Initial analyses of the relationship between “Thresholds” of toxicity for individual chemicals and “Interaction Thresholds” for chemical mixtures

https://doi.org/10.1016/j.taap.2006.11.016Get rights and content

Abstract

The inter-relationship of “Thresholds” between chemical mixtures and their respective component single chemicals was studied using three sets of data and two types of analyses. Two in vitro data sets involve cytotoxicity in human keratinocytes from treatment of metals and a metal mixture [Bae, D.S., Gennings, C., Carter, Jr., W.H., Yang, R.S.H., Campain, J.A., 2001. Toxicological interactions among arsenic, cadmium, chromium, and lead in human keratinocytes. Toxicol. Sci. 63, 132–142; Gennings, C., Carter, Jr., W.H., Campain, J.A., Bae, D.S., Yang, R.S.H., 2002. Statistical analysis of interactive cytotoxicity in human epidermal keratinocytes following exposure to a mixture of four metals. J. Agric. Biol. Environ. Stat. 7, 58–73], and induction of estrogen receptor alpha (ER-α) reporter gene in MCF-7 human breast cancer cells by estrogenic xenobiotics [Gennings, C., Carter, Jr., W.H., Carney, E.W., Charles, G.D., Gollapudi, B.B., Carchman, R.A., 2004. A novel flexible approach for evaluating fixed ratio mixtures of full and partial agonists. Toxicol. Sci. 80, 134–150]. The third data set came from PBPK modeling of gasoline and its components in the human. For in vitro cellular responses, we employed Benchmark Dose Software (BMDS) to obtain BMD01, BMD05, and BMD10. We then plotted these BMDs against exposure concentrations for the chemical mixture and its components to assess the ranges and slopes of these BMD-concentration lines. In doing so, we consider certain BMDs to be “Interaction Thresholds” or “Thresholds” for mixtures and their component single chemicals and the slope of the line must be a reflection of the potency of the biological effects. For in vivo PBPK modeling, we used 0.1× TLVs, TLVs, and 10× TLVs for gasoline and six component markers as input dosing for PBPK modeling. In this case, the venous blood levels under the hypothetical exposure conditions become our designated “Interaction Thresholds” or “Thresholds” for gasoline and its component single chemicals. Our analyses revealed that the mixture “Interaction Thresholds” appear to stay within the bounds of the “Thresholds” of its respective component single chemicals. Although such a trend appears to be emerging, nevertheless, it should be emphasized that our analyses are based on limited data sets and further analyses on data sets, preferably the more comprehensive experimental data sets, are needed before a definitive conclusion can be drawn.

Introduction

In February 2005, the Society of Toxicology (SOT), jointly with the Agency for Toxic Substances and Disease Registry (ATSDR) and other sponsors, held a workshop in Atlanta, Georgia. The central theme of the workshop was Contemporary Concepts in Toxicology: Charting the Future: Building the Scientific Foundation for Mixtures Joint Toxicity and Risk Assessment. This workshop was a follow-up based upon the recommendations of a 2002 SOT Expert Panel that identified specific biologically based hypotheses and experimental approaches to generate data useful to enhance assessments and develop realistic policies for chemical mixtures. One of the nine hypotheses generated by the Expert Panel in the 2002 meeting was “Apparent dose thresholds for interactions are higher than individual chemical thresholds.” The present paper, which was part of the program on “Dose to Response” in the abovementioned workshop, was assigned to the first author of this paper to specifically address this hypothesis.

The idea of “Interaction Thresholds” was introduced in 1996 by El-Masri et al. (1996) as the minimal level of change in tissue dosimetry associated with a significant health effect. That research work was based on PBPK modeling of the competitive inhibition of trichloroethylene metabolism by 1,1-dichloroethylene. To estimate the “Interaction Threshold,” the model of El-Masri et al. (1996) included a description of the percentage of enzyme (i.e., CYP2E1) sites occupied by either chemical in the presence or absence of the other.

When two or more interactive chemicals are studied together, theoretically, there could be infinite interaction thresholds depending on the dose levels used for the individual chemicals in the studies. However, if we specify certain occupational or environmental exposure concentrations for all the other component chemicals in the mixture except one, we may obtain an interaction threshold for that set of specific exposure conditions. Even though the concept sounds quite straightforward, there are very few such dose–response data sets for chemical mixtures. It is therefore even more of a challenge to attempt to analyze and compare interaction thresholds of chemical mixtures with their respective dose thresholds of individual component chemicals. In addition to the very limited availability of experimental data sets for such analyses and comparisons, two other issues arose in carrying out this assignment from the SOT. First, how do we define “Interaction Threshold”? Do we simply follow our earlier definition by El-Masri et al. (1996)? Given some of the data sets to be utilized are from cell culture systems and PBPK modeling was not carried out in these studies, the dose metrics are different. What do we do in this instance? The second issue relates to “interaction.” How do we know that toxiõcological interactions indeed happened in the mixtures that we selected, particularly at the low dose region? These issues are discussed, respectively, in the Methods and Discussion sections.

To explore this relatively new area of inter-relationship of “Thresholds” between chemical mixtures and their respective component single chemicals, we used three different sets of data and two different types of data analyses. Two of the data sets were from our own laboratory: one on cytotoxicity in normal human epidermal keratinocytes (NHEK) from treatment of metals and a metal mixture (Bae et al., 2001, Gennings et al., 2002), the other on venous blood levels from physiologically based pharmacokinetic (PBPK) modeling of gasoline and its components in the human based on a human PBPK model extrapolated from an earlier published rat PBPK model (Dennison et al., 2003, Dennison et al., 2004). The other data set came from Gennings et al. (2004) on induction of estrogen receptor alpha (ER-α) reporter gene in MCF-7 human breast cancer cells by estrogenic xenobiotics and mixture. We employed Benchmark Dose Software (BMDS) (USEPA, NCEA; http://cfpub1.epa.gov/ncea/cfm/bnchmrk/versions.cfm?ActType=default) and PBPK modeling to assess the ranges and slopes related to “Interaction Thresholds” or “Thresholds” for mixtures and their component single chemicals. The experimental details, as well as our thoughts on the significance of the findings are given below.

Section snippets

Origins and contents of the data sets

The first data set was from the normal human epidermal keratinocytes (NHEK) studies reported by Bae et al. (2001) and Gennings et al. (2002). The cells were treated with As, Cr, Cd, Pb, or a mixture of the four metals for a 24-h period, re-fed with fresh metal-free medium for 3 days prior to viability (cytotoxicity) analysis by the MTT assay (Mossman, 1983). At least three replicate experiments were conducted and five to seven dose levels covering a concentration range of about 100- to

NHEK cytotoxicity from individual metals and a metal mixture

Of the numerous BMDS dose–response plots, we chose to present three plots (Fig. 1, Fig. 2, Fig. 3) from the metal mixture experiments as an illustration. Fig. 1 shows the BMDS plot mean response of NHEK cytotoxicity as a result of exposure to the metal mixture. The dose–response curve is U-shaped because NHEK cytotoxic interaction from the metal mixture is dose-dependent with the highest concentrations causing antagonistic interactive cytotoxicity (Bae et al., 2001, Gennings et al., 2002). To

Discussion

This work represents a first attempt to explore the relationship between “Thresholds” of toxicity for individual chemicals and “Interaction Thresholds” for chemical mixtures and the SOT Expert Panel provided the impetus for this endeavor. The approach we took is different from the treatment of the subject related to NOAELs/LOAELs of mixtures and corresponding single chemicals (Groten et al., 1997). It is interesting to note that despite the three entirely different types of studies (i.e.,

Acknowledgments

We thank ATSDR (Cooperative Agreement U61/ATU881475), NIEHS (Superfund Basic Research Program Project P42 ES05949; research grants R01 ES09655 and RO3 ES10116 ZES1; training grant T32 ES 07321), US Air Force (research grants F33615-91-C-0538; F49620-94-1-0304), and NIOSH/CDC (1 RO1 OH07556-01). Without the generous support of these agencies, the development of research described herein could have never been possible. The experimental data from NHEK cytotoxicity were generated by Dr. Dong-soon

References (10)

  • J.P. Groten et al.

    Subacute toxicity of a mixture of nine chemicals in rats: detecting interactive effects with a fractionated two-level factorial design

    Fundam. Appl. Toxicol.

    (1997)
  • D.S. Bae et al.

    Toxicological interactions among arsenic, cadmium, chromium, and lead in human keratinocytes

    Toxicol. Sci.

    (2001)
  • J.E. Dennison et al.

    Characterization of the pharmacokinetics of gasoline using PBPK modeling with a complex mixture chemical lumping approach

    Inhalation Toxicol.

    (2003)
  • J.E. Dennison et al.

    Development of a PBPK model for volatile fractions of gasoline using chemical lumping analyses

    Environ. Sci. Technol.

    (2004)
  • I. Dobrev et al.

    In silico toxicology: simulating interaction thresholds for human exposure to mixtures of trichloroethylene, tetrachloroethylene, and 1,1,1-trichloroethane

    Environ. Health Perspect.

    (2002)
There are more references available in the full text version of this article.

Cited by (22)

  • Chemical mixtures: toxicologic interactions and risk assessment

    2020, Information Resources in Toxicology, Volume 1: Background, Resources, and Tools
  • Review and recommendations on criteria to evaluate the relevance of pesticide interaction data for ecological risk assessments

    2018, Chemosphere
    Citation Excerpt :

    The dose-dependent nature of GTA responses has been termed an “interaction threshold” (Gennings et al., 2007). An interaction threshold is defined as “the dose region where the magnitude of toxicological interactions between components of a mixture significantly deviate from additivity” and is illustrated in Fig. 3 (Yang and Dennison, 2007; El Masri, 2010). Statistical methods have been developed to address the challenge of quantifying an interaction threshold (Gennings, 2010; El Masri, 2010).

  • Effect of estrogenic binary mixtures in the yeast estrogen screen (YES)

    2014, Regulatory Toxicology and Pharmacology
    Citation Excerpt :

    The investigation of potential risk to human health as a result of exposure to EDCs and mixtures thereof is a major topic for consumer safety (Bjarnason, 2004; Rudel, 1997; Sarigiannis and Hansen, 2012; SCHER/SCENIHR/SCCS, 2011; U.S. EPA, 2000; Vandenberg et al., 2012). It has been reported that combinations of estrogenic chemicals can act together at dosage levels at which the single test compounds do not cause the observed effects (Kortenkamp, 2007; Yang and Dennison, 2007). Combined exposure to chemical mixtures with estrogenic effects have been investigated in several models (Andersen et al., 1999; Brion et al., 2012; Delclos et al., 2009; Kolle et al., 2011).

  • Multiple Chemical Exposures and Risk Assessment

    2013, Computational Toxicology: Methods and Applications for Risk Assessment
  • Cumulative risk assessment of chemical exposures in urban environments

    2011, Environment International
    Citation Excerpt :

    Although the reasoning behind the calculation procedure is sound, the underlying scientific evidence is weak and many of the values used have a weak empirical basis, such as the evidence factor (fij) and the maximum interaction effect (Mij). One of the issues that needs to be resolved is whether there is a threshold below which substances do not interact (Yang and Dennison, 2007). For example, in our case study, the hazard quotient of all pesticides is below 0.01.

View all citing articles on Scopus
View full text