In NAMs we trust – an innovative paradigm-shift in risk-based chemicals management for globally harmonized protection goals

KEYWORDS:

• New approach methods
• exposure assessment
• chemicals safety
• toxicological risk assessment

1. Introduction

Toxicological risk assessment for chemicals is on the brink of revolution. New approach methods (NAMs) have been developed over the past two decades, and case studies are presented about how to integrate NAM data in safety assessments [1–3]. NAMs include a variety of non-animal in vitro species-relevant cell-based assays, computational approaches, and chemistry-and biology-based knowledge that can be integrated in the context of adverse outcome pathways (AOPs) to provide evidence for assessing the likelihood of adverse biological responses and disease states arising from chemical exposures [4]. The general principle of classical toxicology, that risk is a function of effects response and exposure dose, still holds true when using NAMs in next generation risk assessment (NGRA) for chemical safety evaluation. The dose of a substance (and where and how it is bio-transformed into chemicals delivered to the site of a mode/mechanism of biological action) defines the poison. In using NAMs to evaluate the potential for a systemic exposure dose to a chemical to cause adverse effects, the intent is not to predict the current classical toxicology hazards that would be listed in an endpoint driven EU REACH dossier today, for example, but to look at the risks of adverse outcomes arising from real-life exposure scenarios, from the point of view of ensuring worker, consumer, general public or environmental protection from harm and particularly where current data are lacking for thousands of substances [5].

2. The paradigm shift

The biggest challenge for a paradigm-shift in toxicological risk assessment is to move from the mid-20^th century science of animal models to the application of more species relevant exposure estimation tools and modern biological models from 21^st century science. This includes chemical dosing of human cells and organoids and using transcriptomics, proteomics, receptor-based assays, metabolomics, cheminformatics, and mathematical modeling analyses. Some NAMs, particularly those using omics analyses, can generate large amounts of data and need sophisticated but easy-to-understand statistical analysis to develop confidence in using the data to support decision-making. The safety case needs clear communication to capture the biological relevance, assay constraints, the underlying mechanisms of adverse outcomes, sensitivity analysis in the data, and the sources of uncertainty.

Aiming to assess risk inevitably leads to the need for chemical exposure estimation. NAMs can be developed to investigate the potential for effects/hazard identification and exposure estimation, such as the use of physiologically-based kinetic (PBK) modeling to define internal doses [6]. Having information on which cells, tissues, and organs are likely to be exposed to chemicals in the body/organism can help to define the types of NAM effects-based assays that are useful to select from a broad palette of assays and focus on in a risk assessment context. To ensure people and wildlife continue to be protected to a high standard whilst also enabling innovation, it is necessary to consider what is likely to happen at the doses to which people and wildlife are really exposed. This puts a greater emphasis on problem formulation, exposure estimation, and hypothesis generation at the beginning than is the case in the current empirical hazard-led paradigm using animal data that is extrapolated to yield acceptable doses by applying large >100-fold default uncertainty factors to crude point of departure estimates.

3. The acceptance of NAMs for chemical safety decision-making

In a paper by Hilton et al. [7], the progression and acceptance of new scientific paradigms relating to Thomas Kuhn’s ‘The Structure of Scientific Revolutions’ is discussed philosophically in the context of the acceptance of the new science of NAMs. In a training session using a case study on how NAMs can be used for safety decision-making with no reference to animal data at all, the audience indicated their relatively high level of confidence in the exposure assessment outputs using NAMs such as PBK models, and their relatively high confidence in the biological response data from a broad suite of in vitro bioactivity assays covering hundreds of human cellular and molecular pathways. Based on the evidence presented, the situation being evaluated was agreed by the majority as low risk, as there was minimal bioactivity. However, the confidence of the participants in taking a decision to allow the use of the chemical in a product using evidentially conservative and moderate to high confidence data remained overall low and reticent. This leads one to think that the acceptance of NAM data in making safety decisions, will not just be about the quality or demonstrable biological relevance of the scientific evidence but also in the familiarity with the new data types, the psychology of decision-making and the trust and credibility of a new NGRA process. Natural scientists will need to work together with stakeholders, including risk assessors, social scientists, and decision-makers, to shift the regulatory paradigm.

The desire to accept chemical safety evaluation with the full replacement of animal testing becomes a reality not through science alone but through political will and the building of trust and confidence in the new science applied in context [8]. It is pleasing to see the 2023 report from the European Chemicals Agency [9] on the use of alternatives to testing on animals that approximately 90% of studies between 2019 and 2022 for skin corrosion/irritation, eye damage/irritation, and skin sensitization were performed in vitro, compared to 50% between 1990 and 2019. The increase in trust in NAMs may also have to come alongside a decrease in trust of traditional animal testing when the shortcomings become more apparent as biological science understanding increases. Such developments may result in international policy differences in chemical trade and require sensitive management, hence perhaps some hesitancy to adopt.

Animals have been used as trusted models, but no model is a complete representation of the species under consideration; a rat is not a human being. Trust in NAMs is not likely to come through simply validating NAMs as compared with how they predict animal hazard tests; a desire to not have to change complex existing laws and guidance may be a factor in advocating for this approach, so the new thing can be easily slotted in for the old thing in law. It is not likely to be so easy. Trust and confidence will come from science diplomacy and case-study demonstrations of how NAMs can realistically be applied in a tiered and integrated way to making safety decisions, considering exposure first. Van der Zalm et al. [10] proposed five essential elements to build scientific confidence in NAMs: fitness for purpose, human biological relevance, technical characterization, data integrity and transparency, and independent review. One thing is for sure, there will need to be a re-write of technical guidance.

4. Tiered approaches to data generation and new concepts for quantitative risk estimation

In practical terms, it is likely that data generation workflows will be tiered to make for cost-effective and pragmatic ways forward. This is already the case for exposure estimation, where highly conservative assumptions are made in the simpler Tier 1, and in higher Tiers where more information and data are built into probabilistic model [11], up to the ultimate Tier 3 of measuring chemicals in human blood and urine through biomonitoring. A tiered approach to ‘read-across’ is also possible, including both chemical and biological read-across evidence, within the context of a 10-step framework and decisions taken iteratively on when there is enough evidence to demonstrate a common mode or mechanism of action [11]. The Bioactivity-Exposure Ratio (BER) [1,12] or the Dietary Comparator Ratio (DCR) [13] concepts have emerged where internal systemic exposure dose concentrations (µmol) estimated from PBK modeling tools are compared to concentrations of chemicals (in µmol) causing bioactivity responses in cell-based or receptor binding assays. However, the question remains – what is an acceptable BER? Risk assessment methods for answering this question are likely to be probabilistic, Bayesian, and statistical in nature [14] and require discussion at the science-policy interface to harmonize protection goals.

5. Responsible and ethical capacity building in NAMs technology

Applying NAMs in risk-based paradigms will shift the practices, technical guidance, capability areas, and skills that are going to be needed within industry and regulatory bodies. The fast pace of scientific developments will also likely lead to the need to change chemicals regulations. It is important for any future capacity building activities (for example, in United Nations work on a new Global Framework for Chemicals (UN GFC, 2023) [15] and the establishment of a science-policy panel (SPP) for chemicals, waste, and pollution prevention) that a future-proofed understanding of the paradigm shift on the horizon for chemical safety testing is understood. It would not be morally ethical to train up scientists, build infrastructures and regulatory regimes in the developing world on the principles and practices of animal testing alone, when the rest of the world is using 21^st century science. Legacy in vivo data should be shared on open data platforms to avoid duplication; such data still has technical value. However, the future should be about building a shared trust and confidence in the NAMs of modern day safety science, through increased access to global open training platforms (e.g. from the US EPA), exposure to case studies and global scientific collaborations.

Declarations of interest

C Alexander-White is a consultant working in the cosmetics sector and is an independent science adviser to the government in the United Kingdom. The opinions and views are of the author.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.