Toxicity Testing in the 21st Century: A View from the Chemical Industry

The rapid onset of the science and experimental tools of molecular biology and genomics has presented the toxicology community with an unprecedented opportunity to reexamine, and potentially significantly restructure, many of the decades-long principles and practices guiding translation of conventional animal toxicology data to scientifically informed assessments of adverse human health risks. Implementation of this energizing and transformational technology into toxicology has of course been recently catalyzed by the report of the U.S. National Research Council (NRC) report, Toxicity Testing in the 21st Century: A Vision and a Strategy (National Research Council, Committee on Toxicity Testing and Assessment of Environmental Agents, Board of Environmental Studies and Toxicology, Institute for Laboratory Animal Research, 2007a), and was overviewed in Part I of this Forum series (Andersen and Krewski, 2009). The NRC report outlined the promise that the new technologies could dramatically increase both the number of chemicals comprehensively evaluated as well as broadening and improving the human relevance of toxicity end points assessed. Such outcomes, if successfully achieved, would significantly improve and inform science-based health assessments on a broad range of environmental chemicals. It has long been recognized that the ability to fully characterize the potential hazard properties of the many thousands of environmental chemicals associated with chemical production by conventional toxicity test methods is pragmatically constrained by laboratory, animal, financial, and expertise resource limitations. And, of course, there are yet many thousands more chemicals originating as by-products of other human endeavors or as natural substances in the environment, all of which cannot be ignored if we are truly interested in understanding the contributions of environmental chemicals to adverse public health outcomes. The resource limitations associated with conventional toxicity testing are further emphasized when examining the testing resources necessary to develop pharmaceutical or pesticide products. Testing of these agents mandate use of an extensive and regulatory-proscribed core battery of sequential tests that cost millions of dollars per chemical and 4–5 years at a minimum to complete (and consume many thousands of animals). The chemical industry, facing the challenge of testing many thousands of more chemicals (to which human exposures may already be occurring) relative to the limited numbers of drug and pesticide products, has generally applied a tier-based testing framework for evaluation of chemicals and by-products of chemical production (Becker et al., 2007). The test protocols used are the same or very similar to those used in drug and pesticide evaluations; however, toxicity test implementation is sequenced using triggers to guide decisions on which tests of the total test battery are needed. Triggering decisions are driven by considerations of both observed toxicity in early-stage tests and/or information on potential margins of exposure anticipated or identified from substance use. Although the expressed intent of the NRC vision was to largely replace the need for live animal testing, the 21st century technologies might well be productively applied to further examining the scientific merits of tier-based testing strategies versus conduct of full test batteries. Enthusiasm engendered by the promise of the transformational 21st century technologies to de-bottleneck resource needs for improving understanding of environmental chemicals hazards and risks has rightly led to calls to establish government-funded programmatic investments to facilitate effective and coordinated implementation of these new technologies into the future mainstream of toxicity testing and human risk evaluation. In an immediate follow-up to the 21st Century report, the NRC report Applications of Toxicogenomic Technologies to Predictive Toxicology and Risk Assessment (National Research Council, Committee on Applications of Toxicogenomic Technologies to Predictive Toxicology, Board of Environmental Studies and Toxicology, Board of Life Studies, Division of Earth and Life Studies, 2007b) recommended funding a large-scale and coordinated program aimed specifically at addressing key implementation issues associated with translation of toxicogenomic technologies into actual practice. The Applications report also outlined some of the key challenges as well as opportunities facing effective and efficient implementation of this new technology and included specific commentary on some of the important issues specifically associated with assessment of environmental chemicals. Although the NRC reports provided important high-level insights of how the new technologies could serve as a foundational mechanism to reframe a modern era of toxicity testing, they did not (and reasonably could not be expected to) define a precise blueprint of how the objectives of the vision would or should be constructed. However, the NRC reports suggest that the road to achieving the objectives of the vision, the rewards of which will be high, will nonetheless be complicated, technically challenging, and likely filled with many potential turns and possible dead ends. That perspective alone mandates that research efforts to integrate these new technologies into the future practice of toxicology and risk assessment must involve an extraordinary level of creative and closely coordinated planning that is highly transparent and participatory to all segments of toxicology and risk scientists, i.e., academia, government, industry, nongovernment organizations, and ultimately the public. One of the values of this Forum series is to capture a spectrum of diverse experiences and perspectives of toxicologists from differing sectors of our discipline, with the intent that these insights may further stimulate effective and efficient integration of these exciting new technologies into a world of improved toxicity and risk evaluations. The purpose of this Forum piece was to offer the perspective of the chemical industry on this issue, with our particular focus and experience with industrial and environmental chemicals. The NRC Applications report (National Research Council, Committee on Applications of Toxicogenomic Technologies to Predictive Toxicology, Board of Environmental Studies and Toxicology, Board of Life Studies, Division of Earth and Life Studies, 2007b) acknowledged that evaluation of new compounds in development in highly regulated and contained environments, e.g., pharmaceuticals and pesticides, presented different types of toxicity assessment and interpretative challenges relative to the world of industrial chemicals and other environmental substances. Specifically, the public health implications of either false positives or negatives is distinctly different for compounds within the contained environments of precommercial development processes as compared to chemicals that are already in widespread direct use by the public and/or are widely distributed throughout the environment as by-products of production processes, consumer uses, or pollutants associated with other human endeavors (Table 1). Differential Implications of False-Positive/Negative Outcomes of High-Throughput 21st Century Toxicity Testing Approaches for Pharmaceutical Compounds in Development Versus Environmental Chemicals Already Presenting Real-World Human Exposures Differential Implications of False-Positive/Negative Outcomes of High-Throughput 21st Century Toxicity Testing Approaches for Pharmaceutical Compounds in Development Versus Environmental Chemicals Already Presenting Real-World Human Exposures For compounds in development programs, false-positive responses may mean that compounds with potential valuable health, societal, and/or commercial promise are unnecessarily abandoned. Such outcomes may result in ultimate introduction of substitutes that are not as effective in achieving commercial and/or societal expectations, or alternatively, cause continued use of existing materials possessing less than desirable properties, e.g., emerging “green chemistry” concerns. Conversely, false negatives may delay timely decisions to halt costly development on substances ultimately doomed to failure. In both of these cases, assuming the false negatives do not survive to commercial development, the immediate impact is largely to the institution developing the chemistry and/or associated products and technologies. In the specific case of pharmaceuticals, human clinical evaluations also further backstop development decisions associated with either potential false-negative or false-positive findings from preclinical toxicity evaluations, including those incorporating information from application of the new molecular/cellular technologies. Importantly, such clinical trial backstop options do not exist, for ethical reasons, for assessment of environmental chemicals. Apart from contained development scenarios, environmental chemicals present an entirely different set of health implications associated with false-positive or false-negative testing outcomes. Because many of these chemicals are already in the real-world environment, false-positive responses can result in ill-informed regulatory or other public responses, e.g., product bans or deselection pressures, resulting in society being denied continued access to valued agents. In addition, false positives may misdirect valued toxicological and other resources from problems ultimately of much higher public health concern. False negatives are equally a concern in that they may delay or even eliminate actions that in fact should be higher priorities for protecting public health. The false-positive/negative issues uniquely associated with environmental chemical demand that any transition to a future testing paradigm based predominately on nonwhole-animal testing be implemented with great caution. Such a transition, if it is to be successful, must be strongly focused on identifying true human health risks with a higher degree of confidence than that associated with existing test systems. It will always be necessary to evaluate relevance, reliability, sensitivity, and specificity of advanced high-throughput molecular screening and computational profiling methods prior to regulatory acceptance so that regulatory agencies, the regulated community, and the public have sufficient confidence in the decisions based on such methods. While traditional structures for conducting method validation and demonstrating model predictivity may not be practical, approaches such as those discussed by the National Research Council, Committee on Applications of Toxicogenomic Technologies to Predictive Toxicology, Board of Environmental Studies and Toxicology, Board of Life Studies, Division of Earth and Life Studies (2007b) with respect to validation of toxicogenomic technologies as well as practices embodied in the Organization for Economic Cooperation and Development (OECD) principles and guidance for the validation of quantitative structure activity relationships (OECD, 2007) and evidence-based toxicology (Guzelian et al., 2005; Hoffmann and Hartung, 2006) should be considered. The emergence of “green chemistry,” as a discipline devoted to promoting development of chemical products and processes that reduce or eliminate the use and generation of hazardous substances, carries with it a need to conduct side-by-side comparisons of hazard (and risk) profiles across a series of molecules with comparable beneficial uses. These comparisons would, theoretically, permit selection of the least hazardous (and risky) molecule(s). Such comparisons, however, will require a sufficient degree of confidence in the hazard profiles of the substances being compared. As noted above, it will not be efficient or cost-effective to conduct extensive and complex in vivo longer-term toxicity tests for all possible “green” molecules. So, if green chemistry is to have a substantial impact, there is a pressing need for more efficient tools to provide “data” on toxicity end points of concern and to support science-based decisions to select the “greenest” molecules in terms of potential hazards and risks. The approaches outlined by the 21st Century Report and now under development by Environmental Protection Agency (EPA), National Institute of Environmental Health Sciences, and National Institutes of Health (Collins et al., 2008) as well as researchers at the Hamner Institute (Thomas et al., 2007) have great promise in increasing the efficiency of selection. But again, as discussed above, the confidence in green chemistry decisions based on such methods will be dependent on demonstration of the relevance, reliability, and predictivity of the methods used to generate the “equivalent” hazard profile information. In an earlier commentary in this Forum series, our pharmaceutical colleagues MacDonald and Robertson (2009) noted the distinct challenge for the mostly in vitro testing paradigm proposed in the NRC vision is for it to adequately encompass the broad range of intracellular and intraorgan and interorgan pharmacokinetic and pharmacodynamic phenomena driving expression of whole-animal toxicity responses. These concerns are equally shared by environmental chemicals, and the many examples of such complexity in the toxicology literature are far too numerous to productively enumerate here. Thus, the expectation that any battery of in vitro technologies will be able to adequately replicate such biological and toxicological complexity is likely to prove an exceedingly difficult challenge. This is not to say, however, that marrying the new technologies with modified approaches to whole-animal toxicity testing, e.g., using these technologies to eliminate lifetime bioassays or other rodent toxicity studies, might be one of the major “turns in road” associated with implementation of the vision. Exactly where such future forks in road decisions will take toxicology is as yet uncertain but the end objective must be to develop toxicity evaluation programs to better inform decisions on potential adverse human health outcomes than conventional toxicity testing and risk assessment practices now in practice. Over the last several decades, the advent of government regulatory and industry product stewardship practices has significantly reduced the concentrations of many chemicals in the environment. For the most part, these efforts have increasingly distanced the doses used in conventional environmental chemical toxicity testing from human exposure-doses actually encountered in the real world. This growing disparity, and its importance to the future of toxicity testing, was noted 10 years ago by the Society of Toxicology Risk Assessment Task (Conolly et al., 1999). The Society of Toxicology Task Force stated: “The relevance of using doses that are many multiples of conceivable human exposures…is, at most, quite dubious. [and]… the predicted risks may have little or no relationship to risk in the real world.” A distinct advantage of the new technologies is their ability to rapidly and relatively efficiently explore a much wider-range of dose and response and particularly to explore the shape of the dose-response curve at doses/concentrations much more reflective of real-world human exposures to environmental chemicals. Examination of a wider, and particularly a lower, range of chemical dose will afford the opportunity to significantly increase the human relevance of many environmental chemical assessments. Interpretation of low-dose effects observed in in vitro or other high-throughput systems will be significantly facilitated if accompanied by contextual relationships to both no-effect and effect level doses reported in whole-animal toxicity tests. Although the advent of sophisticated physiologically based pharmacokinetic and other exposure-response models is significantly advancing these objectives, there remains an important opportunity to routinely assess internal dosimetry in animal bioassays of environmental chemicals. All too often existing toxicity studies of environmental chemicals express dose simply as external milligram per kilogram or parts per million when in fact the interpretive value of dose information really lies in what is delivered to the systemic circulation and target tissues. With the relatively recent advances in analytical sciences technologies, there is a distinct opportunity for capturing this vital information with little modification to existing toxicology studies (Saghir et al., 2006). In addition, when animal internal dosimetry findings are integrated with the rapidly expanding data emerging from human biomonitoring programs and further with dose-response information from high-throughput technologies, the collective body of information will be of much great value in informing future health risks. These types of opportunities have recently been illustrated in efforts to develop “Biomonitoring Equivalents” assessments of environmental chemical exposures. Such efforts, which bridge actual human exposures to doses in toxicity test systems and current regulatory exposure standards, will further define the strategic path for efficient selection of environmental chemicals deserving of more detailed health assessments (Hays et al., 2007; et al., can serve to exposure concentrations used in high-throughput and screening to concentrations in that to health risk assessment such as and 2009). Thus, there is little that application of high-throughput technologies to dose-response evaluations at exposures will to better informing the potential for adverse health outcomes in The future focus on toxicity evaluations at doses must be however, by to what an a concern of particular for environmental chemicals for which segments of the are already The NRC Vision the opportunity for emerging technologies to rapidly which in fact are and that toxicity must be within the toxicological community in to to environmental chemicals. experiences have strongly that if the future from in vitro or other high-throughput technologies are simply as toxicity an immediate is that such effects must be As well by many such dependent on may and health responses to chemical with as has been in an earlier commentary by and While some have to an effect in it is a A is a more as long as doses cause responses that within of the no adverse effects would be This would to even the most and to with there is a pressing need for more research using systems biology of toxicity to on the and of responses and the which such systems from the to an and at higher to a reflective of adverse of the human health risk assessments of environmental chemicals are based on from animal toxicity A of risk assessment is the scientifically of from animal toxicity studies and potential or to human health risks in a necessary understanding of the biology the toxicity responses, risk routinely use or to for the information. These can range from or multiples of to the that some chemical responses are with no The application of and the scientific and on their is often a cause of and in scientifically decisions on environmental chemicals of concern (National Research Council, Committee on Risk Approaches by the U.S. Board of Environmental Studies and Toxicology, Division of Earth and Life Studies, The new 21st Century technologies offer an exciting opportunity to test the scientific of many of the used As outlined in both the NRC Applications report (National Research Council, Committee on Applications of Toxicogenomic Technologies to Predictive Toxicology, Board of Environmental Studies and Toxicology, Board of Life Studies, Division of Earth and Life Studies, 2007b) and in a of Toxicology and report et al., those offer the promise of direct or of the used with biological and toxicological information. The toxicology community is already well on the road of how this science may impact the of key examples et have the potential value of to explore the shape of the dose-response curve for and particularly to more comprehensively test for the or of low-dose responses. opportunities to more the potential for agents to based a significantly risk evaluations of many environmental chemicals et al., 2007). The systems biology outlined in the NRC 21st Century Vision (Andersen and Krewski, National Research Council, Committee on Toxicity Testing and Assessment of Environmental Agents, Board of Environmental Studies and Toxicology, Institute for Laboratory Animal Research, an valuable opportunity to or replace with and based information. In an earlier commentary in this Forum series, MacDonald and Robertson (2009) noted that the new technologies are tools in rapid and more understanding of chemical of for These same opportunities to environmental chemicals and likely present one of the immediate and opportunities to with actual science in environmental chemical risk assessments. The high-throughput technologies have the ability to rapidly inform and other of issues current toxicology and risk A understanding of specific on toxicity expression by technologies has catalyzed development of improved animal which more closely human responses to environmental chemicals. of models specifically to chemical of now of particularly in the low-dose of environmental chemicals that could not be in Such efforts can provide important experimental insights into the impact of as of low-dose responses to environmental chemicals. of is one of how such in vivo models can significantly improve understanding of human risks associated with low-dose exposures to chemicals National Research Council, Committee on Applications of Toxicogenomic Technologies to Predictive Toxicology, Board of Environmental Studies and Toxicology, Board of Life Studies, Division of Earth and Life Studies, for that are to numbers of environmental chemicals, the of to the toxicity potential of particularly at environmental is a growing human health focus et al., 2007; National Research Council, Committee on Applications of Toxicogenomic Technologies to Predictive Toxicology, Board of Environmental Studies and Toxicology, Board of Life Studies, Division of Earth and Life Studies, et al., The of the new technologies to rapidly and comprehensively the full range of dose-response and associated of will likely provide insights into scientifically of and risk models applied to risk assessments. important perspective that must be into such however, is the fact that the exposures to chemicals from the of natural chemicals present in and other natural chemicals many if not all the toxicity properties of a key challenge (and will be to how the new technologies might better from and to how the of environmental chemical can be more from of natural chemicals 2007). The advent of the tools and technologies associated with 21st Century Vision a in a opportunity for toxicologists to a of a toxicity testing and risk assessment paradigm that has been of Although there is to this strategic vision into the mainstream of toxicity evaluations as rapidly as that must be with the understanding that those same many of toxicology experience and practice have a of that must be for effective implementation of the most the new technologies now afford opportunities to more the toxicological of the foundational of “The dose the and how this across the range of and With this as the the and turns of the road toxicologists can be that implementation of the NRC 21st Century Vision is in the