Abstract
Addressing inter- and intra-species differences in potential hazardous effects of chemicals remains a long-standing challenge in human health risk assessment that is typically addressed heuristically through use of 10-fold default “uncertainty” or “safety” factors. Although it has long been recognized that chemical-specific data would be preferable to replace the “defaults,” only recently have there emerged experimental model systems and organisms with the potential to experimentally quantify the population variability in both toxicokinetics and toxicodynamics for specific chemicals. Progress is most evident in the use of population in vitro human cell-based models and population in vivo mouse models. Multiple case studies were published in the past 10–15 years that clearly demonstrate the utility of such models to derive data with direct application to quantifying variability at hazard identification, exposure-response assessment, and mechanistic understanding of toxicity steps of traditional risk assessments. Here, we review recent efforts to develop fit-for-purpose approaches utilizing these novel population-based in vitro and in vivo models in the context of risk assessment. We also describe key challenges and opportunities to broadening application of population-based experimental approaches. We conclude that population-based models are now beginning to realize their potential to address long-standing data gaps in inter- and intra-species variability.
70 Years of Using Default Factors in Risk Assessment: Quo Vadis?
One of the earliest proposals to quantitatively address both inter- and intra-species variability in potential toxicity of chemicals was that by the US FDA (Lehman, 1954) defining the basis for the so called “acceptable daily intake,” an amount of a substance thought to be without appreciable risks of adverse health effects alter chronic intake by a population of humans. Specifically, it was proposed that a “safety factor” of 100 be used to account for the uncertainties in extrapolation from dose-response toxicity data in animal studies to the whole of human population. This suggestion for a “default” factor that could be used when no chemical-specific data were available was swiftly implemented in decision-making on pesticide safety by the World Health Organization (WHO, 1965). WHO “adopted the commonly used empirical method: the maximum no-effect dietary level obtained by animal experiment […] was divided by a ‘factor’, generally 100”. The evolution of the concept of “default factors,” including sub-division of the 100-fold first into two 10-fold factors (for inter- and intra-species) and then further into four 3-fold factors (for toxicokinetics and toxicodynamics), as well as additions of other 10-fold components that were meant to account for the weaknesses in the database and the types of point-of-departure (POD) values that are available, and their application in risk assessment of not only pesticides but also environmental chemicals, drugs and other agents, has been extensively reviewed elsewhere (Dourson et al., 1996; Dourson and Stara, 1983; Meek et al., 2002). While it could still be argued that such “factors” are largely arbitrary (Dourson et al., 2022), there is an appreciable amount of historical data from studies in humans and animals that suggest they are reasonable estimates scientifically, though not always conservative in the absence of chemical-specific data (Dorne et al., 2005; Renwick, 1993; U.S. Environmental Protection Agency, 2005; WHO/IPCS, 2018).
Importantly, as early as 1960s, it was already appreciated by the regulatory bodies that “genetically determined differences in the enzymatic pattern of man can affect his reaction to toxic agents” and, because there were little data at that time to quantitatively characterize such differences, the “acceptable daily intakes should not be applied too rigidly” (WHO, 1965). Moreover, this gap in suitable scientific data that could be used to increase confidence in proper accounting for population variability in quantitative risk assessment has been reiterated multiple times ever since (National Research Council, 1983; National Research Council, 2009). Perhaps this was most clearly stated by Dr. George Gray who named four principal pitfalls in risk assessment: (1) ignoring variability, (2) ignoring uncertainty, (3) favoring consistency over science, and (4) not evaluating the influence of assumptions (Institute of Medicine, 2001). All of these points either directly or indirectly relate to the common deficiencies in the understanding of the variability in responses between individuals and overreliance on default assumptions. Indeed, numerous National Academies’ reports made recommendations to formally include considerations of human variability not only in non-cancer risk assessment, but also in cancer dose-response modeling and risk characterization (National Research Council, 2009; National Research Council, 2011; National Research Council, 2014).
To quantitatively characterize the extent of intra-species variability for a particular chemical, data are needed from a population of subjects where variables other than inter-individual differences in metabolism or other biological factors are standardized or minimized. Such conditions can be best achieved in either standardized animal tests, or in human clinical trials. Indeed, early (though still post-hoc) support to the use of 10-fold factor for intra-species variability was derived, in part, from the data for acute lethality from studies in rats (Weil, 1972). However, most of the historical data from guideline animal experiments that are available in public domain come from studies in a small number of inbred or hybrid (i.e., genetically homogeneous) strains of rats or mice (Chhabra et al., 1990; Watford et al., 2019; Wignall et al., 2014); such data are not suitable for addressing inter-individual variability. Data for therapeutic drugs have been used for the analysis of human variability for both kinetic and the dynamic aspects, these studies generally suggested that the default intra-species variability factor is often inadequate, even when accounting for the differences in kinetics alone (Dorne et al., 2012; Dorne et al., 2005). Recent meta-analyses of enzyme and transporter activities showed wide ranges of chemical-specific factors, including those greater and less than the default (Buratti et al., 2021; Darney et al., 2020a; Darney et al., 2020b; Di Consiglio et al., 2021). Suggestions have been made that a factor of 25, and not 10, may be a more sensible starting point for estimating overall human variability in susceptibility to cancer (National Research Council, 2009).
More recently, several strategies have been proposed to improve confidence in estimations of inter-individual human variability and increase reliance on scientific data rather than use default factors. These approaches result from rapid advances in our understanding of genetic variability within and across species, most notably humans and rodents, and the spread of the -omics research tools that enable improved characterization of the mechanisms of toxicity. The concepts of pathway-related uncertainty factors (Dorne et al., 2012), and of chemical-specific adjustment factors (Bhat et al., 2017; Gentry et al., 2002) both fundamentally rely on better understanding of (i) genetic variability in relevant species (e.g., humans and rodents), and (ii) molecular pathways of toxicity.
Can Population-Based Experimental Models Provide the Lever Long Enough and a Place to Stand for Moving Away from Defaults?
Sequencing of the human genome and decrease in the cost of genotyping have revolutionized the field of genetics. Most published large-scale genome-wide association studies (GWAS) of disease associations and adverse drug effects have focused on identifying susceptibility loci and devising strategies (i.e., genetic tests) to prevent drug toxicity by identifying those who are at risk of adverse drug reactions. Such studies are typically referred to as pharmacogenomics and aim to use individual patient’s genetic information to predict drug response(s) and/or guide optimal drug and dose selection to enable safer, more effective, and cost-effective treatments (Motsinger-Reif et al., 2013; Relling and Evans, 2015). Over the past two decades, such studies focused either on the discovery of novel genomic loci that may be associated with susceptibility to a particular disease (Giacomini et al., 2017), or on modifying effects of a relatively narrow compendium of “pharmaco-genes” (Bush et al., 2016). Favorable evidence of cost-saving and effectiveness of using pharmacogenomics information to improve treatment outcomes and/or prevent adverse events is accumulating rapidly (Krebs and Milani, 2019). Such studies also provide very useful evidence of gene-drug associations and penetrance of various toxicology-relevant alleles (Zhou et al., 2017). It is likely that implementation of pharmacogenomic data in clinical practice will continue to increase with wider adoption of the electronic medical record databases (Reisberg et al., 2019). Even though pharmacogenomics and traditional GWAS are used to identify polymorphisms that may confer susceptibility in human sub-populations, few successful examples of predicting phenotype from genotype, especially for highly polymorphic drug-metabolizing enzymes, have been reported (Gaedigk et al., 2017; Haufroid and Hantson, 2015). In addition, the information on susceptibility alleles, if any, cannot be used directly for the quantitation of the overall population variability in response to each drug.
The revolution in genomic sequencing technologies enabled not only GWAS and pharmacogenomic studies in humans, but also improved our understanding of the genetic makeup of experimental animals, models that are most relied upon in risk assessment of environmental pollutants, industrial chemicals and pesticides (Harrill and McAllister, 2017). It was pointed more than 40 years ago (Festing, 1979) that genetic uniformity of traditional rodent models (inbred strains and F1 hybrids) that are used in regulatory toxicology has a number of positives, such as phenotypic uniformity, high long-term genetic stability, high identifiability, and large differences between strains. However, the primary disadvantage of using the data from these “genetic clones” for risk assessment is the jeopardy of drawing conclusions from an “outlier” strain not representative of the heterogeneous population (Chiu and Rusyn, 2018).
Population-based in vivo and in vitro models
Early suggestions for reducing uncertainty in estimating population-wide dose-response in rodent studies included the use of several isogenic strains with a factorial experimental design as opposed to the reliance on the data from a single strain (Festing, 1979); however, these did not find much use in regulatory toxicology. The efforts to deep sequence a number of commonly used inbred mouse strains (Frazer et al., 2007) have given renewed impetus to the proposal to use a small battery of inbred strains instead of the single strain for toxicity screening (Festing, 2010). Furthermore, conception of a number of “designed” populations of inbred mouse lines that derive from common genetically heterogenous ancestor strains, so called Diversity Outbred (Churchill et al., 2012) and Collaborative Cross (Threadgill and Churchill, 2012) populations, have further expanded the toolbox for quantitative characterization of variability in effects of drugs and chemicals.
Another experimental approach to evaluate population variability in effects of chemicals and drugs is to use cells from multiple individuals as an in vitro population-based model. Two most commonly used strategies are the use of human lymphoblastoid cell lines (LCLs) and induced pluripotent stem cell (iPSC)-derived organotypic models. For pharmacogenomic studies, thousands of human LCLs have been sequenced or very densely genotyped (1000 Genomes Project Consortium et al., 2012) and are now a commonly used model (Wheeler and Dolan, 2012). These cells have been used to identify genetic variants associated with pharmacologic phenotypes, or to screen individual patient’s cells to guide pharmacotherapy. Many GWAS for drug-induced phenotypes have been performed in LCLs, often incorporating gene expression data (Lappalainen et al., 2013). These cells have also been used to conduct population-based in vitro hazard and concentration-response assessment of hundreds of chemicals (Abdo et al., 2015b; Eduati et al., 2015), as well as mixtures and complex substances (Abdo et al., 2015a). It was also recently found that combining these population-based in vitro data with population-based reverse toxicokinetic in vitro- to-in vivo extrapolation could yield toxicity values that correlated highly with those derived from in vivo experimental animal data (Chiu and Paoli, 2021). Human iPSC-derived population-based models are most advanced for cardiomyocytes (Burnett et al., 2021b); these cells have been used to characterize hazard, dose-response and mechanisms for hundreds of environmental chemicals and drugs (Blanchette et al., 2020; Burnett et al., 2021a; Burnett et al., 2019; Grimm et al., 2018). Moreover, as discussed further below, it has been shown that if appropriate in vitro-to-in vivo extrapolation approaches are utilized, population-based iPSC-derived cardiomyocytes can accurately predict in vivo concentration-response relationships for “long QT”, an electrophysiological syndrome associated with disruption of ion channel function in cardiomyocytes that is a risk factor for ventricular arrythmia and death (Blanchette et al., 2019). More recently, pluripotent stem cells (PSCs) were derived from different mouse strains (Ortmann et al., 2020); this study showed that lines from distinct genetic backgrounds have divergent differentiation capacity.
Obtaining information on population variability from gene by environment (G*E) studies
One potential barrier to wider adoption of these models in risk assessment contexts for estimating population variability is a lack of awareness by the academic scientists of the potential value of their data for G*E studies. If replication is used, studies of cell line responses to drugs (Haverty et al., 2016) or environmental chemicals (Abdo et al., 2015b), or studies in rodent models of exposure to chemicals (Cichocki et al., 2017) can be used to estimate toxicodynamic variability factors. Essentially, these estimates are based on estimates of intraclass correlation (ICC) (Bartko, 1966) with the individual (or strain) as the grouping class, in order to partition sampling and technical variability from true population variability (Pleil et al., 2018). The ICC multiplied by the total variability is an estimate of population variability; however, this fairly straightforward analysis is seldom applied and reported. Population variability may reflect underlying genetic sources of population variation (Byers, 2008), and researchers involved in G*E studies should be encouraged to report evidence of population variability even if no individual significant genetic region is found to be associated with an exposure.
In the absence of replication per individual, the overall estimated variability may serve as a conservative upper bound for true population variability since it includes both inter- and intra-individual variation. In addition, replication is not necessary to estimate genetic sources of variation, if high-quality genetic data are available, because the pairwise comparisons of similarity of genotypes to similarity of phenotypes can be used to estimate broad-sense heritability of response to exposure with software such as GCTA (Yang et al., 2011). These heritability estimates may be thought of as a lower bound to the intra-class correlation, as they address only the genetic source of variation across individuals.
Using Experimental Population-Based Models for Risk Assessment Applications: Matching Models to Regulatory Applications
While academic research is primarily aimed at advancing fundamental knowledge, regulatory agencies need to ensure that their statutory processes are following established guidelines while also taking advantage of new discoveries that are fit for regulatory purpose. The opportunities for using population-based experimental models have been outlined not only in a number of review articles (Chiu and Rusyn, 2018; Harrill and McAllister, 2017; Rusyn et al., 2018; Zeise et al., 2013), but also in the official position by some agencies (Cote et al., 2016). Specifically, the US EPA identified several approaches that can be used to help with better characterizations of population variability. These include the use of (1) adverse outcome pathway (AOP) networks for identifying chemicals and other environmental stressors that appear to act by the same mechanisms and could contribute to risk; (2) in vivo and in vitro test results from genetically diverse populations for capturing the range of genetically determined risk; and (3) epidemiology studies for capturing variability due to molecular biological differences in response to chemical and nonchemical stressor exposures. Most salient to this review is the second point above – what available non-human experimental models can provide data that fits the regulatory context? Here, we identify published case studies that provide practical examples of how various model systems have been already used to address various risk assessment applications (Table 1).
Table 1.
— = not appropriate model for this application
N/A = Studies have not been conducted
Exposure Assessment
While exposure scenarios vary among individuals based on their employment, place of residence and other conditions, the inter-individual variability in the internal dose is also a result of variability in toxicokinetics. Human in vitro models are widely used to inform physiologically-based pharmacokinetic modeling (PBPK), specifically to refine estimates of population variability in toxicokinetic parameters (Kostewicz et al., 2014). Cells from different individuals are used widely to estimate intrinsic clearance, transporter and metabolic activity (Pathak et al., 2017); however, these data are not typically associated with concurrent characterization of the genetic polymorphisms. Many PBPK models offer “virtual populations” (Zhang et al., 2020), albeit they are based on modeled population variability rather than on population-based in vitro data as inputs. The lack of iPSC-derived population-based panels for human or animal hepatocytes, renal proximal tubule epithelial cells, or enterocytes is a considerable gap in existing repertoire of models to obtain chemical-specific experimental data and quantify toxicokinetic variability.
A number of studies were conducted in genetically-defined mouse panels to quantify population toxicokinetic variability in internal dose of two high production volume chlorinated solvents, trichloroethylene and tetrachloroethylene. For trichloroethylene, the first study to quantify strain-specific absorption, distribution, metabolism, and excretion (ADME) after a single oral dose of trichloroethylene was conducted in a panel of 15 standard inbred mouse strains (Bradford et al., 2011). Serum levels of oxidative and glutathione conjugation metabolites of trichloroethylene, which are considered the toxic moieties, were then used in combination with data from two additional strains (17 strains in all) (Chiu et al., 2014) to calibrate and extend a PBPK model by adding one-compartment models for glutathione metabolites and a two-compartment model for dichloroacetic acid. A Bayesian population analysis of inter-strain variability to quantify variability in trichloroethylene TK showed that, when extrapolated to lower doses more relevant to environmental exposures, mouse population-derived variability estimates closely matched population variability estimates previously derived from human toxicokinetic studies with trichloroethylene. Subsequent studies of repeat exposure (for up to 4 weeks) were conducted in a smaller panel of seven inbred mouse strains to quantify TK of oxidative and glutathione conjugation metabolites of trichloroethylene not only in serum, but also in liver (Yoo et al., 2015a) and kidney (Yoo et al., 2015b). Overall, these studies demonstrated the utility of mouse inter-strain ADME studies for addressing human toxicokinetic variability for risk assessment applications, giving results consistent with human studies showing that variability was highly dependent on the metabolic pathway, with TK variability in oxidative metabolites less than the default (~2-fold or less) and variability in conjugative metabolites greater than the default (~7-fold).
For tetrachloroethylene, several studies evaluated oxidative and glutathione conjugation metabolite formation in different mouse tissues in a Collaborative Cross panel of 45 strains and in several inbred mouse strains (Cichocki et al., 2017; Luo et al., 2019; Luo et al., 2018b). In every tissue examined in the Collaborative Cross study, significant variability among strains, almost an order of magnitude, was observed in toxicokinetics of tetrachloroethylene and its oxidative and glutathione conjugation metabolites. For the parent molecule and its primary oxidative metabolite trichloroacetic acid, the observed range (max to min) of variability across ~50 strains was about 8-fold (Cichocki et al., 2017), but for glutathione conjugation metabolites it was up to 50-fold (Luo et al., 2019). In addition, Collaborative Cross population provided a better characterization for the population variability in toxicokinetics of tetrachloroethylene as compared to study of only three inbred strains (Luo et al., 2018b). The three strains exhibited oxidative metabolism of tetrachloroethylene at the lower end of the broader Collaborative Cross population, while glutathione conjugation was in the middle; these data demonstrate the validity of a concern that a single or few strains may not be representative of a larger population responses (Chiu and Rusyn, 2018). In addition, these population-based data were used to incorporate glutathione conjugation into a PBPK model for tetrachloroethylene (Dalaijamts et al., 2018) and to provide data-informed quantitative characterization of population-wide tissue and metabolite-specific variability of tetrachloroethylene toxicokinetics (Dalaijamts et al., 2020). Together, these studies found that, after accounting for technical and intra-strain variability, toxicokinetic inter-strain variability is generally less than the default factor of ~3 for oxidation and comparable to or slightly larger than the default for conjugation.
Hazard Identification
There are long-standing arguments that studies in rodents may be un-informative about chemical’s potential to be a human health hazard (Akhtar, 2015; Hartung, 2008). Others reason that studies in animals provide great value to protection of humans from chemical risks (Krewski et al., 2019; Olson et al., 2000). Both sides acknowledge that homogeneity of the animals (i.e., use of inbred or F1 hybrid strains) that are commonly used for regulatory studies is, at least partially, responsible for the perceived lack of species concordance (Festing, 1986; Festing and Wilkinson, 2007). In this regard, a number of population-based model systems may aid in addressing such challenges in hazard identification as showing that perceived lack of species concordance may have been due to the use of an “outlier” strain, and filling in the gaps where hazard testing in rodents is either non-informative (e.g., cardiac toxicity) or has not been conducted yet due to cost and other constraints. Below, we provide examples of case studies that address these challenges in hazard identification.
It is well appreciated in drug development that some laboratory animals may not be “pharmacologically relevant” to evaluate efficacy or safety of certain pharmaceuticals (Bussiere et al., 2009). There are also well-documented cases of species differences in nuclear receptor-mediated signaling that is related to induction of xenobiotic metabolism and other liver effects (Corton et al., 2014; Rusyn et al., 2006). These examples point to the conclusions that animals are not always the most appropriate or relevant models for determining human efficacy and safety of chemicals and drugs. However, there are also cases when human cancer hazard and organ-specific toxicity are not concordant with studies in rodents (e.g., cancer bioassays) because of the lack of genetic diversity in the strains that are typically used in regulatory studies (Chiu and Rusyn, 2018; Harrill and McAllister, 2017; Krewski et al., 2019). Inter-strain variability in toxicity of various environmental chemicals and drugs is well documented (Church et al., 2015; Church et al., 2014; Harrill et al., 2012; Harrill et al., 2018; Harrill et al., 2009b; Koturbash et al., 2011; Nguyen et al., 2017; Tsuchiya et al., 2012; Yoo et al., 2015b; You et al., 2020) and has been attributed to either polymorphisms in xenobiotic metabolism genes, or other toxicity-relevant pathways such as immune response and tissue repair. While most of these studies used inter-strain variability to better characterize the mechanisms of susceptibility, several studies demonstrated that the failure to recognize mouse as “relevant” species in studies of some drugs and chemicals may have been largely due to limited genetic diversity (Harrill et al., 2012; Yoo et al., 2015b).
In addition, to fill the gaps in hazard data on chemicals in commerce and the environment, in the past 15 years, thousands of chemicals have been tested in hundreds of in vitro models by large-scale government consortia (Judson et al., 2009; Paul Friedman et al., 2020; Richard et al., 2021). However, only some of these cell-based models are population-based (Abdo et al., 2015b; Burnett et al., 2021a; Burnett et al., 2019). For example, the LCL model that is widely used in pharmacogenomic testing of drugs was applied to testing of over one hundred environmental chemicals (Abdo et al., 2015b). This study derived chemical-specific points of departure (POD) for cytotoxicity and explored potential genetic modifiers of inter-individual variability. In addition, a series of studies were recently published using the population-based model of human iPSC-derived cardiomyocytes from healthy humans of diverse genetic backgrounds. This experimental model is not only now widely used is studies of preclinical safety of drug candidates, but also can be used to evaluate potential cardiotoxicity of environmental and industrial chemicals, a major gap in safety assessment of non-pharmaceuticals because rodent models are unsuitable for testing of the potential causative involvement of environmental factors in cardiovascular disease (Burnett et al., 2021b). In vitro iPSC-derived cardiomyocyte data on interindividual variability in baseline beating parameters and drug-induced effects are highly reproducible (Grimm et al., 2018). Recent studies evaluated hundreds of chemicals and drugs in the population-based model and demonstrated its utility for rapid, high-throughput hazard characterization of chemicals for which little to no cardiotoxicity data are available from guideline studies in animals (Blanchette et al., 2020; Blanchette et al., 2022; Burnett et al., 2021a; Burnett et al., 2019). These studies showed not only that an overall cardiotoxicity hazard can be predicted, but also that the types of a hazardous effect (e.g., changes in beating frequency and/or electrophysiology) can be predicted with confidence (Blanchette et al., 2020). Finally, in vitro studies using cultured primary hepatocytes from a panel of inbred mouse strains were shown to be a feasible population-based chemical toxicity testing model (Martinez et al., 2010). This study demonstrated that inter-strain variability in liver toxicity of acetaminophen, WY-14,643 and rifampin can be evaluated and concentration-response effects on viability and function shown.
Exposure-Response Assessment
Similar to the challenges with hazard identification, adequate dose-, or concentration-response data for quantitative part of risk assessment are most often not available from human subjects. Data from animal studies are frequently used to determine the exposure-response relationships, in which case extrapolation must be done to human exposures to account for inter-species and inter-individual variability using uncertainty factors. In this respect, both human in vitro and animal in vivo population-based models have provided chemical-specific information to inform exposure-response assessments. Two regulatory contexts with a number of examples detailed below are (i) increasing confidence in the point of departure and/or margin of safety estimates, and (ii) obtaining experimentally-derived chemical-specific estimates of uncertainty in toxicokinetic and/or toxicodynamic variability.
The genetic homogeneity of toxicity testing data is well-established. For instance, in two studies that identified over 500 experimental animal studies that included more than three dose groups (including the control) and thus could be modeled using benchmark dose approach (Watford et al., 2019; Wignall et al., 2014), most of these utilized a single rodent strain: the B6C3F1 hybrid. Because the inter-strain variability in toxicokinetics and toxicodynamics has been observed in virtually every multi-strain study (see above), concerns exist about the use PODs derived from a non-heterogenous animal model as a starting point is exposure-response assessments (Chiu and Rusyn, 2018). Few examples are available of dose-response studies in multi-strain panels; the dearth of such studies is largely due to the expense and complexity. The most relevant to traditional risk assessment is a study of benzene inhalation over 28 days in a Diversity Outbred model (French et al., 2015). It was shown that an increase in benzene-induced chromosomal damage was dose-dependent and reproducible, and that the POD estimated from these population-based studies was an order of magnitude below the POD value estimated using the standard B6C3F1 strain. This single sex and single species study employed one control and three exposed dose groups, as compared to a traditional 28-day study where 5 animals per group (i.e. male mice) would be used (Hayashi et al., 1994). Still, the Diversity Outbred study was 15 times the number of animals of a traditional 28-day mouse study, which is not a realistic strategy to replace current guideline study designs. A number of other studies of exposure-response design in panels of inbred strains or Collaborative Cross population were published; however, most of those focused on chemical metabolism and were of very short duration (Bradford et al., 2011; Cichocki et al., 2017; Venkatratnam et al., 2017).
A number of studies that used population-based mouse models quantified toxicodynamic variability, hence providing chemical-specific data for risk assessment to replace default uncertainty factors. For example, in a study of tetrachloroethylene in Collaborative Cross mouse population (Luo et al., 2019), the total uncertainty factor (both kinetics and dynamics) was calculated for liver and kidney effects and those were compared to default assumptions used by the US EPA in their risk assessment (U.S. EPA, 2012). For liver effects, experimentally-derived total UFH (1.2) was smaller than the default value. For the kidney effects, study-derived total UFH (7.6) was close to the default UF of 10. Collectively, this study confirmed the appropriate use of the default values; however, it also provided scientific rationale for using lower UFs in deriving candidate reference doses for liver and kidney. Two studies of 1,3-butadiene in Collaborative Cross mouse population calculated uncertainty factors for DNA damage and epigenetic effects to provide quantitative estimates that may be used in evaluation of inter-individual differences in cancer susceptibility for this high production volume chemical. Interestingly, the UFs derived from the data on butadiene-associated DNA damage or histone methylation in lung, liver or kidneys (likely representing cumulative variability in toxicokinetics and toxicodynamics) were below 2; only DNA methylation-derived UFs were greater (2.7 to 7.9 depending on the tissue), yet still below the default UF of 10 (Lewis et al., 2019). A study of urinary adducts of 1,3-butadiene showed similar results for mice, UF were 3 or less depending on sex (Erber et al., 2021), while the variability in humans, based on the data from occupationally-exposed workers, was about an order of magnitude higher. Collectively, these mouse and human population-based studies of 1,3-butadiene showed that UF derived based on the data from several small human occupational studies are sufficiently conservative.
A more realistic, in terms of cost and throughput, approach to derive population-based PODs and margins of exposure/safety may be the approach based on human in vitro models. Such data are available for dozens to hundreds of chemicals and drugs that were tested in human population-derived LCLs (Abdo et al., 2015b), iPSC-derived cardiomyocytes (Burnett et al., 2019), and fibroblasts (Burnett et al., 2021c). Even though direct comparisons with human in vivo data are difficult, when such comparisons were possible, the outcomes are highly encouraging. With respect to LCLs, an indirect comparison was made in a recent study analyzing data from 10 pesticides. Specifically, it was found that combining population-based reverse toxicokinetics with population-based LCL toxicodynamic variability resulted in human toxicity values that were highly concordant with those extrapolated from experimental animal data after accounting for inter-species uncertainty and human variability (Chiu and Paoli, 2021). More direct comparisons have been possible with iPSC-derived cardiomyocytes. For example, a study of 13 drugs with clinical trial data on the risk of long QT showed that in vitro data from a population-based iPSC-derived cardiomyocyte model accurately predicted blood concentrations in patients at which long QT was observed (Blanchette et al., 2019). Likewise, when in vitro population-derived PODs were compared to measured or estimated in vivo blood concentrations of drugs or chemicals, the lowest margins of exposure were concordant with known cardiotoxic effects for certain drugs and chemicals (Blanchette et al., 2020).
Human in vitro studies can also be used to derive chemical-specific intra-species UFs, which would be most likely only reflective of toxicodynamic variability because of the limited capacity of the existing population-based in vitro models (LCLs, cardiomyocytes and fibroblasts) to metabolize chemicals. For example, the (Abdo et al., 2015b) study derived quantitative estimates of chemical-specific variability in cytotoxicity and explored potential genetic modifiers of inter-individual variability. It was found that for about half of tested compounds, cytotoxic response in the 1% most “sensitive” individual occurred at concentrations within a factor of 10½ (i.e., approximately 3) of that in the median individual; however, for some compounds, this factor was >>10. It was concluded that this experimental approach can fill critical gaps in the large-scale toxicity testing programs, and provide quantitative, experimentally-based and chemical-specific estimates of human toxicodynamic variability. Subsequent studies conducted similar analyses using iPSC-derived cardiomyocytes (Blanchette et al., 2020) and human skin fibroblasts (Burnett et al., 2021c), as well as compared derived toxicodynamic variability factors with human in vivo data from clinical and epidemiological studies (WHO/IPCS, 2018). Importantly for future use of these models in risk assessment, even though the number and type of chemicals and individuals used varied between studies, the toxicodynamic variability factor estimates for cytotoxicity derived in each study were generally close to the default value of 3.16 (Burnett et al., 2021c).
Similar to the challenge of the study size when an animal population-based model is proposed, in vitro studies in cells from multiple individuals are also more complex. However, it was shown that in vitro studies for exposure-response analysis could be highly informative even when performed in a limited number of genetically-diverse individuals. For LCLs, a tiered Bayesian strategy for fit-for-purpose population variability estimates showed that a pilot experiment using samples sizes of ~20 individuals would reduce prior uncertainty by >50% with > 80% balanced accuracy for classification and a high confidence experiment could be performed with sample sizes of ~50–100 individual LCLs (Chiu et al., 2017). For iPSC-derived cardiomyocytes, for potency and clinical risk of long QT, a cohort of 5 randomly-selected unique donors would provide accurate and precise estimates (Blanchette et al., 2022). For estimating inter-individual variability, cohorts of at least 20 donors would be needed, with smaller populations on average showing bias towards underestimation in population variance. No such analysis is available yet for human fibroblasts because only limited population size has been evaluated (Burnett et al., 2021c).
Finally, in vitro models have been used to evaluate inter-species differences in a variety of risk assessment-relevant outcomes such as chemical metabolism, cytotoxicity, oxidative stress and DNA repair capacity. Most typically, dermal fibroblasts were used as a model that was accessible from multiple species and individuals (Souci and Denesvre, 2021). A number of studies showed that cultured dermal fibroblasts retain differences in longevity across mammals, features that are preserved at the level of global gene expression and metabolite concentrations, and that cells derived from longer-lived species are more resistant to cellular stress and cytotoxicity elicited by several chemicals and stress conditions (Csiszar et al., 2012; Harper et al., 2007; Harper et al., 2011; Kapahi et al., 1999). To extend this conceptual study design to risk assessment, a follow-up study characterized inter-species variability in cytotoxicity for 40 chemicals using primary dermal fibroblasts from 68 individuals of 54 taxonomically diverse species (Burnett et al., 2021c). It was shown that in vitro-derived estimates of inter-species variability were, on average, similar to the default UFs used in risk assessment; however, substantial chemical-to-chemical differences were noted, with numerous chemicals exceeding default UFs by up to an order of magnitude or more. Thus, this study supports the use of in vitro-derived chemical-specific data in the new approach methods-based risk assessment framework to reduce uncertainty and more reliably characterize inter-species and inter-individual variability.
Mechanistic Evidence
Most risk assessment procedures for cancer and non-cancer adverse health effects include consideration of the molecular and cellular events to determine the biological plausibility of the findings in experimental systems to human health. For example, when assessing a chemical for potential carcinogenicity, current US EPA practice is to focus on analysis of a mode of action, a sequence of key events and processes, starting with interaction of an agent and a cell, proceeding through operational and anatomical changes, and resulting in cancer formation (U.S. Environmental Protection Agency, 2005). Related concepts of adverse outcome pathways (Ankley et al., 2010) and key characteristics (Smith et al., 2020) have been proposed. In either case, whichever way the mechanistic evidence is collected and evaluated, the knowledge of the potential for inter-individual variability in the key molecular regulators and effectors of toxicity, as well as the potential to identify the loci that determine susceptibility, is critically important. Even though genetic variability is but one component of the “intrinsic and extrinsic factors that increase susceptibility or exposure” that also include “pregnant women, infants, children, families living near current and former industrial sites, or any other potentially highly exposed sub-populations” (Koman et al., 2019), most population-based experimental studies provide the data on the former (Miller et al., 2001). A number of approaches have been used to either identify loci that may confer susceptibility in human sub-populations, or quantify population variability in a chemical-specific manner to reduce uncertainty in risk estimates. These include data from molecular epidemiology studies (Grandjean, 1992), experimental model systems (Chiu and Rusyn, 2018; Lash et al., 2003), or purely computational predictions (Kosnik et al., 2021).
Conclusions
Lack of quantitative characterization of inter- and intra-species variability has been long recognized as key issue in chemical risk assessment, but for the first 50+ years since the “acceptable daily dose” concept was introduced (Lehman, 1954), there were few experimental options for addressing them beyond dividing by “default factors” of 10. Moreover, although biomonitoring- and biomarker-based risk assessments have become increasingly popular, they still largely need to on default factors for addressing human variability due to the lack of TK or TD data (Apel et al., 2020; Aylward et al., 2013; Faure et al., 2020). Since the sequencing of the human genome, there has been considerable excitement as to the possibility of addressing this data gap in risk assessment, with many authoritative recommendations to explore how to utilize novel biological models and approaches (Cote et al., 2016). However, it is only in the last decade that the promise has begun to bear fruit through the development of robust and reproducible in vivo and in vitro model systems, and demonstration of their utility in chemical risk assessment for a number of chemicals. Indeed, not only has there been a better understanding of where such approaches fit into the risk assessment paradigm, but also there have been multiple cases studies for several of these applications. Overall, the most progress has been made on use of population based human in vitro cells and population-based in vivo mouse models, with clear demonstration as to their utility in hazard identification, exposure-response assessment, and mechanistic understanding. For human in vitro models, a key limitation remains the limited number of cell types for which population-based models are available (currently LCLs, cardiomyocytes, and fibroblasts), which limits the types of toxicity endpoints that can be assessed. For population-based in vivo models, key challenges include the complexity of the study design and analyses necessary, as well as the mouse being the only robust and widely-used population model for studies of chemical toxicity. Still, there is some hope that development of non-human in vitro population-based models could help ameliorate these issues, perhaps enabling in vivo studies to narrow their focus or even replacing in vivo studies in some cases; however, this area is relatively undeveloped in part for technical reasons (challenges in developing reproducible models). Overall, population-based models not only have a great potential to address numerous long-standing data gaps and limitations in chemical risk assessment, but the substantial growth in case studies and applications in the last decade shows that this potential is now beginning to be realized.
Acknowledgements
These studies were supported, in part, by the grants from the US Environmental Protection Agency (RD83561202) and the National Institute of Environmental health Sciences (R01 ES029911 and P42 ES027704). The views expressed in this manuscript do not reflect those of the funding agencies. The use of specific commercial products in this work does not constitute endorsement by the funding agencies.
References
- 1000 Genomes Project Consortium, et al. , 2012. An integrated map of genetic variation from 1,092 human genomes. Nature. 491, 56–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Abdo N, et al. , 2015a. In vitro screening for population variability in toxicity of pesticide-containing mixtures. Environ Int. 85, 147–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Abdo N, et al. , 2015b. 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]
- Akhtar A, 2015. The flaws and human harms of animal experimentation. Camb Q Healthc Ethics. 24, 407–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ankley GT, et al. , 2010. Adverse outcome pathways: a conceptual framework to support ecotoxicology research and risk assessment. Environ Toxicol Chem. 29, 730–41. [DOI] [PubMed] [Google Scholar]
- Apel P, et al. , 2020. Human biomonitoring initiative (HBM4EU) - Strategy to derive human biomonitoring guidance values (HBM-GVs) for health risk assessment. Int J Hyg Environ Health. 230, 113622. [DOI] [PubMed] [Google Scholar]
- Aylward LL, et al. , 2013. Evaluation of biomonitoring data from the CDC National Exposure Report in a risk assessment context: perspectives across chemicals. Environ Health Perspect. 121, 287–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bartko JJ, 1966. The intraclass correlation coefficient as a measure of reliability. Psychol Rep. 19, 3–11. [DOI] [PubMed] [Google Scholar]
- Bhat VS, et al. , 2017. Evolution of chemical-specific adjustment factors (CSAF) based on recent international experience; increasing utility and facilitating regulatory acceptance. Crit Rev Toxicol. 47, 729–749. [DOI] [PubMed] [Google Scholar]
- Blanchette AD, et al. , 2020. A Bayesian Method for Population-wide Cardiotoxicity Hazard and Risk Characterization Using an In Vitro Human Model. Toxicol Sci. 178, 391–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Blanchette AD, et al. , 2022. A tiered approach to population-based in vitro testing for cardiotoxicity: Balancing estimates of potency and variability. J Pharmacol Toxicol Methods. 107154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Blanchette AD, et al. , 2019. Thorough QT/QTc in a Dish: An In Vitro Human Model That Accurately Predicts Clinical Concentration-QTc Relationships. Clin Pharmacol Ther. 105, 1175–1186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bradford BU, et al. , 2011. Interstrain differences in the liver effects of trichloroethylene in a multistrain panel of inbred mice. Toxicol Sci. 120, 206–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Buratti FM, et al. , 2021. Human variability in glutathione-S-transferase activities, tissue distribution and major polymorphic variants: Meta-analysis and implication for chemical risk assessment. Toxicol Lett. 337, 78–90. [DOI] [PubMed] [Google Scholar]
- Burnett SD, et al. , 2021a. Cardiotoxicity Hazard and Risk Characterization of ToxCast Chemicals Using Human Induced Pluripotent Stem Cell-Derived Cardiomyocytes from Multiple Donors. Chem Res Toxicol. 34, 2110–2124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Burnett SD, et al. , 2021b. Human induced pluripotent stem cell (iPSC)-derived cardiomyocytes as an in vitro model in toxicology: strengths and weaknesses for hazard identification and risk characterization. Expert Opin Drug Metab Toxicol. 17, 887–902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Burnett SD, et al. , 2019. Population-based toxicity screening in human induced pluripotent stem cell-derived cardiomyocytes. Toxicol Appl Pharmacol. 381, 114711. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Burnett SD, et al. , 2021c. A new approach method for characterizing inter-species toxicodynamic variability. J Toxicol Environ Health A. 84, 1020–1039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bush WS, et al. , 2016. Genetic Variation Among 82 Pharmacogenes: The PGRNseq Data From the eMERGE Network. Clinical Pharmacology & Therapeutics. 100, 160–169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bussiere JL, et al. , 2009. Alternative strategies for toxicity testing of species-specific biopharmaceuticals. Int J Toxicol. 28, 230–53. [DOI] [PubMed] [Google Scholar]
- Byers D, 2008. Components of phenotypic variance. Nature Ed. 1, 161. [Google Scholar]
- 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]
- Chhabra RS, et al. , 1990. An overview of prechronic and chronic toxicity/carcinogenicity experimental study designs and criteria used by the National Toxicology Program. Environ Health Perspect. 86, 313–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- Chiu WA, Paoli GM, 2021. Recent Advances in Probabilistic Dose-Response Assessment to Inform Risk-Based Decision Making. Risk Anal. 41, 596–609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- Chiu WA, et al. , 2017. A tiered, Bayesian approach to estimating of population variability for regulatory decision-making. ALTEX. 34, 377–388. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Church RJ, et al. , 2015. Sensitivity to hepatotoxicity due to epigallocatechin gallate is affected by genetic background in diversity outbred mice. Food Chem Toxicol. 76, 19–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Church RJ, et al. , 2014. A systems biology approach utilizing a mouse diversity panel identifies genetic differences influencing isoniazid-induced microvesicular steatosis. Toxicol Sci. 140, 481–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Churchill GA, et al. , 2012. The Diversity Outbred mouse population. Mamm Genome. 23, 713–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- Corton JC, et al. , 2014. Mode of action framework analysis for receptor-mediated toxicity: The peroxisome proliferator-activated receptor alpha (PPARalpha) as a case study. Crit Rev Toxicol. 44, 1–49. [DOI] [PubMed] [Google Scholar]
- Cote I, et al. , 2016. The Next Generation of Risk Assessment Multi-Year Study-Highlights of Findings, Applications to Risk Assessment, and Future Directions. Environ Health Perspect. 124, 1671–1682. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Csiszar A, et al. , 2012. Testing the oxidative stress hypothesis of aging in primate fibroblasts: is there a correlation between species longevity and cellular ROS production? J Gerontol A Biol Sci Med Sci. 67, 841–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dalaijamts C, et al. , 2018. Incorporation of the glutathione conjugation pathway in an updated physiologically-based pharmacokinetic model for perchloroethylene in mice. Toxicol Appl Pharmacol. 352, 142–152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- 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]
- Darney K, et al. , 2020a. Bayesian meta-analysis of inter-phenotypic differences in human serum paraoxonase-1 activity for chemical risk assessment. Environ Int. 138, 105609. [DOI] [PubMed] [Google Scholar]
- Darney K, et al. , 2020b. Human variability in influx and efflux transporters in relation to uncertainty factors for chemical risk assessment. Food Chem Toxicol. 140, 111305. [DOI] [PubMed] [Google Scholar]
- Di Consiglio E, et al. , 2021. Human Variability in Carboxylesterases and carboxylesterase-related Uncertainty Factors for Chemical Risk Assessment. Toxicol Lett. 350, 162–170. [DOI] [PubMed] [Google Scholar]
- Dorne JL, et al. , 2012. Population effects and variability. Methods Mol Biol. 929, 521–81. [DOI] [PubMed] [Google Scholar]
- Dorne JL, et al. , 2005. Human variability in xenobiotic metabolism and pathway-related uncertainty factors for chemical risk assessment: a review. Food Chem Toxicol. 43, 203–16. [DOI] [PubMed] [Google Scholar]
- Dourson M, et al. , 2022. The Future of Uncertainty Factors With In Vitro Studies Using Human Cells. Toxicol Sci. 186, 12–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dourson ML, et al. , 1996. Evolution of science-based uncertainty factors in noncancer risk assessment. Regul Toxicol Pharmacol. 24, 108–20. [DOI] [PubMed] [Google Scholar]
- Dourson ML, Stara JF, 1983. Regulatory history and experimental support of uncertainty (safety) factors. Regul Toxicol Pharmacol. 3, 224–38. [DOI] [PubMed] [Google Scholar]
- Eduati F, et al. , 2015. Prediction of human population responses to toxic compounds by a collaborative competition. Nat Biotechnol. 33, 933–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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. 34, 2375–2383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Faure S, et al. , 2020. Evaluation of human biomonitoring data in a health risk based context: An updated analysis of population level data from the Canadian Health Measures Survey. Int J Hyg Environ Health. 223, 267–280. [DOI] [PubMed] [Google Scholar]
- Festing MF, 1979. Properties of inbred strains and outbred stocks, with special reference to toxicity testing. J Toxicol Environ Health. 5, 53–68. [DOI] [PubMed] [Google Scholar]
- Festing MF, 1986. The case for isogenic strains in toxicological screening. Arch Toxicol.Suppl 9, 127–137. [DOI] [PubMed] [Google Scholar]
- Festing MF, 2010. Inbred strains should replace outbred stocks in toxicology, safety testing, and drug development. Toxicol Pathol. 38, 681–90. [DOI] [PubMed] [Google Scholar]
- Festing S, Wilkinson R, 2007. The ethics of animal research. Talking Point on the use of animals in scientific research. EMBO Rep. 8, 526–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Frazer KA, et al. , 2007. A sequence-based variation map of 8.27 million SNPs in inbred mouse strains. Nature. 448, 1050–3. [DOI] [PubMed] [Google Scholar]
- 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]
- Gaedigk A, et al. , 2017. Prediction of CYP2D6 phenotype from genotype across world populations. Genet Med. 19, 69–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gentry PR, et al. , 2002. An approach for the quantitative consideration of genetic polymorphism data in chemical risk assessment: examples with warfarin and parathion. Toxicol Sci. 70, 120–139. [DOI] [PubMed] [Google Scholar]
- Giacomini KM, et al. , 2017. Genome-wide association studies of drug response and toxicity: an opportunity for genome medicine. Nat Rev Drug Discov. 16, 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Grandjean P, 1992. Individual susceptibility to toxicity. Toxicol Lett. 64–65 Spec No, 43–51. [DOI] [PubMed] [Google Scholar]
- Grimm FA, et al. , 2018. A human population-based organotypic in vitro model for cardiotoxicity screening. ALTEX. 35, 441–452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Grimm FA, et al. , 2019. Multi-dimensional in vitro bioactivity profiling for grouping of glycol ethers. Regul Toxicol Pharmacol. 101, 91–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harper JM, et al. , 2007. Skin-derived fibroblasts from long-lived species are resistant to some, but not all, lethal stresses and to the mitochondrial inhibitor rotenone. Aging Cell. 6, 1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harper JM, et al. , 2011. Fibroblasts from long-lived bird species are resistant to multiple forms of stress. J Exp Biol. 214, 1902–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harrill AH, et al. , 2012. A mouse diversity panel approach reveals the potential for clinical kidney injury due to DB289 not predicted by classical rodent models. Toxicol Sci. 130, 416–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harrill AH, et al. , 2018. Mouse population-based evaluation of urinary protein and miRNA biomarker performance associated with cisplatin renal injury. Exp Biol Med (Maywood). 243, 237–247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- Harrill AH, et al. , 2009a. Population-based discovery of toxicogenomics biomarkers for hepatotoxicity using a laboratory strain diversity panel. Toxicol Sci. 110, 235–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harrill AH, et al. , 2009b. Mouse population-guided resequencing reveals that variants in CD44 contribute to acetaminophen-induced liver injury in humans. Genome Res. 19, 1507–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hartung T, 2008. Food for thought… on animal tests. ALTEX. 25, 3–16. [DOI] [PubMed] [Google Scholar]
- Haufroid V, Hantson P, 2015. CYP2D6 genetic polymorphisms and their relevance for poisoning due to amfetamines, opioid analgesics and antidepressants. Clin Toxicol (Phila). 53, 501–10. [DOI] [PubMed] [Google Scholar]
- Haverty PM, et al. , 2016. Reproducible pharmacogenomic profiling of cancer cell line panels. Nature. 533, 333–7. [DOI] [PubMed] [Google Scholar]
- Hayashi M, et al. , 1994. In vivo rodent erythrocyte micronucleus assay Mutation Research. 312, 293–304. [DOI] [PubMed] [Google Scholar]
- Institute of Medicine, 2001. 5: Using Science-Based Risk Assessment to Develop Food Safety Policy. in: Food Forum, (Ed.), Food Safety Policy, Science, and Risk Assessment: Strengthening the Connection: Workshop Proceedings. National Academy Press, Washington, DC. [PubMed] [Google Scholar]
- 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]
- Judson R, et al. , 2009. The toxicity data landscape for environmental chemicals. Environ Health Perspect. 117, 685–695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kapahi P, et al. , 1999. Positive correlation between mammalian life span and cellular resistance to stress. Free Radic Biol Med. 26, 495–500. [DOI] [PubMed] [Google Scholar]
- Koman PD, et al. , 2019. Population susceptibility: A vital consideration in chemical risk evaluation under the Lautenberg Toxic Substances Control Act. PLoS Biol. 17, e3000372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kosnik MB, et al. , 2021. Distinct genetic regions are associated with differential population susceptibility to chemical exposures. Environ Int. 152, 106488. [DOI] [PubMed] [Google Scholar]
- Kostewicz ES, et al. , 2014. PBPK models for the prediction of in vivo performance of oral dosage forms. Eur J Pharm Sci. 57, 300–21. [DOI] [PubMed] [Google Scholar]
- 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]
- Krebs K, Milani L, 2019. Translating pharmacogenomics into clinical decisions: do not let the perfect be the enemy of the good. Hum Genomics. 13, 39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krewski D, et al. , 2019. Concordance between sites of tumor development in humans and in experimental animals for 111 agents that are carcinogenic to humans. J Toxicol Environ Health B Crit Rev. 22, 203–236. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lappalainen T, et al. , 2013. Transcriptome and genome sequencing uncovers functional variation in humans. Nature. 501, 506–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lash LH, et al. , 2003. Genetics and susceptibility to toxic chemicals: do you (or should you) know your genetic profile? J Pharmacol Exp Ther. 305, 403–9. [DOI] [PubMed] [Google Scholar]
- Lehman AF, OG 1954. 100-fold margin of safety Quarterly Bulletin - Association of Food & Drug Officials of the United States. 18, 33–35. [Google Scholar]
- 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]
- Lock EF, et al. , 2012. Quantitative high-throughput screening for chemical toxicity in a population-based in vitro model. Toxicol Sci. 126, 578–588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- Luo YS, et al. , 2018a. 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]
- Luo YS, et al. , 2018b. Comparative analysis of metabolism of trichloroethylene and tetrachloroethylene among mouse tissues and strains. Toxicology. 409, 33–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Martinez SM, et al. , 2010. Evaluation of an in vitro toxicogenetic mouse model for hepatotoxicity. Toxicol Appl Pharmacol. 249, 208–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Meek ME, et al. , 2002. Guidelines for application of chemical-specific adjustment factors in dose/concentration-response assessment. Toxicology. 181–182, 115–20. [DOI] [PubMed] [Google Scholar]
- Miller MC 3rd, et al. , 2001. Genetic variability in susceptibility and response to toxicants. Toxicol Lett. 120, 269–80. [DOI] [PubMed] [Google Scholar]
- Motsinger-Reif AA, et al. , 2013. Genome-wide association studies in pharmacogenomics: successes and lessons. Pharmacogenet Genomics. 23, 383–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- National Research Council, 1983. Risk assessment in the federal government: Managing the process. National Academies Press, Washington, D.C. [PubMed] [Google Scholar]
- National Research Council, 2009. Science and Decisions: Advancing Risk Assessment. National Academies Press, Washington, DC. [PubMed] [Google Scholar]
- National Research Council, 2011. Review of the Environmental Protection Agency’s Draft IRIS Assessment of Formaldehyde. National Academies Press, Washington, DC. [PubMed] [Google Scholar]
- National Research Council, Review of EPA’s Integrated Risk Information System (IRIS) Process. National Academies Press, Washington, DC, 2014. [PubMed] [Google Scholar]
- Nguyen HT, et al. , 2017. Strain differences in the proteome of dioxin-sensitive and dioxin-resistant mice treated with 2,3,7,8-tetrabromodibenzo-p-dioxin. Arch Toxicol. 91, 1763–1782. [DOI] [PubMed] [Google Scholar]
- O’Shea SH, et al. , 2011. In vitro screening for population variability in chemical toxicity. Toxicol Sci. 119, 398–407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Olson H, et al. , 2000. Concordance of the toxicity of pharmaceuticals in humans and in animals. Regul Toxicol Pharmacol. 32, 56–67. [DOI] [PubMed] [Google Scholar]
- Ortmann D, et al. , 2020. Naive Pluripotent Stem Cells Exhibit Phenotypic Variability that Is Driven by Genetic Variation. Cell Stem Cell. 27, 470–481 e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pathak SM, et al. , 2017. Model-Based Analysis of Biopharmaceutic Experiments To Improve Mechanistic Oral Absorption Modeling: An Integrated in Vitro in Vivo Extrapolation Perspective Using Ketoconazole as a Model Drug. Mol Pharm. 14, 4305–4320. [DOI] [PubMed] [Google Scholar]
- Paul Friedman K, et al. , 2020. Utility of In Vitro Bioactivity as a Lower Bound Estimate of In Vivo Adverse Effect Levels and in Risk-Based Prioritization. Toxicol Sci. 173, 202–225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pleil JD, et al. , 2018. Human biomarker interpretation: the importance of intra-class correlation coefficients (ICC) and their calculations based on mixed models, ANOVA, and variance estimates. J Toxicol Environ Health B Crit Rev. 21, 161–180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reisberg S, et al. , 2019. Translating genotype data of 44,000 biobank participants into clinical pharmacogenetic recommendations: challenges and solutions. Genet Med. 21, 1345–1354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Relling MV, Evans WE, 2015. Pharmacogenomics in the clinic. Nature. 526, 343–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Renwick AG, 1993. Data-derived safety factors for the evaluation of food additives and environmental contaminants. Food Addit Contam. 10, 275–305. [DOI] [PubMed] [Google Scholar]
- Richard AM, et al. , 2021. The Tox21 10K Compound Library: Collaborative Chemistry Advancing Toxicology. Chem Res Toxicol. 34, 189–216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rusyn I, et al. , 2018. Introduction to mammalian genome special issue: the combined role of genetics and environment relevant to human disease outcomes. Mamm Genome. 29, 1–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rusyn I, et al. , 2006. Modes of action and species-specific effects of di-(2-ethylhexyl)phthalate in the liver. Crit Rev Toxicol. 36, 459–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Smith MT, et al. , 2020. The Key Characteristics of Carcinogens: Relationship to the Hallmarks of Cancer, Relevant Biomarkers, and Assays to Measure Them. Cancer Epidemiol Biomarkers Prev. 29, 1887–1903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Souci L, Denesvre C, 2021. 3D skin models in domestic animals. Vet Res. 52, 21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Threadgill DW, Churchill GA, 2012. Ten years of the Collaborative Cross. Genetics. 190, 291–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tsuchiya M, et al. , 2012. Interstrain differences in liver injury and one-carbon metabolism in alcohol-fed mice. Hepatology. 56, 130–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- U.S. Environmental Protection Agency, Guidelines for carcinogen risk assessment. Risk Assessment Forum, US Environmental Protection Agency, Washington, DC, 2005. [Google Scholar]
- U.S. EPA, Toxicological Review of Tetrachloroethylene (CAS No. 127-18-4): In Support of Summary Information on the Integrated Risk Information System (IRIS). U.S. Environmental Protection Agency, Washington, DC, 2012. [Google Scholar]
- Venkatratnam A, et al. , 2017. Collaborative Cross Mouse Population Enables Refinements to Characterization of the Variability in Toxicokinetics of Trichloroethylene and Provides Genetic Evidence for the Role of PPAR Pathway in Its Oxidative Metabolism. Toxicol Sci. 158, 48–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- Watford S, et al. , 2019. ToxRefDB version 2.0: Improved utility for predictive and retrospective toxicology analyses. Reprod Toxicol. 89, 145–158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weil CS, 1972. Statistics vs safety factors and scientific judgment in the evaluation of safety for man. Toxicol Appl Pharmacol. 21, 454–63. [DOI] [PubMed] [Google Scholar]
- Wheeler HE, Dolan ME, 2012. Lymphoblastoid cell lines in pharmacogenomic discovery and clinical translation. Pharmacogenomics. 13, 55–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- WHO, Evaluation of the toxicity of pesticide residues in food / deliberations of the Joint Meeting of the FAO Committee on Pesticides in Agriculture and the WHO Expert Committee on Pesticide Residues, Rome, 15–22 March 1965. In: Joint Meeting of the F. A. O. Committee on Pesticides in Agriculture, t. W. H. O. E. C. o. P. R., (Ed.), Rome, Italy, 1965. [Google Scholar]
- 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]
- Wignall JA, et al. , 2014. Standardizing benchmark dose calculations to improve science-based decisions in human health assessments. Environ Health Perspect. 122, 499–505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yang J, et al. , 2011. GCTA: a tool for genome-wide complex trait analysis. Am J Hum Genet. 88, 76–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yoo HS, et al. , 2015a. Comparative analysis of the relationship between trichloroethylene metabolism and tissue-specific toxicity among inbred mouse strains: liver effects. J Toxicol Environ Health A. 78, 15–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yoo HS, et al. , 2015b. 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]
- You D, et al. , 2020. Nitrosative Stress and Lipid Homeostasis as a Mechanism for Zileuton Hepatotoxicity and Resistance in Genetically Sensitive Mice. Toxicol Sci. 175, 220–235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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]
- Zhang X, et al. , 2020. Application of PBPK Modeling and Simulation for Regulatory Decision Making and Its Impact on US Prescribing Information: An Update on the 2018–2019 Submissions to the US FDA’s Office of Clinical Pharmacology. J Clin Pharmacol. 60 Suppl 1, S160–S178. [DOI] [PubMed] [Google Scholar]
- Zhou Y, et al. , 2017. Worldwide Distribution of Cytochrome P450 Alleles: A Meta-analysis of Population-scale Sequencing Projects. Clin Pharmacol Ther. 102, 688–700. [DOI] [PMC free article] [PubMed] [Google Scholar]