown in Figure 4. Plasma [¹⁴C]-BaP metabolites were first detected at 0.5 hours. The greatest yield of metabolites were tetrols (specific stereoisomers unknown) and dihydrodiols (7,8- and 9,10- with perhaps some 4,5-dihydrodiol). The small amount of 3-hydroxy-[¹⁴C]-BaP (in addition to the 4,5-dihydrodiol which is possibly unknown peak C) suggest that metabolism at the bay region predominated. Formation of the mutagenic and carcinogenic bay region 7,8-dihydrodiol-9,10-epoxide and subsequent covalent binding to DNA is a biomarker of risk. In our previous study (46 ng [¹⁴C]-BaP dose) we found 2/4 subjects had detectable (6–7 adducts per 10¹¹ bp, LOD of 0.5 per 10¹¹ bp) DNA adduction in PBMCs at 2–3 hours but little or no adduction at 48–72 hours (Madeen et al., 2019). For comparison, BaP-DNA adducts in occupationally exposed populations are typically 1–10 per 10⁸ bp (Pavanello et al., 2005). Tetrols, the result of hydrolysis of BaP-7,8-DHD-9,10-E have been used as a urinary biomarker of BaP cancer risk (Zhong et al., 2011). The 1,6-dione was present along with 2 unknown peaks that eluted in the region of the chromatogram corresponding to the 3,6-, 6,12- and possibly the 7,8-dione. [¹⁴C]-BaP and metabolites were markedly reduced in plasma 24–48 hours after dosing (Figures 3 and 4). Inter-individual variation in metabolism is shown in Figure S1. If glucuronide and sulfate conjugation were a significant fraction of metabolites, one could expect enterohepatic recirculation following bacterial β-glucuronidase and/or sulfatase cleavage and reuptake of the phenol, dihydrodiol, or tetrol into plasma. The only metabolite exhibiting what could be interpreted as reuptake is unknown D. At these very low concentrations there may not be active deconjugation in intestine. Individual pharmacokinetic constants for [¹⁴C]-BaP metabolites are given in Table 5.

Figure 4.

Plasma levels of [¹⁴C]-BaP and [¹⁴C]-BaP metabolites from a single representative participant at 0.5–48 hours following a dose of 100 ng [¹⁴C]-BaP. No parent compounds were quantified prior to 0.5 h in plasma extractions from this participant.

Table 5.

Summary of pharmacokinetic parameter estimates of [¹⁴C]-BaP metabolites in volunteers, estimated using a non-compartmental model. Results are listed as mean ± standard deviation except T_max, which is the median T_max for the dose/compound and includes all seven participants and all quantifiable data points.

Compound	Dose (ng)	Cmax (fg BaP/mL)	Tmax (h)	AUC (fg BaP/mL×h)	AUC0-∞ (fg BaP/mL×h)	Half-life (h)
Unknown A	25	5.06 ± 3.24	1.5	29.8 ± 20.8	45.9 ± 30.8	9.79 ± 5.23
	50	11.3 ± 9.47	1.5	73.1 ± 63.9	187 ± 267	69.3 ± 114
	100	38.7 ± 16.2	1.0	352 ± 201	514 ± 296	28.1 ± 23.0
	250	54.9 ± 36.1	1.5	522 ± 477	653 ± 614	16.7 ± 7.71
Tetrols	25	3.14 ± 0.84	2.0	43.1 ± 39.2	85.0 ± 53.8	31.4 ± 26.4
	50	7.93 ± 5.23	1.5	64.5 ± 54.6	110.0 ± 83.2	25.7 ± 22.4
	100	32.2 ± 20.6	1.5	416 ± 316	676 ± 429	45.3 ± 54.2
	250	33.9 ± 19.4	1.5	475 ± 432	672 ± 598	24.4 ± 22.0
9,10-dihydrodiol	25	2.33 ± 0.90	2.0	18.6 ± 14.6	25.9 ± 25.3	19.1 ± 14.5
	50	6.88 ± 4.01	1.5	47.7 ± 33.9	105 ± 58.2	43.2 ± 38.3
	100	21.7 ± 14.5	1.0	249 ± 175	2040 ± 4130	336 ± 724
	250	24.6 ± 11.9	1.0	298 ± 173	505 ± 474	33.3 ± 40.2
Unknown B	25	2.12 ± 0.42	1.5	5.93 ± 2.60	12.9 ± 7.86	5.22 ± 2.79
	50	7.39 ± 7.02	1.5	26.7 ± 29.1	107 ± 221	27.9 ± 57.7
	100	21.4 ± 16.1	1.0	156 ± 124	185 ± 129	19.4 ± 9.58
	250	24.8 ± 14.7	1.0	176 ± 141	215 ± 169	14.2 ± 8.23
7,8-dihydrodiol	25	1.24 ± 1.24	1.5	3.90 ± 2.11	13.8 ± 11.0	8.13 ± 5.96
	50	3.25 ± 3.25	1.5	24.7 ± 23.2	24.1 ± 18.8	9.08 ± 8.41
	100	13.9 ± 13.9	1.0	127 ± 125	184 ± 142	25.1 ± 17.9
	250	17.2 ± 17.2	1.5	155 ± 155	186 ± 176	13.3 ± 6.93
Unknown C	25	1.40 ± 0.34	3.0	17.6 ± 19.8	31.9 ± 30.7	14.3 ± 11.1
	50	3.29 ± 2.21	2.0	44.4 ± 36.4	80.8 ± 77.3	27.5 ± 23.3
	100	12.8 ± 10.5	1.5	250.0 ± 241	401 ± 326	35.7 ± 27.4
	250	16.3 ± 10.7	1.5	264 ± 261	500.4 ± 639	27.0 ± 33.1
Unknown D	25	2.63 ± 1.15	6.0	56.5 ± 41.0	97.1 ± 50.9	22.8 ± 9.01
	50	5.75 ± 3.33	3.0	138 ± 88.8	209 ± 107	34.7 ± 26.5
	100	18.08 ± 10.0	4.0	451 ± 269	136 ± 148	60.4 ± 63.8
	250	16.6 ± 10.6	2.0	389 ± 337	426 ± 390	25.4 ± 11.6
1,6-dione	25	1.17 ± 0.24	4.0	14.3 ± 16.0	21.4 ± 13.4	14.4 ± 6.69
	50	2.63 ± 1.96	3.0	44.7 ± 38.5	124 ± 45.7	69.2 ± 57.1
	100	6.56 ± 3.06	1.5	90.6 ± 46.8	208 ± 172	57.4 ± 61.0
	250	7.88 ± 4.63	1.0	107 ± 27.0	363 ± 459	108 ± 188
3,6-dione	25	NA ± NA	NA	NA ± NA	NA ± NA	NA ± NA
	50	0.85 ± 0.24	1.5	10.3 ± 12.7	350 ± 485	315 ± 439
	100	1.62 ± 0.694	1.5	17.09 ± 23.8	12.0 ± 7.65	4.79 ± 1.74
	250	2.22 ± 1.09	1.0	9.66 ± 6.54	30.6 ± 44.6	12.6 ± 20.4
9-OH	25	0.66 ± 0.06	0.5	NA ± NA	NA ± NA	NA ± NA
	50	1.90 ± 2.28	1.0	9.53 ± NA	10.79 ± NA	1.19 ± NA
	100	8.13 ± 6.96	1.0	21.9 ± 17.9	104 ± 213	67.6 ± 167
	250	6.26 ± 3.64	0.75	32.3 ± 18.9	65.05 ± 64.9	30.3 ± 49.2
3-OH	25	1.47 ± 0.74	1.0	1.88 ± 0.47	21.04 ± 27.9	14.02 ± 20.5
	50	3.12 ± 2.83	1.25	8.50 ± 6.07	11.0 ± 6.38	2.28 ± 0.51
	100	9.48 ± 6.63	1.0	20.3 ± 18.4	30.7 ± 19.6	8.030 ± 11.2
	250	9.34 ± 6.91	1.5	35.7 ± 28.9	63.09 ± 62.0	17.02 ± 25.8

A comparison of C_max and AUC_{0–48 hr} for the sum of all plasma metabolites relative to parent BaP plus metabolites as a function of dose is shown in Figure 5. At the highest dose of 250 ng parent BaP makes up 2.7% of the total [¹⁴C] (AUC_{0–48 hr}) in plasma. The significant linear inverse relationship of C_max with dose raises the possibility of predicting plasma levels across a wide environmental exposure range which includes daily estimates of exposure for non-smokers in the U.S. This would certainly be of value in BaP risk assessment. From the C_max data, the x-intercept (100% metabolites) for dose of BaP administered would be 380 ng with a 95% confidence level of 258–1019 ng which is in the range of average daily dietary BaP exposure in the U.S. (U.S. EPA, 2017). There is, of course, the caveat that our results are from a single bolus dose in fasted individuals which does not represent daily dietary exposure. Our previous study with co-administration of food (smoked salmon) found marked differences in pharmacokinetics between fasted and fed individuals (Hummel et al., 2018). In addition, we speculate that the majority of metabolism is occurring at the bay region of BaP (tetrols from hydrolysis of BaP-7,8-DHD-9,10-E, 7,8- and 9,10-DHDs and the 9-phenol). However, until the identity of the conjugates (unknown A) and unknowns B, C, and D is established the ratio of bioactivation to detoxication is uncertain. We are currently attempting identification of these unknown metabolites in plasma and in urine. This extensive metabolism of BaP at ng levels, seen in this study following oral ingestion, may not apply for exposures through inhalation or dermal contact characteristic of smokers or individuals with high occupational exposures.

Figure 5.

Pharmacokinetic constants C_max and AUC_0–48hr for [¹⁴C]-BaP in plasma as a percentage of C_max and AUC_0–48hr for total [¹⁴C] (BaP plus metabolites) as a function of dose. The percent of [¹⁴C]-BaP in plasma decreased with increasing dose for both C_max (r²=0.950, p=0.025) and AUC_0–48hr (r²=0.691, p=0.169) although the latter was not significant.

3.4. Limitations

There are limitations associated with this study, including the unknown metabolites that cannot be further identified due to a lack of standards. The small sample size limited assessment of potential confounders and sources of the large inter-individual variability. Further, the inclusion and exclusion criteria for recruitment and enrollment of so few participants creates a group that may not be representative of the wider population. Dietary questionnaires collected for the three days immediately preceding dosing through the final blood draw may be less useful than quantitative data for exposure and consumption assessments; however, given the limited sample pool any potential variations introduced by diet would not be statistically significant. Third, variability in background levels of PAHs, including BaP, at levels lower than the GC-MS/MS LOD, could impact pharmacokinetics. Again, potential effects are beyond the scope of this study and limited by sample size. Fourth, participants received a low oral bolus dose of a single PAH, which is not how humans are typically exposed. People are continuously exposed to PAH mixtures at various concentrations primarily (≥ 95%) through diet.

3.5. Conclusions

The IARC class 1 known human environmental carcinogen, BaP, when given as a microdose to humans, exhibited rapid uptake (T_max, 0.5–1 hours), extensive metabolism, and rapid elimination (average half-life across doses, 46.5 ± 58.2 hours). Parent [¹⁴C]-BaP was a minor component (based on C_max and AUC) in plasma relative to total metabolites and declined with dose. The pharmacokinetic parameters for most of the metabolites exhibited a linear dose-response relationship. As these doses (25–250 ng) cover the range of likely human dietary exposure, our results should be of use in risk assessment not just for BaP but for carcinogenic PAH mixtures as levels of BaP in environmental samples are predictive of concentrations of other carcinogenic PAHs (Schneider et al., 2002). Our finding that most metabolism is likely occurring at the bay region also has implications for risk assessment. As in previous studies employing AMS and [¹⁴C]-PAHs, there is a large inter-individual variability in [¹⁴C]-BaP pharmacokinetics. The interface of AMS to UPLC via the moving wire (Thomas et al., 2011) allows for tentative identification of [¹⁴C]-BaP metabolites based on coelution with known standards. This is a unique dataset on the uptake, metabolism and disappearance from plasma of an important environmental carcinogen. The extensive metabolism of [¹⁴C]-BaP, as reflected in the time course of plasma levels, is presumably a consequence of robust “first-pass” metabolism by the liver (with perhaps a significant contribution from intestine) and should be incorporated into revisions of current PBPK models. The high inter-individual variability in pharmacokinetics may be due in part to a wide range in expression of CYPs, GSTs, UGTs, SULTs, and transporters (and allelic variants) important in the ADME of BaP and similar environmental carcinogens.

Highlights.

• Human dose-response study of carcinogen BaP at environmental levels with UPLC-AMS.

• UPLC-AMS of [¹⁴C]-BaP and metabolites over 4 doses in human plasma was determined.

• Inter-individual differences in pharmacokinetics with 25–250 ng doses were observed.

• UPLC-AMS analysis showed extensive metabolism predominantly at the BaP bay region.

• This metabolic profile at relevant exposures should impact cancer risk assessment.

Abbreviations

AFB₁

aflatoxin B₁

AMS

accelerator mass spectrometry

AHR

aryl hydrocarbon receptor

AUC

area under the curve

ATSDR

Agency for Toxic Substances and Disease Registry

BaP

benzo[a]pyrene

BaP-DHD

(± trans-benzo[a]pyrene-7,8-dihydrodiol

BaP-DHDE

BaP-DHD-9,10-epoxide

BMI

body mass index

C_max

maximum concentration in plasma

CYP

cytochrome P-450

DBC

dibenzo[def,p]chrysene

FDA-IND

U.S. Food and Drug Administration Investigative New Drug

IARC

International Agency for Research on Cancer

IPCS

International Programme on Chemical Safety

IQ

2-amino-3-methylimidazo[4,5-f]quinolone

IRB

Institutional Review Board

LLNL

Lawrence Livermore National Laboratory

LOD

limit of detection

LOQ

limit of quantitation

MeIQx

2-amino-3,8-dimethylimidazo[4,5-b]quinoxaline

NPL

National Priorities List

N²-dG

nitrogen at the 2 position of deoxyguanosine

OSU

Oregon State University

PAH

polycyclic aromatic hydrocarbon

PBMCs

peripheral blood mononucleated cell

PBPK

physiologically based pharmacokinetics

PhIP

2-amino-3-methyl-6-phenylimidazo[4,5-b]pyridine

RPF

Relative Potency Factor

T_max

time at C_max

UPLC

ultra-pressure liquid chromatography (Waters)

U.S. EPA

United States Environmental Protection Agency