ToxPoint: In the Era of Precision Medicine, Diversity Should Not Be Neglected in Chemical Safety Assessment

There is tremendous interest in personalized and precision medicine that account for individual susceptibilities due in large part to genetic diversity. Paradoxically, toxicologists typically limit their investigations to using genetically inbred mouse strains or cells or cell lines with limited genetic heterogeneity. This is important to note because decades of research in pharmacogenomics have yielded hundreds of genetic variants that influence the pharmacokinetics and pharmacodynamics of drugs and chemicals. Differences in disease risk among racial/ethnic groups are widely recognized in the fields of medicine, epidemiology, and social sciences. In an era when the National Institutes of Health is investing in population-based studies examining millions of individuals such as “All of Us,” there remains a need to develop tools and approaches that evaluate the influence of genetic heterogeneity on toxicological responses. It should be noted that precision medicine is still evolving and has inherent biases to overcome, such as gene-disease association studies focused primarily on individuals of European ancestry (Ioannidis et al., 2004). However, precision medicine and pharmacogenomics benefit from human exposure studies that are most often not possible in the safety evaluation of industrial chemicals and environmental pollutants.

There are well-established tools available today that enable scientists to screen chemicals within genetically diverse populations. However, before illustrating these models, it is necessary to discuss the value of testing chemicals in a population space. Chemical testing in genetically diverse mouse models has been shown to reflect differential sensitivities for human-relevant toxicities that occur only in genetically sensitive subsets of humans. This was observed for acetaminophen, which caused liver biomarker elevations in healthy subjects of Hispanic descent (Harrill et al., 2009), and for green tea extract, which caused “idiosyncratic” liver failure in a small number of susceptible humans (Church et al., 2015). These studies highlight the need for population-based approaches for hazard identification because they demonstrate that selection of an animal model for testing can lead to erroneous conclusions that there is not a human-relevant effect, when in fact the adverse effect can only be observed in certain genetic contexts (eg, rodent strains or susceptible human donor cells). A challenge is that it is generally not possible to know a priori which mouse strains will be susceptible; this is in part why it is critical to screen chemicals within a genetically diverse population.

Second, dose-response testing in a population space enables assessment of shifts in the point of departure that occur across individuals(Venkatratnam et al., 2018). Although there are tools available in pharmacokinetic modeling that allow scientists to account for interindividual differences in metabolism, there are few computational strategies that can accurately predict differences in downstream pharmacodynamic processes. Gaining insight into the distribution of dose-responsive adverse effects across a population is foundational to setting chemical-specific uncertainty factors that can replace default uncertainty factors in setting the human reference dose (Chiu et al., 2017). Indeed, it has been shown in population-based human lymphoblastoid cell line models that some chemicals exhibit interhuman variability that exceeds the default uncertainty factors and, therefore, a data-driven approach to generate variability factors would be more protective of human health.

There are population-based animal models available that could replace conventional rodent strains in guideline toxicity testing. One such model is the Diversity Outbred (J: DO) mouse stock, in which (similar to humans) each individual mouse is genetically unique. A benefit to using this model is a high level of hybrid vigor and randomized genetic polymorphisms that facilitate assessment of gene by environment interactions in a chemical exposure context. A tradeoff of using a highly diverse mouse population for testing is the requirement for a larger samples size so the studies have appropriate power to detect differences (Harrill et al., 2018). However, in cases where animal testing is necessary for decision making encompassing human health protection, Diversity Outbred mice provide an advantage of detecting human-relevant health hazards over genetically limited strains or conventional lines of inbred rodents. For testing contexts where a genetically matched control is needed, the highly diverse Collaborative Cross recombinant inbred lines are an attractive alternative.

In parallel, advances in human cell banking and engineering make population-based assay development more tractable for in vitro applications. Development of well-characterized human induced pluripotent stem cell banks is an ongoing international effort and there are thousands of cell lines available, many of which include extensive genetic characterization. International stem cell banks and numbers of human-derived lines available and their characterization status are summarized in a recent report (Kim et al., 2017). Furthermore, protocols for programing differentiation of stem cells into various mature types, such as neuronal cells and myocytes, are becoming more common (Pawlowski et al., 2017). These types of resources should continue to be developed to generate higher throughput human cell-based screening platforms for assessing interindividual differences in chemical dose-response. Data from these models could feed into statistical frameworks for determining chemical-specific uncertainty factors that address interindividual variability.

In summary, there are genetically diverse models available the enable query of population dynamics in response to chemicals. Their use is critical for the protection of public health and, in particular, genetically sensitive subpopulations that may exist within the broader public health context. Given the available toolsets, extrapolating data from a single rodent strain or a limited pool of human cells/cell lines may be useful for pilot studies but should not be solely relied upon for risk-based decision making. Population-based toxicology testing will enable a precision toxicology approach that protects not just a subset of the population, but all of us.

DECLARATION OF CONFLICTING INTERESTS

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Disclaimer: Views expressed are the author’s own and do not reflect positions or official policy of the US Government or the National Institutes of Health.