As described in earlier chapters, workshop participants discussed recent advances in knowledge related to the exposome and the spectrum of neurotoxicants to which humans are exposed; the mechanisms by which these exposures may impact the brain; and the broad array of neurodevelopmental and neurodegenerative disorders associated with neurotoxicant exposure. At the same time, individual workshop participants highlighted a number of research gaps.
Important unanswered questions regarding environmental exposures in human populations include identifying subpopulations at risk, their vulnerability at different points across the life span, and how genetic and environmental factors may be working together, said Jason Richardson. For example, in exploring the effects of air pollution in animal models, Deborah Cory-Slechta said her lab has begun to focus on specific developmental periods as opposed to lifetime or cross-developmental windows of exposures. She added that scientists are just beginning to investigate the transgenerational effects of lead exposure in mice. For example, she said, effects of lead exposure in the F-0 generation have been demonstrated even in the F-3 generation.
Brenda Eskenazi agreed on the importance of this issue, noting the need for more human studies as well. Different outcomes have been detected with different windows of exposure, she said, but more longitudinal studies are needed to assess exposures over the lifetime. Richardson said that answering these questions will require multidisciplinary research teams and the development of new technologies and methodologies to address polygenic and multiple environmental contributors as well as additional validated animal and cell-based models for various diseases.
Meanwhile, the coronavirus disease 2019 (COVID-19) pandemic has elevated the need to better understand how pathogen-driven infections interact with environmental toxicants, particularly as climate change and rising sea levels increase exposure to mosquito vectors in coastal areas, said Caleb Finch (Finch and Kulminski, 2020).
Other research gaps arise from deficits in availability and use of exposure monitoring tools, according to Trevor Penning, the Thelma Brown and Henry Charles Molinoff Professor of Pharmacology in the University of Pennsylvania Perelman School of Medicine. For example, he noted that disclosure to the Environmental Protection Agency’s (EPA’s) Toxic Release Inventory (TRI) Program is voluntary, unless a facility meets three specific criteria in which case they are required to report covered emissions.1 Proprietary chemicals in the TRI may not be identified, and industry can claim exemptions from using the TRI to disclose their emissions. In addition, he
1 For more information, see https://www.epa.gov/toxics-release-inventory-tri-program/basics-tri-reporting (accessed October 15, 2020).
said, there are EPA monitoring deserts in which no monitors are available to get air quality data.
With regard to mechanisms of toxicity, Helena Hogberg, deputy director of the Center for Alternatives to Animal Testing at the Johns Hopkins Bloomberg School of Public Health, said the knowledge gap is large. Little toxicology data exist for most chemicals on the market, and even less information is available regarding toxic effects on the nervous system. Gene–environmental interactions are also largely unexplored; more attention is needed to ensure the human relevance of mechanisms being modeled in labs, said Hogberg. Beate Ritz noted that she and colleagues from Harvard will soon begin a project to screen pesticides using induced pluripotent stem cells (iPSCs) from carriers of different genetic mutations linked to Parkinson’s disease (PD) in order to better understand gene–environment interactions.
Considerable evidence shows that developmental exposures to chemicals may be linked to developmental adult neurological disorders and brain aging, said Hogberg. She advocated using an adverse outcome pathway approach, looking at mechanisms in animal and cellular models and then trying to link those mechanisms to adverse outcomes such as the development of PD.
Epidemiology will be needed to definitively link compounds identified through in vitro screening to adverse human outcomes, said Mark Zylka. Animal models and in vitro studies are both important for identifying classes of chemicals that target certain molecular pathways implicated in disease, he said, but to narrow down the hundreds of chemicals to those that represent exposure threats to humans will require working with scientists who can assess levels in the environment. Epidemiologists can then look at a smaller subset of chemicals linked to biological pathways to see if humans are indeed exposed and if that exposure increases the relative risk for certain disorders.
Stanley Barone, deputy director of the risk assessment division of the Office of Pollution Prevention and Toxics in the Office of Chemical Safety and Pollution Prevention at EPA, agreed that putting data on biological processes such as proliferation, migration, differentiation, apoptosis, myelination, and synaptic plasticity into the context of adverse outcomes is essential to show their relevance to disease, morbidity, and mortality and support decision making based on data. He said this predictive toxicology approach can enable agencies such as EPA to bridge data gaps and make risk assessments in a chemically agnostic way.
J. Timothy Greenamyre added that genetics may provide hints that can help identify adverse outcome pathways. For example, the many genes that have been identified as causing PD can be grouped by mechanism (e.g., endolysosomal/autophagy functioning, proteostasis, inflammation, etc.). Chemicals and other environmental factors present in environmentally relevant concentrations can then be screened to assess impact on those pathways, he added.
Future research is also needed to identify factors that impact the susceptibility to potential adverse effects of environmental exposures such as air pollution, said Petkus. For example, he said, a recent study in Sweden showed that the adverse effect of exposure to air pollution on the development of dementia was larger in people with comorbid cardiovascular disease (Grande et al., 2020). Lifestyle factors may also interact with air pollution to increase or reduce the risk of adverse outcomes, said Petkus. Future studies of exposure effects should also examine associations with the accumulation of biomarkers of Alzheimer’s disease, including imaging biomarkers, he added.
National chemical, pesticide, and potential toxicant exposure data collected using standardized methods and driven by legislative efforts impermeable to change in leadership are essential, said Allison Willis, associate professor of neurology and epidemiology in the University of Pennsylvania Perelman School of Medicine. Richardson said some national data are available, including pesticide usage data from EPA. However, these data reflect primarily agricultural use, but not household use of pesticides, he said. Richardson also noted that the National Health and Nutrition Examination Survey (NHANES) has published a National Report on Environmental Exposures that includes urinary and blood metabolite data, and as described by Ritz in Chapter 2, these data can be used to assess population-exposure levels cross-sectionally for a limited number of agents while and GIS modeled data can be used for long-term exposure estimation, said Ritz. However, Tracey Woodruff noted that the NHANES data only cover about 300 chemicals out of the 2,000 to 4,000 chemicals that are highly used. Moreover, she said, regulatory policies and the proprietary nature of some of these data may limit access.
Woodruff added that while the Clean Air Act2 requires the monitoring of certain air pollutants that are then required to be reported to a national database, and thus providing a central access point, there is no requirement
for central reporting of water pollution data. California has geographically resolved water contamination data that is available on a statewide level, along with air pollutants and pesticide use, she said. Eva Feldman said monitoring with uniform standards has not been implemented in other states such as Michigan, where there is a high prevalence of amyotrophic lateral sclerosis (ALS). For ALS at least, some of the clusters appear to be associated with the use of well water, which may not be tested for presence of pollutants, added Feldman.
The environmental health science community is developing sophisticated tools to study the exposome and the totality of exposures over the course of a lifetime, said Richard Woychik. Meanwhile, neuroscientists bring to the table powerful imaging tools and other technologies and experimental approaches for studying neurodevelopment and function, and genome scientists bring new sequencing tools and genome analysis capabilities to better understand the complex traits involved in susceptibility to environmental exposure, he said.
Many new tools, or new uses of existing tools, were suggested to advance the field. For example, Jennifer McPartland advocated using a variety of methods, tools, and frameworks to define the universe of substances that affect the brain. That would require evaluating chemicals against known neurobiological targets, ensuring that knowledge about the etiology and pathology of neurological conditions is reflected in the chemical evaluation methods used to screen and characterize potential neurotoxicants, employing tools like exposomics to elucidate how different exposure patterns across populations relate to different neurological conditions, and considering these exposures in the context of other stressors. Other new technologies discussed included
- Machine-learning approaches: Andrew Petkus and others are using machine-learning approaches applied to high-dimensional neuroimaging data to examine associations between exposure and indexes of structural brain heath. Using imaging data from the Women’s Health Initiative Memory Study (WHIMS), they generated a disease pattern similarity score (AD-PS) that represents an individual’s neuroanatomical risk for AD (Casanova et al., 2018). Over a 5-year period, they showed that women exposed to higher amounts of PM2.5 had higher annual increases in the AD-PS score, which corresponded to approximately a 24 percent increase in dementia risk (Younan et al., 2020). Higher AD-PS scores were
- also associated with increased gray matter atrophy and declines in episodic memory. Petkus noted, however, that longitudinal studies are needed to validate this model.
- Three-dimensional cellular models: Hogberg described a new tool developed in her lab with funding from the National Center for Advancing Translational Sciences (NCATS)—a three-dimensional brain microphysiological system derived from iPSCs, composed of neurons, astrocytes, and oligodendrocytes (Pamies et al., 2017). Hogberg said microglia are not naturally present, but can be added to these “BrainSpheres” (Abreu et al., 2018). Her laboratory is using these BrainSpheres primarily for toxicology studies and is also partnering with neuroscientists to explore other uses. She noted that using iPSCs of different genetic backgrounds opens up multiple opportunities to explore gene–environmental interactions and link these findings with mechanistic studies.
- High-throughput screens: To help assess gene–environment interactions, Zylka advocated applying sophisticated new ways of conducting high-throughput sequencing-based screens.
- Gene targeting: Greenamyre added that manipulating gene expression with gene-targeting technologies such as antisense oligonucleotides, shRNAs, or viral vectors may also be helpful in teasing out regional vulnerabilities and the specificity of certain mechanisms. He noted that selective vulnerability is key to understanding neurodegenerative diseases.
- New uses for existing models: Greenamyre also suggested new ways of using existing animal models. For example, in the rotenone PD model, his lab has been studying the brain not only after parkinsonian symptoms develop, but also the quiescent period between exposure and subsequent development of symptoms to identify possible biomarkers of inevitable or preventable degeneration. He noted that initiating factors and factors that drive disease progression may be different; thus biomarkers and therapeutic targets may also differ across the continuum of the disease.
- Animal models more relevant to humans: Payne-Sturges added that animal models are needed that are more relevant to the human experience, particularly with regard to social stressor exposures. For example, she said restraint models are often used to mimic stress in animals, but may not be relevant to stress as it relates to living in poverty. Developing such models would require collaboration across disciplines, particularly with environmental health scientists interested in addressing issues around health disparities and cumulative risk, she said. Cory-Slechta said her lab has been working on a model of stress that reflects income inequality.
- Neurotoxicant monitoring badges: Walter Koroshetz, meanwhile, imagined tools that may be available in the future, such as a badge that measures organophosphate or pesticide exposures, akin to the badges that radiologists wear to alert them when their radiation exposure has exceeded a certain level. Building tools that capture exposure over time and that build a bridge to animal studies will require first the development of signatures of toxicant exposure in humans; an example is measurements in hair or the nasal mucosa, said Koroshetz.
Population-based cohorts from broad geographic regions with phenotypic and omics data are needed to advance understanding of the relationship between environmental exposures and human diseases, said Eva Feldman. Carl Hill, vice president of scientific engagement at the Alzheimer’s Association, suggested that these studies could also be used to explore what environmental factors are most important for which disproportionately affected populations, and why; for example, the role of cognitive reserve and resilience in resisting the effects of environmental toxicants. Eskenazi advocated for large investments in exposure sciences and environmental epidemiology similar to the investments that were made for genetics. Expanding exposure sciences so they can be applied on a large scale to large populations will be essential, she said.
Koroshetz suggested combining population studies that include genome-wide association study (GWAS) data with exposome data to better understand the interaction between genetics and environment. One of the challenges of this approach, Gary Miller noted, is developing statistical approaches appropriate for such complicated data. A concerted effort with advanced analytics would be needed to assess gene by environment in an unbiased way, he said. However, because both GWAS and exposure studies link directly to biological pathways, there may be an opportunity to “meet in the middle” to assess these connections, added Miller.
To better understand the real-world effects of environmental exposures, Koroshetz suggested analyzing what happens when a policy change affects use of a chemical. For example, as Woodruff described earlier, California banned the use of polybrominated diphenyl ethers (PBDEs) in 2003 and eventually banned all flame-retardant chemicals in the state. McPartland noted that because California is a large market, this ban had ripple effects across the country and, indeed, throughout the world. Cory-Slechta also noted that once lead was removed from paint and gasoline, studies showed that blood lead levels declined in every segment of the population in subsequent years. More recently, she said, studies have analyzed changes in
standardized student test scores since the removal of lead and shown positive outcomes (Reyes, 2012). However, Woodruff said investment has been limited in assessing how system-level changes such as this influence exposures and, moreover, it is difficult to link exposures to health effects because other risk factors can also contribute to those effects.
Cindy Lawler, acting chief of the Cellular, Organs, and Systems Pathobiology Branch in the Division of Extramural Research and Training at the National Institute of Environmental Health Sciences, said the institute is currently supporting a study to take advantage of a natural experiment in which China wanted to reduce air pollution in areas surrounding Beijing during the 2008 Olympics. The investigators identified women who were pregnant at the time in four urban districts for which data on the concentrations of various air pollutants were available. They showed that for babies whose eighth month of gestation occurred during the Olympics, decreased levels of air pollution were associated with higher birthweights (Rich et al., 2015). Other studies are examining whether this resulted in a decrease in conditions associated with low birthweight, including neurodevelopmental conditions such as autism, said Lawler.
Air quality changes have also been documented in large cities as a consequence of people sheltering in place during the COVID-19 pandemic. Woodruff said that the Environmental Influences on Child Health Outcomes (ECHO) study3 has pivoted to some COVID-related research to understand what kind of environmental risk factors may be changing during the pandemic, including not only a reduction in air pollution, but also changes in diet that could result in changes in exposure to chemicals such as phthalates. While assessment of neurodevelopmental outcomes would have to come later, she suggested this could be an excellent opportunity to collect biospecimens.
Eskenazi suggested that combining emerging geospatial methods with biosamples from existing cohort studies such as the Adolescent Brain Cognitive Development (ABCD) study4 could provide valuable and low-cost opportunities to examine environmental exposures. The estimation of exposures is now possible, she said, through satellite data and remote sensing methods. Comparing this information with data from cell phones and other body-worn sensors could enable estimation of real-time exposures in populations, she said. In addition, longitudinal and multigenerational studies that collect biosamples and exposure data during relevant periods are needed, particularly because new discoveries in epigenetics suggest multigenerational effects of environmental exposures, Eskenazi added. Residence-level monitoring of air quality and the use of wearable detectors
that monitor toxicants and microbes could also provide real-world data to better understand the exposome, said Finch. Incorporating data on social determinants of health and multigenerational life history would also be valuable, he said.
Cory-Slechta noted that many years ago, the National Research Council Committee on the Health Risks of Phthalates developed an algorithm for assessing the effect of exposure on a specific disorder: Start with a known disease or disorder and identify those chemicals that are known to individually influence this disorder (NRC, 2008). Because they may have different mechanisms that converge downstream within the same physiological system, statistical methods can be used to quantify the risks of the different combined exposures. The economic costs of addressing risk from these chemicals could also be calculated for a cost–benefit analysis.
McPartland noted, however, that new regulatory frameworks and policies will be needed to account for exposures to mixtures of chemicals as well as those mixtures alongside other stressors. Indeed, said Eskenazi, measuring one chemical at a time does not reflect real-life human exposures. Yet, she noted that there are statistical challenges that will need to be addressed to enable multifactor analyses of toxic exposures. Ritz suggested implementing “pesticidovigilance,” mirrored on the concept of “pharmacovigilance,” to ensure that regulatory standards are aligned with public policy interests.