Abstract
The International Academy of Toxicologic Pathology (IATP) Satellite Symposium on “New Approach Methodologies (NAMs) for Neurotoxicity Assessment and Regulatory Perspectives,” organized in Spain, addressed the growing need for improved assessment of neurotoxicity. Traditional neurotoxicity assessment using in vivo animal studies are impractical for testing the substantial number of environmental chemicals that currently lack data and in the early detection of neuro-related adverse reactions in drug discovery. The NAMs, including human in vitro assays and small model organisms, have been developed for faster and cost-effective assessment of neurotoxic potential. While NAMs offer improved practicality, utility, and valuable mechanistic insights, their integration into regulatory decision-making requires robust scientific validation and technical characterization. Confidence in and regulatory application of NAMs data can be supported by mapping cellular outcomes to neuropathological findings in mammals, including humans, through the Adverse Outcome Pathway (AOP) framework, and the Integrated Approach to Testing and Assessment (IATA). Case studies presented demonstrated the application of NAMs in chemical and drug safety evaluations, focusing on developmental neurotoxicity (DNT), Parkinson’s disease, and drug-induced seizures. In conjunction with in vivo toxicology studies, NAMs represent a significant step toward advancing chemical and drug toxicity assessment via hazard identification and drug screening safety assessments.
Keywords
Introduction
The International Academy of Toxicologic Pathology (IATP) is a professional scientific organization that establishes the criteria of excellence and accomplishments in the field of toxicologic pathology for acceptance of Associate Fellows and accreditation of Fellows worldwide. The IATP serves as a global source of experts in toxicologic pathology and organizes unique educational opportunities for toxicologic pathologists, scientists in related fields, and trainees. The IATP Satellite Symposium is typically a half-day session held during annual meetings of global toxicologic pathology societies, such as the European Society of Toxicology Pathology (ESTP) congress. In 2024, the annual symposium of the ESTP was held as part of the 5th Cutting Edge Pathology (CEP) Congress, and as a joint venture with sister societies, namely, the European College of Veterinary Pathologists (ECVP) and the European Society of Veterinary Pathology (ESVP) in San Lorenzo de El Escorial, Spain.
The focus of the 2024 IATP Satellite Symposium was “New Approach Methodologies (NAMs) for Neurotoxicity Assessment and Regulatory Perspectives,” to complement the theme of the ESTP meeting, which was Neuropathology: The Vast Pink Wonderland. The NAMs refer to technologies, strategies, or their combinations that evaluate chemical hazards and risks, serving as alternatives to replace, reduce, or refine (3R) the use of complex animal studies. 65 The 2024 IATP Satellite Symposium was specifically proposed as a platform to discuss the use and implementation of specific NAMs employed in neurotoxicity assessments to an audience primarily engaged in animal toxicology studies. The IATP fellows Drs Deepa Rao and Sibylle Groeters chaired the IATP Satellite Symposium which had four invited speakers: Professor Ellen Fritsche and Drs Helena Hogberg, Matthew Winter, and Andrea Terron.
The following narrative details their lectures on the use and implementation of specific NAMs in neurotoxicology, including chemical and drug screening initiatives for developmental neurotoxicity (DNT), Parkinson’s Disease (PD) and the assessment of chemically induced seizures. In addition, mapping cellular-level outcomes from NAMs to neuropathological findings in mammals, including humans, to build confidence in the translational value of this type of data were exemplified.
The Need for Improved Assessment of Development Neurotoxicity and Adult Neurotoxicity
Professor Fritsche (SCAHT—Swiss Centre for Applied Human Toxicology) and Dr Hogberg (NICEATM—National Toxicology Program Interagency Center for the Evaluation of Alternative Toxicological Methods) presented the need for a new framework in toxicology to assess DNT and adult neurotoxicity (ANT). In recent decades, there has been a notable increase in neurodevelopmental disorders among children.10,55 Today, 15% to 20% of children are diagnosed with a neurodevelopmental disorder such as autism spectrum disorder (ASD), attention-deficit/(hyperactivity) disorder (AD(H)D), and mental retardation 85 to name a few (Figure 1). Although the knowledge about these disorders have increased and they have become easier to diagnose, the increased prevalence is still not understood. Moreover, genetic causes as the only factor in the increased prevalence of these diseases seem unlikely, and evidence indicates that environmental/chemical exposures may contribute significantly to their etiology.35,36 Crucially, the majority of chemicals in our environment (~98.5%) have not been tested for DNT. 17 Current testing guidelines (OECD TG426 and 443)69,72 are based on traditional in vivo animal studies that are costly, time-consuming, and require large numbers of animals. Moreover, these guidelines are often accompanied by high uncertainties due to variability of data, making the interpretation and translation to human diseases difficult. 75 In order to provide timely and efficient information on the DNT potential of the ~100,000+ chemicals in widespread use, a new framework for DNT assessment using NAMs is needed.
Prevalence of neurodevelopmental disorders in 1990 and 2021, presenting the significant increase over time [adapted from IHME, Global Burden of Disease (2024) https://www.healthdata.org/research-analysis/gbd].
Similar to DNT testing, a limited amount of chemicals has been tested for ANT, with acute toxicity data (including acute neurotoxicity hazard) available for just 60% of the chemicals registered for commercial use. 45 Although acute neurotoxicity might lead to hospital admission due to its immediate and noticeable effects on the nervous system, low-dose, chronic exposures cause effects that are more subtle in appearance or might lead to diseases later in life are more challenging for the regulatory risk assessment of environmental/chemical exposures to assess. Consequently, with regard to increasing worldwide incidences of neurological and psychiatric illnesses including neurodegenerative diseases 62 and the suspected contribution of chemicals to such illnesses,27,87 current regulatory testing32,93 is far from sufficient for the protection of human health. Furthermore, the prevalence of neurological effects, such as seizure, as an adverse drug reaction in new medicines remains stubbornly high. 92 Although great progress has been made to develop in silico and in vitro screens for early stage seizure detection, this still remains a major cause of attrition during new drug development. 79 Importantly, many seizure-related failures occur relatively late in drug development and the gold standard approach for detecting seizures in nonclinical toxicology studies (rodent electroencephalography [EEG]) is highly invasive and resource-intensive. Consequently, the development and application of NAMs for better detecting seizure liability offer three advantages: (1) higher throughput that can be applied in earlier/discovery phases of drug development when the costs of failure are lower; (2) potential replacement of invasive EEGs in animals; and (3) identification of side effects earlier in development leading to minimized animal testing on candidate compounds destined to fail.
New Approach Methodologies to Assess Development Neurotoxicity and Adult Neurotoxicity
In response to the challenge with the in vivo DNT approach, significant collaborative efforts were undertaken by regulatory bodies, scientific communities, and stakeholders to develop an in vitro screening approach to enhance DNT assessment. 82 Starting in the early 2000s, several workshops organized by the Center for Alternatives to Animal Testing (CAAT), the Joint Research Centre (JRC) and US Environmental Protection Agency (EPA) brought these various stakeholders together to build the foundation for such a framework. 84 During these workshops, key processes that are important for brain development and cell systems capable of modeling such processes were identified. Subsequently, the European Food Safety Authority (EFSA) and the Organization for Economic Cooperation and Development (OECD) supported efforts that eventually led to the development of a DNT in vitro battery (IVB) designed for regulatory applications. An expert group on DNT was also established at the OECD that recently released a key guidance document titled “Initial Recommendations on Evaluation of Data from the Developmental Neurotoxicity (DNT) In Vitro Testing Battery.” This document provides essential insights into the implementation and interpretation of the DNT IVB. 71
The DNT IVB described in this guidance document consists of 17 assays that measure various key processes of neurodevelopment including human neural progenitor cell (hNPC) proliferation, apoptosis, cell migration, cell differentiation, neurite outgrowth, neurite maturation, synaptogenesis, and neural network formation and function (Table 1). These assays are made up of various 2D and 3D cell cultures from human and rodent cells, including cell lines, induced pluripotent stem cells (iPSCs), and primary cells. Due to the multicellular nature of some of the DNT IVB assays (3D neurosphere and rat primary cortical), cell-cell interactions are also captured in data from the battery.
Assays included in the initial OECD DNT-IVB.
Abbreviations: CNS, Central Nervous System; IUF, Leibniz Research Institute for Environmental medicine; LUHMES, Lund Human Mesencephalic; PNS, Peripheral Nervous System; UKN, University of Konstanz; US EPA, United States Environmental Protection Agency.
The use of the assays in the DNT IVB is based on the assumption that chemical perturbations in these key processes may result in adverse DNT outcomes at the organ and individual level. 53
In contrast to the DNT IVB, which embraces a multitude of potentially unknown mechanisms, ANT testing methods driven by well-described knowledge of underlying mechanisms may be preferred. Such a battery should be able to detect various human neurotoxic mode of actions (MoAs) ranging from acute to chronic neurotoxicity with an emphasis on neurodegenerative diseases.88,93 Here, one could envision, eg, a battery of assays assessing individual MoAs. A list of such assays currently embedded in the ToxCast screening platform was recently assembled 56 and addressed gaps in the ToxCast assay battery concerning neurotoxicity targets. Combining this list, 56 a neurotoxicity MoA list elaborated in Masjosthusmann et al, 57 mitochondrial toxicity mechanism (relevant to PD), 24 and Hallmarks and mechanism of neurodegeneration, 93 resulted in a comprehensive collection of important mechanisms that should be covered in an ANT IVB (Table 2). The MoA to induce DNT and ANT can be the same; the difference lies in the window of exposure (during development vs in adulthood), but in general, the developing brain is considered more susceptible to perturbations. 34 As such, the same NAM can be applicable to both DNT and ANT assessment but may need different experimental designs. For example, the neurite outgrowth assay UKN4 has been used both for DNT and Parkinsonian assessment but at different stages of differentiation. The DNT is assessed during early differentiation by measuring inhibition of neurite growth, while PD is assessed when the cell culture is more mature by measuring degeneration of neurites due to exposure. Similar, the network formation assay is exposed to chemicals during the development and maturation of the network to assess DNT effects, while the assay can be used to assess acute ANT if exposed when more matured.
Modes-of-Action (MoA) of adult neurotoxicity (ANT) that could be transferred to in vitro methods.
These are also key events in the adverse outcome pathway (AOP) on “Parkinsonian Motor Deficits” (AOP 3, AOP Wiki).
Instead of screening for many mechanisms or key processes individually, more complex, mixed cultured test systems containing different neuronal subtypes as well as different glia (ie, astrocytes, oligodendrocytes, and microglia) can be explored. Taking this path would entail a thorough characterization of test systems for individual neurotoxicity-specific MoAs to understand their applicability domain. Such an approach may not reveal the toxicity mechanism but could identify a compound of concern for further followed-up studies. An example of test systems that are very well suited for this are 3D BrainSpheres. These 3D specimens contain differentiated neurons and glia that can be grown in different media with distinct cell type compositions.39,54,73,74 Besides cellular endpoints, BrainSpheres can also reveal adverse effects on neuronal function, eg, by toxicity assessment using microelectrode arrays.39,42 While already characterized for a variety of neurotoxic MoAs, a comprehensive study on the BrainSpheres` biological/toxicological applicability domain is still missing. Moreover, the addition of microglia to this promising multicellular test system is pending and will be approached within two major European Commission projects (the Horizon Europe project CHIASMA 1 and the EFSA-funded project Brain Health). Indeed, immune-mediated toxicities, such as cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS), are increasingly recognized as critical safety issues in biotherapeutics.63,83 This likely requires more complex NAMs, incorporating also other types of immune cells such as T- and B-cells, and might be a critical research need in the future. 41
Small Model Organism Behavior and Functional Brain Imaging Endpoints as a Complement to In Vitro Assays
The use of in vitro assays can provide timely data for decision-making and to prioritize compounds with potential for DNT and ANT. Ultimately, however, whole organismal assessments of neurotoxicity including inferential measures such as behavior (eg, locomotor activity) are considered critical for assessing neurotoxicity risk, and these are not possible using in vitro or in silico approaches. Small model organisms including early life stage vertebrates such as zebrafish embryos, or invertebrates such as planarians are considered NAMs and have shown potential as translational models of neurological effects in a whole animal context, including for DNT, neurodegeneration, and seizure.19,86
Whole organismal NAM approaches employing automated assessment of behavioral endpoints are generally considered higher throughput and lower-cost alternatives to traditional in vivo rodent studies. In terms of translational value to mammals, including humans, vertebrate species offer the most potential. Genomic conservation among these species, such as the zebrafish (Danio rerio), is relatively high, eg, around 86% of human drug targets have identified orthologues in zebrafish. 37 With respect to neurotoxicity specifically, there is also broad conservation of CNS structure and function between zebrafish and mammals. 78 For example, despite the absence of a cerebral cortex, zebrafish are able to perform a range of executive functions, have well-conserved neurochemistry and are responsive to pharmacological modulation of all major neurotransmitter systems. 49 Furthermore, behavioral assays in zebrafish appear comparable at a functional level with mammalian behavior with links to neural circuitry underlying the basic form of behavioral regulation. 64 The most common behavior assay for DNT assessment is the locomotor light-dark transition assay, where various parameters of the swimming behavior of embryo-larval zebrafish are measured. 77 The utility of zebrafish for toxicity testing is expanding across the globe with hopes that it can serve as a chemical screening tool, predict adverse health outcomes, or even replace rodent models used for risk assessment. Unlike other in vivo systems such as rodents, zebrafish reproduce quickly with large clutches and can be directly exposed to many toxicants via immersion in treated water. Zebrafish also provide added value compared with in vitro assays, eg, as a comparatively metabolically competent organism with the ability to address potential ADME (absorption, distribution, metabolism, and excretion) issues. 16 The dopaminergic (DA) neuronal network of the central nervous system in particular has been extensively characterized in both early life stage and adult zebrafish and shows several similarities with mammalian species.48,89 Various tools for visualizing DA neurons have been employed, and these allow researchers to study the development of the nervous system and model disease states such as PD. 98
Planarians are another small model organism group suitable for DNT/ANT which provide comparable advantages to zebrafish as they are small, develop quickly, and display a wide range of behaviors. The genome of the planarian is less well studied than that of vertebrates, such as zebrafish, but it has been shown to exhibit high homology among neuronal genes with humans. 59 In some contexts, however, behavioral assessment of neurotoxicological outcomes in NAMs is less appropriate for the prediction of comparable outcomes in mammals. For example, complex behaviors in zebrafish such as social interactions 29 are not displayed until later in development (eg, after the point of protection in Europe), and behavioral convulsions can be ambiguous depending on the initiating mechanism of action and neural circuits affected. 22 Such considerations have led to the development of approaches in which more direct measures of the CNS response to chemical treatment can be undertaken in early life stage of zebrafish, when they are still considered NAMs. One such approach utilizes recent advances in functional neuroimaging, in which genetically encoded Ca2+ sensitive fluorescent dyes combined with high-speed fluorescence microscopy allow the assessment of whole brain functional responses to stimulation (reviewed by Vanwalleghem et al 91 ). This approach is beginning to be used as a tool for the screening of neurotoxic chemical effects including for the identification of seizure liability amongst new drugs,94,96 which is the focus of case study 3 presented below.
Building Confidence in Using New Approach Methodologies Data to Characterize Hazard and Risk
Although NAM data can provide valuable mechanistic information on a chemical’s potential to be hazardous, their use for regulatory decisions requires high confidence in said data. Traditionally, this was achieved through a lengthy and rigid validation process (OECD TG34). 70 Recently, the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) published a report on Validation, Qualification, and Regulatory Acceptance of NAMs, that recommends a more flexible approach to help test developers, end users, and regulatory agencies build confidence in using NAMs. 65 Revision of the OECD TG34 is also ongoing, and harmonization with the ICCVAM report is taken into consideration. The report includes key concepts such as the intended context of use, human relevance, technical characterization, data integrity, information transparency, and independent review. The application of some of these key concepts to DNT and ANT is described below.
Scientific Validation and Technical Characterization of New Approach Methodologies for Assessing Neurotoxicity
Ensuring relevance, reproducibility, and predictivity of NAMs is a prerequisite for regulatory trust and hence the use of NAMs in a regulatory context. 18 Therefore, NAMs for DNT and ANT have been evaluated thoroughly for their biological relevance, robustness, and ability to predict reference compounds with known DNT or ANT potential. Here, extensive international contributions of the last >20 years to move DNT NAMs into regulatory acceptance are the reason for the advanced status of DNT. Although a similar effort was not undertaken for ANT, the existence of chemicals known to induce Parkinsonian syndrome in human (MPTP/MPP+), or to reproduce hallmarks of the Parkinsonian syndrome, is expected to facilitate the regulatory implementation of the test battery already available, and for which experimental evidence exists (see case study 2).
Biological relevance aims to demonstrate similarities between the biology and physiology of the test system with the in vivo situation. Here, cell morphology, marker expression, cell function, responses to physiological cues as well as disturbance of biology due to, eg, toxins, are evaluated. In the specific case of the DNT IVB, NAMs were a priori developed to cover critical key neurodevelopmental processes that are indispensable for normal human brain development. 4 However, within numerous publications of the respective test method developers at the Leibniz Research Institute for Environmental medicine (IUF), the University of Konstanz and the US EPA, the biology of the DNT IVB has been thoroughly described.20,51 For ANT, there are so far no systematic assessments of biological plausibility for certain test systems. However, multiple cellular systems exist that allow measuring a broad variety of mechanisms concerning ANT. A systematic update of the literature after a previous inventory assembled in 2018 57 would be helpful.
Technical characterization of a test method must be shown on different levels (Figure 2). To start with, a test method with necessary controls needs to be established. Crofton et al 21 published a very useful instruction on how to establish a test method for DNT that will suffice regulatory use. This instruction is not specific for DNT but also useful for other domains. Once the test method is established, reproducibility can be shown intraexperimentally by assessing intraplate variability, interexperimentally by calculating variance of assay-specific controls across different experiments. Within the laboratory (intralaboratory) can be assessed by observing reproducibility between different experimenters within the same laboratory and between laboratories by performing a lab-to-lab transferability (interlaboratory) study. Assays of the DNT IVB have been evaluated on all levels up to the interlaboratory reproducibility and are documented in the DNT IVB ToxTemps. 71 The lab-to-lab transferability of all 17 assays from all test method developers’ labs to the contract research organization (CRO) DNTOX GmbH is currently taking place. The method transfer to a CRO bears the advantage that not only the transferability of the methods into a naïve lab is shown but also their availability for end users is guaranteed at the same time.
Gaining confidence in DNT IVB assays: Biological Relevance, robustness, and predictivity of positive and negative reference compounds were shown. Images produced with BioRender; the “Predictivity” image is adapted from Blum et al. 9
For ANT, formal reproducibility has not been systematically followed as there is no agreed-on ANT IVB yet. However, test systems for studying brain function and disease are manifold (Figure 3) and are used by the large community of neuroscientists for a broad variety of neurological research questions. Standardizing such test systems for ANT regulatory use is the first step. As an example, significant progress has been made for the 3D BrainSpheres. BrainSpheres can be produced in different laboratories from different human-induced pluripotent stem cell (hiPSC) lines with different media. However, depending on the protocol, the BrainSphere outcomes will be different with regard to cell-type composition and function.39,73 Intra- and inter-experimental variability for different endpoints studied in BrainSpheres in different laboratories has been shown, and mainly myelin15,73,80 and neuronal subtype-specific activity39,42 has been the focus. Recently, a case study using zebrafish behavior and BrainSpheres as readouts identified the GABAA-receptor as a target for chlorophen-induced neurotoxicity. 54 Likewise, LUHMES cells have been widely used for investigating cellular effects relevant for the pathophysiological process of developing Parkinsonian motor deficits24,25 (see case study 2 below). Understanding the presence and function of targets relevant for ANT MoAs in such test systems will greatly increase trust in using such microphysiological systems for regulatory purposes.
Human-induced pluripotent stem cell (hiPSC) derived neural cultures: hiPSC can differentiate in 2D and 3D into human-induced neural progenitor cells (hiNPC). Using different media and protocols, hiNPC can be further differentiated and matured into spontaneous mixed 2D or 3D cultures including organoids, which add anatomical structures to the 3D BrainSpheres. Alternatively, genetically or pharmacologically driven iNeurons and iAstrocytes can be produced, which can be utilized as single cultures or can subsequently be placed in certain ratios into defined mixed cultures. Finally, the goal is to also add microglia to the neuroectodermal in vitro systems for including immune cell function into the mixed neural cultures. Images produced with BioRender.
The NAM predictivity is not necessarily an easy task when it comes to DNT, as the number of human reference compounds is relatively small. Two workshops in the past have produced chemical lists for defining DNT reference compounds.3,60 These reference compounds were used for assessing sensitivity and specificity of the DNT IVB. 20 Current chemical testing in the DNT IVB increases the data base for DNT effects of compounds and will further aid in gaining confidence in data produced with these novel methods. Such analyses of test accuracy have not been performed for ANT yet.
Recently, van der Zalm et al 90 published a framework on how to establish scientific confidence in NAMs for regulatory application. This multistakeholder guidance points out that NAMs have to be fit-for-purpose, need to be of human relevance, have to be thoroughly technically characterized, and need to have undergone independent review. Data also need to be of integrity and transparent. This was published after the data basis for the initial OECD recommendations for the DNT IVB 71 had already been produced, yet indicated that the homework for the DNT IVB was done properly.
Linkage Between New Approach Methodologies and Human Diseases
Professor Fritsche presented another way of gaining trust in NAMs, by studying the presence and functionality of molecular pathways, in respective NAMs, that are contributing to human disease. For example, mutations in the WDR81 gene lead to severe microcephaly and disturbed gyrification in humans. 14 Mechanistic studies revealed that mutated WDR81 alters the functionality and signaling pathway of the epidermal growth factor receptor (EGFR), which leads to inhibited NPC proliferation that is responsible for the observed microcephaly. 11 The EGFR is functional in some DNT IVB assays, as its activation and inhibition stimulate and reduce NPC proliferation and migration, respectively.51,57 Hence, a compound interfering with this pathway will be detected by the NPC proliferation and migration assays and can potentially be associated with human disease. More studies concerning other pathways are currently being performed for increasing trust in the relevance of the DNT IVB assays for human protection.
Another possibility as a proof-of-concept for studying human relevance of neural in vitro models is the generation of cultures from patient cells. Here, one example is the generation of hiPSC that can be used for further differentiation into mixed neuronal and glia cultures, like the 3D BrainSpheres. A BrainSphere model of the rare disease Cockayne Syndrom B (CSB) modeled clinical neurological phenotypes of this human disease, while the respective mouse model did not. 47 Both examples demonstrate the value of considering disease pathways in NAMs for gaining trust in their protective abilities for human health.
The Adverse Outcome Pathway Framework and Integrated Approach to Testing and Assessment
Another approach to illustrate biological relevance is the adverse outcome pathway (AOP) framework. This framework was developed to address the need of the regulatory and scientific community to organize and streamline existing mechanistic knowledge and link to adverse outcomes (AOs) of regulatory significance. 2 An AOP starts from a molecular initiating event (MIE), leading to subsequent key events (KEs) and finally an AO, eg, impaired learning and memory. However, due to the complexity of the endpoint, few putative and reviewed AOPs exist for DNT. 76 Combining the current endorsed DNT AOPs leading to impaired learning and memory into a network revealed that KEs in the network align with the neurodevelopmental processes included in the DNT IVB, therefore increasing the confidence in human relevance (Figure 4). The advantage of such key processes or KEs measurements lies in the integration of a large variety of cell type-specific MIEs within the individual assays, allowing resource-effective compound screening aiding the fact that knowledge on DNT MIEs is currently sparse. However, data generated using the DNT IVB can also further assist AOP development and improve understanding of mechanisms underlying DNT.
Examples of molecular initiating events (MIEs) followed by Key Events (KEs) that can be measured by the DNT IVB (yellow and orange box), leading to DNT adverse outcomes such as impaired cognition.
The AOP framework also supports the Integrated Approach to Testing and Assessment (IATA) that can promote the use of NAM data in regulatory decisions as it combines data from multiple sources, including mechanistic information and toxicological pathology, to conclude on the toxicity of chemicals. 82 While it is currently not envisioned that NAMs will be a direct replacement for in vivo Test Guidelines, there are several regulatory relevant scenarios for which data from these assays could be applied to inform decision-making. In an appendix to the published OECD DNT guidance document, 76 several IATA case studies have been published that exemplify various regulatory needs, including weight of evidence for hazard assessment, 67 waiving of a DNT in vivo guideline study, 28 and prioritization for further testing.12,68
The speakers in the scientific session illustrated three case studies where NAMs were applied for chemical and drug safety assessment for DNT (case study 1), PD (case study 2), and drug-induced seizures (case study 3).
Chemical Screening Efforts for Development Neurotoxicity and Adult Neurotoxicity Using New Approach Methodologies
The DNT Health Effects Innovation (HEI) program within the Division of Translational Toxicology (DTT) at the National Institute of Environmental Health Sciences (NIEHS) was developed in 2019, with the aim to apply an integrated testing strategy to evaluate hazard of chemical exposure to the developing nervous system (https://www.niehs.nih.gov/research/atniehs/dtt/strategic-plan/health/developmental). One of the objectives of the program is to generate screening-level information using NAMs as an interim means to evaluate the hazard and prioritize compounds for further evaluation. This screening effort is conducted in the majority of the assays in the OECD DNT IVB (Table 1) and in zebrafish embryo and planarian behavior assays. Compounds with concern to induce DNT were nominated and justified by various stakeholders. The DNT HEI program selected and distributed chemicals based on specific criteria including known human exposure, DNT in vivo guideline study performed, incomplete data from the DNT IVB, nomination from multiple stakeholders, or compounds of high priority for NIEHS. For practical reasons in terms of time, cost, and data analysis the compounds were divided into three testing phases (https://www.niehs.nih.gov/research/atniehs/dtt/strategic-plan/health/developmental/chemical-list). The first testing phase of 115 chemicals included mainly chemicals with high concern for DNT such as pesticides and industrial compounds. Testing of phase 1 is finalized and shows high activity in the DNT-IVB, especially among the pesticides (unpublished data). Testing in the second phase (108 chemicals), also included DNT negative compounds, cannabinoids, and mixtures, is foreseen to be finalized in early 2025. Selection of 100+ chemicals for phase 3 testing is currently ongoing and will be distributed to contracting labs in early 2025.
In Europe, EFSA has supported chemical testing in the DNT IVB 20 and recently awarded contracts for additional testing of pesticides with high importance to the European Union. The acceleration of testing in the DNT IVB is important to understand the applicability domain of the DNT IVB and to identify uncertainties. Moreover, EFSA support interlaboratory transferability of the DNT IVB assays, compare in vitro results to in vivo guideline findings, investigate the use of the DNT IVB to increase knowledge on human brain health and disease, and enhance development of AOPs for DNT.
In the field of ANT, the assay methods implemented to describe the fully OECD endorsed AOP 3 (inhibition of mitochondrial complex I leading to Parkinsonian syndrome-related motor dysfunction, https://aopwiki.org/aops/3) were used to fine tune test methods and screen several chemicals affecting the mitochondrial electron transport chain. An IATA case study was consequently developed by EFSA1,40 followed by a new EFSA project where many European authorized pesticides will be tested in these methods. This project will be implemented in a screening and testing phase, followed by the development of selected IATA case studies using the already existing examples. This project will be complemented by a parallel activity intended to close the gap between the KE tested in the in vitro assays and the AO of AOP 3 in the zebrafish as a whole organism model.
Case Study 1: Integrated Approach to Testing and Assessment Case Study to Prioritize Compounds for Further Development Neurotoxicity Studies
Dr Hogberg presented a case study that used NAMs to prioritize a class of compounds for further DNT assessments. Organophosphorus flame retardants (OPFRs) are a class of compounds that consists of 20 to 50 chemicals that have replaced phased-out brominated flame retardants (BFRs). 61 The OPFRs are often present as commercial and isomeric mixtures, and a continual rise in their use has led to increased human exposure. 8 Furthermore, there is concern that these chemicals may induce DNT effects as they resemble the structure of organophosphorus pesticides that are well known to be DNT/ANT.13,58 Despite this, OPFRs currently lack DNT information, and it is not feasible or practical to run in vivo guideline studies on all members of this class. Therefore, an IATA case study was developed to demonstrate how a battery of in vitro and complementary non-mammalian animal models (zebrafish and planarians) could be used to prioritize a class of compounds for further testing. Herein, the relative toxicity of OPFRs was compared with some of the phased-out BFRs. 52
An IATA always starts with a problem formulation (1), in this case study, the aim was to select high priority compounds within a class of OPFRs for further DNT hazard evaluation (Figure 5). Existing information was gathered from the literature and the Integrated Chemical Environment (ICE) (https://ice.ntp.niehs.nih.gov) (2). The ICE is a database that was developed by NICEATM, NIEHS to support the development, evaluation, and application of NAMs,7,23 and includes curated high-throughput screening (cHTS) data sets from the US federal Tox21 collaboration and EPA’s ToxCast program for ~10 000 chemicals. Data in ICE are annotated using controlled vocabulary terminology from the Open Biological and Biomedical Ontology (OBO) Foundry (http://obofoundry.org/) for toxicity endpoints and mechanisms.
Workflow of IATA case study for prioritization of compounds for further DNT testing by integrating data from various sources (yellow boxes). These sources produce different types of data (purple) that going through data analysis pipelines (green). The final output data (orange) can then be combined to conclude on the toxicity of compounds and select high-priority examples for further testing.
A weight of evidence assessment was performed to evaluate if the information was adequate to select compounds for prioritization (3). The gathered data from various sources provided limited information on DNT effects for the OPFRs. Therefore, the decision was to generate additional screening data using DNT NAMs (4).
Overall, the replacement OPFRs appear to have comparable activity in the DNT NAM battery to the phased-out BFRs. Zebrafish embryo behavior assays were the most sensitive in the battery, followed by differentiation of oligodendrocytes. Integration of data from ICE and the literature identified other sensitive endpoints, such as endocrine disruption and effects on astrocytes and microglia populations that are currently not measured in the battery. These results indicated that the DNT IVB should be extended to include such endpoints to decrease the uncertainty in using NAMs for regulatory decisions on DNT.
After the integration of the analyzed data from the various sources, physiological-based pharmacokinetic (PBPK) and in vitro to in vivo extrapolation (IVIVE) modeling was performed. The activity concentrations from the in vitro assays overlapped with predicted human exposure, calculated from biomonitoring data in the literature (2), for some OPFRs, indicating potential concern for human health. 52 Moreover, the in vitro activity was within the order of magnitude of the in vivo point of departure (PoD) in regulatory studies for few compounds where such data were available. This demonstrates and builds confidence that data from the DNT IVB provide similar information as in vivo studies. Still, additional information for some compounds was desired to understand the translation of the NAMs data to DNT AOs (5).
Based on the IATA case study, DTT selected two OPFRs to be further evaluated for DNT using tailored in vivo studies that are still ongoing (6). 97 In conclusion, integration of data from multiple sources can reduce uncertainties and provide more mechanistic understanding of chemicals potential to induce DNT.
Case Study 2: Adverse Outcome Pathway Informed Integrated Approach to Testing and Assessment Case Study for Parkinson’s Disease Hazard Assessment
Dr Terron from EFSA presented a case study using NAMs for PD hazard assessment. The PD is a human chronic progressive neurodegenerative disorder with a higher prevalence in the aged male population. Loss of specific DA neuronal populations leads to well-characterized clinical symptoms, which include slowness of movement, resting tremor, rigidity, and disturbances in balance. When Parkinsonism is the prominent part of the disorder, these are referred to as “Parkinsonian disorders” and include PD. 26 However, a primary pathology is common to all Parkinsonian disorders and consists of a selective degeneration of DA neurons in the substantia nigra pars compacta (SNpc), mainly projecting to the striatum, development of cytoplasmic, protein-rich inclusions, called Lewy bodies (LB), and decreased levels of striatal dopamine, which is eventually the cause of the clinical symptoms. Although the precise molecular etiology of the disease is unknown, it is most likely caused by a complex interplay of genetic and environmental factors with multiple interacting pathways. 33 Some cases may have a clear genetic cause, while others can be caused by effects of toxins and/or a gene-environment interaction.
The role of pesticides as potential environmental risk factors for PD has long been suspected and recurrent through multiple epidemiological meta-analyses, although the specific causative agents and the mechanisms underlying the disease are not fully understood. 31 The capacity of chemicals to damage this specific neuronal subpopulation in humans is well documented: several poisonings with the chemical 1-methyl-4-phenyl-tetrahydropyridine (MPTP) occurred during the 1980s and have been found to result in severe Parkinsonian motor deficits due to the loss of nigrostriatal DA neurons. The target of the bioactive MPTP metabolite 1-methyl-4-phenylpyridinium (MPP+) is the complex I (cI) of the mitochondrial electron transfer chain (ETC) and the pesticide rotenone has the same binding site on cI as MPP+. It has also been shown to cause nigrostriatal DA neuron loss associated with Parkinsonian motor deficits in rats.
Because different pesticides induce mitochondrial dysfunction and oxidative stress as part of their pesticide MoA, it may be assumed that chemicals that inhibit cI of DA nigrostriatal neurons have a propensity/risk factor to cause Parkinsonian motor deficits. However, the assessment of these chemicals remains challenging because regulatory toxicology studies are not really designed to understand relevant mechanisms of toxicity or risk factors. Furthermore, regulatory tests can be of limited sensitivity when hazards are likely the consequence of long-term/chronic diseases, low-dose continuous exposure to toxicants, or when multiple toxicants are interacting on the same AO through different MIEs. With the aim of reducing this uncertainty, EFSA started a project based on a scientific and regulatory problem formulation, questioning whether understanding a mechanism of pathogenesis could provide a plausible mechanistic link between pesticide exposure and an AO relevant to PD. This resulted in the development of the OECD-endorsed AOP 3 that provides a theoretical basis explaining the link between cI inhibition and neurotoxic AOs (manifesting in loss of DA neurons) (Figure 6).
AOP 3: Inhibition of the mitochondrial complex I, II, or III of nigrostriatal neurons leads to Parkinsonian motor deficits, as implemented in Delp et al. 24
This case study is a test/pilot case for exploring the use of AOP informed IATA in risk assessment and implement method description for AOPs. The basis of this approach is to use the AOP 3 to define a relevant PoD for risk assessment using a defined testing battery (Figure 6) in the LUHMES cell line. These cells combine the advantages of conditional immortalization (unlimited supply of a homogenous population of cells) with the option of their differentiation into fully postmitotic cells that represent cardinal features of nigrostriatal DA neurons. This approach allows for screening a large number of chemicals for selected MIEs 24 and the ability to convert the defined PoD by an IVIVE to a threshold dose and to use this to define the margin of internal exposure and the setting of regulatory threshold values.
Subsequent work explored whether other mitochondrial inhibitors might also show similar effects, and an extensive case study was performed on tebufenpyrad.1,40
A critical step in this case study intended to use mechanistic data for public health protection was to resolve the “scaling problem,” to develop an aggregated PBPK model for brain target, and to identify an internal margin of exposure as summarized in Figure 7.
Tebufenpyrad case study integrating NAM data, exposure, and PBPK modeling to predict a margin of internal exposure (MoIE).
Based on the obtained MoIE (margin of internal exposure), it was not possible to exclude an inhibition of cI in human and the occurrence of neurological AOs given the intended uses assessed. The PBPK model and the low MoIE obtained tend to show that professionals (operators and workers) are likely to be the most exposed population. It is, however, important to acknowledge that this internal exposure depends on uncertainties in existing data on the chemical used in the case study, uncertainties related to the PBPK model, and those related to the external exposure estimates. The case study is, therefore, representing a model on how to use NAM-based mechanistic data in the regulatory risk assessment process, define best practice in the process, and identify critical uncertainties in the AOP, in the PBPK model, and external exposure estimate.
Case Study 3: Larval Zebrafish as an Alternative In Vivo Model for Assessing Seizure Liability—From Behavior to Functional Brain Imaging
Dr Winter presented a case study that used the larval zebrafish as an NAM to identify seizure liability as an adverse reaction during the drug discovery and development process. The presentation focused on the initial development of a behavioral screening approach to the application of the latest functional imaging approaches, with the ultimate aim of differentiating between “normal” and seizure-like responses to drugs.
Seizures are characterized by periods of rhythmic, synchronized, abnormal neural activity that, depending on which brain regions are affected, can result in a range of adverse health effects. 30 Seizures can be the result of disease, injury, or as an unwanted side effect of drugs. Importantly, seizures are electrographically consistent events with a high degree of conservation across diverse taxa, including lower vertebrates such as fish. 43 Recurrent unprovoked seizures define the condition epilepsy, which occurs as a standalone condition or in association with other disorders including Alzheimer’s disease, autism, and Down’s syndrome. 43 Seizures (and epilepsy) are complex neurological disorders and as such the use of animal models are critical for understanding their causes and developing new treatments. 46
As previously stated, seizures can also occur as a side effect of new drugs, with an estimated 10% to 20% of drugs in development failing to reach patients due to this “liability.” 79 During drug safety assessment, the gold standard approach for detecting seizure liability is rodent encephalography (EEG). The EEGs in rats or mouse, however, are slow, expensive, and highly invasive and are, therefore, unsuitable for early detection of seizure liability as well as being economically and ethically problematic. As well as the procedure itself being invasive, an inability to deploy EEGs early in discovery means that many seizure-related failures occur at a relatively late stage. The impact of this is twofold, first, vast resources have often already been invested, and second, many animals may have been used in other preclinical studies, supporting the progression of a new drug that never reaches the clinic. Consequently, over recent years, the zebrafish has emerged as a credible alternative model to study seizures and epilepsy, showing conservation of basic brain structure and neurochemistry, as well as remarkably similar electrographic features compared with mammals. 5 Importantly for early stage safety assessment, the larval life stage (typically <7 days postfertilization [dpf]) offers higher throughput amenability, meaning it can be applied earlier in preclinical development than traditional rodent-based approaches. Furthermore, in its preindependent feeding form (<5 dpf), embryo-larval zebrafish are considered an NAM.
The first studies employing larval zebrafish in seizure liability assessment focused on convulsive behavior as an indicator of seizure-potential. 95 Although promising for identifying pharmacology that results in hyperlocomotive behavior, these approaches generally lack the sensitivity to identify mechanisms associated with more subtle non-hyper-locomotive phenotypes. To counter these limitations, we developed a functional imaging-based approach, using a non-protected life stage of a transgenic zebrafish (4dpf—preindependent feeding) that possesses a pan-neuronal genetically encoded fluorescent Ca2+ sensor (GCaMP6s), in which we can directly observe changes in neural activation or suppression across the whole brain at a near cellular-level resolution. 96 When combined with high-speed, high-resolution, light sheet fluorescence microscopy, this approach allows us to image across a whole brain volume in around 1.5 seconds. Images are processed using a custom Python image analysis pipeline that registers image sequences against a standardized anatomical brain atlas in which around 50 brain regions are spatially defined in 3D. The result is a neural activity “fingerprint” reflecting activation or suppression of neural circuits associated with the mechanism of action of the chemical to which the larvae are exposed. 96
Using this approach, 65 compounds comprising representative modulators of pharmacological targets associated with the induction of seizures in mammals were assessed, alongside several compounds with no known seizuregenic propensity (negative controls). From these data, we were able to identify mechanism-specific patterns of brain activity, and insights into anatomical regions that appear important for seizure initiation and progression, such as the cerebellum. 94 Multivariate analysis of the functional connectivity between brain regions (defined as the correlation between the temporal activity profile of pairs of regions) also revealed that exposure to many seizuregenic compounds was associated with abnormal synchronization, which is considered a key feature of epileptogenic networks. 44 The approach offers the potential to detect seizure liability and the mechanisms driving such seizure, as part of a preclinical safety screening strategy. Moreover, by providing detailed information on which brain regions are affected by specific drug treatments, this approach also has the potential to provide novel data on how seizure events are generated and propagated across the brain, and how localized neuronal activation can lead to the network-wide hyperexcitation typical of a full-blown seizure. This is important as most current animal models for seizure and epilepsy research rely upon the application of seizuregenic drugs to study epileptogenic network development, as well as for assessing the efficacy of new antiepileptic drugs. However, they do not offer the levels of whole brain coverage or spatiotemporal resolution that our model can. Furthermore, this approach is highly relevant for other branches of neurotoxicology and neuropharmacology outside the study of seizures, eg, we are already applying this technique to the study of the impact of pharmaceuticals in the environment and the efficacy and tolerability of anesthetics in fish.
Discussion
A short panel discussion was held following the lectures moderated by the Chairs Drs Rao and Groeters. Questions from the audience included the relevance of translation of in vitro and in silico endpoints to whole organisms. Specific discussions included translation of data specifically for neurotoxicity assessments, where the complexity of the various components of the nervous system may not be captured in NAM methodologies. For example, the brain of zebrafish may not reflect the complexity of mammalian systems with reference to neuroanatomy, neurophysiology, and/or neurochemistry. Even though the complexity in such methodologies remains a limitation for pivotal go/no-go decisions on the safety evaluation of candidate compounds, it was agreed that positive data could still be useful in screening compounds for potential neurotoxicity in the early phases of drug discovery and development. For chemicals, where the numbers of compounds without any neurotoxicity information is staggering, NAM approaches still allow affordable screening of potential hazard identification of large numbers of compounds where alternatives are meager. Overall, because of the spatial and temporal complexity of the nervous system on many levels, there was general consensus that multiple endpoints are needed to assess potential neurotoxicity and that NAMs remain a critical component in that matrix of neurotoxicity assessment.
Footnotes
Acknowledgements
The authors thank IATP and ESTP for funding to host this session.
Author Contributions
Conceptualization, S.G., R.C.K. and D.B.R.; Investigation, H.T.H., E.F., A.T. and M.J.F; Methodology, H.T.H., E.F., A.T. and M.J.W.; Visualization, H.T.H., E.F., A.T. and M.J.W.; Writing—original draft, H.T.H., E.F., A.T. and M.J.W.; Writing—review and editing, S.G., R.C.K. and D.B.R. All authors have read and agreed to the published version of the manuscript.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Professor EF is shareholders of DNTOX GmbH, offering neurotoxicity testing services. The other authors declared no real, perceived, or potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work presented by HTH was supported by the Division of Translational Toxicology, National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services ZIA ES103387-02 and under Contract No. HHSN273201500010C. EF receives funding from the Horizon Europe project PARC (No. 101057014). MJW receives funding from the National Centre for the Replacement, Refinement and Reduction of Animals in Research (NC3Rs), the Biotechnology and Biological Sciences Research Council (BBSRC), and the Royal Society.
