**Background:** The paper addresses the urgent need for efficient, human-relevant chemical safety assessment methods. Traditional animal studies are slow, expensive, ethically challenging, and often poorly predictive of human responses due to species differences in physiology, metabolism, and genetics. The One Health concept recognizes the interconnected health of people, animals, plants, and the environment, demanding a collaborative chemical strategy. New approach methodologies (NAMs)—including in vitro, in silico, in chemico, and alternative animal models—offer higher throughput, mechanism-based, and human-relevant data at lower costs. However, their adoption has been limited by insufficient validation, standardization, and harmonization. This paper provides a pragmatic decision framework to standardize NAM application in risk assessment, accounting for system complexities and uncertainties.
**Methods:** The framework is an iterative decision tree with four major components: (1) Framing the Assessment—problem formulation, identifying key components, and considering toxicokinetics (ADME) early to ensure in vitro conditions reflect in vivo target tissue concentrations. (2) Dynamics-Bioactivity—determining whether the adverse outcome and its associated adverse outcome pathway (AOP) or mode of action (MoA) are known. If known, a targeted approach measures specific key events (KEs); if unknown, a nontargeted approach (e.g., toxicogenomics) broadly investigates bioactivity. (3) Method Interpretation—evaluating whether guidance and scientifically robust methods are available, whether bioactivation (metabolism) plays a role, and whether data were generated in a dose-response format. If not, data may be used semiquantitatively or qualitatively. (4) Kinetics—using physiologically based kinetic (PBK) models and quantitative in vitro to in vivo extrapolation (QIVIVE) to estimate human equivalent administered doses (EADs) or internal doses. Forward dosimetry estimates internal dose from known external exposure; reverse dosimetry estimates EAD from in vitro bioactivity concentrations. Uncertainty factors (e.g., for interindividual variability and method variability) are applied to derive reference doses (RfDs) or margins of safety (MoS). The paper provides extensive resources for each decision point, including OECD test guidelines, the OECD AOP Knowledge-Base, the Integrated Chemical Environment (ICE), the httk R package, and good in vitro method practice (GIVIMP) guidelines.
**Key Results:** Three case studies illustrate the framework. (1) Benzophenone (BP) in drinking water: Using ToxCast in vitro estrogen receptor assays and QIVIVE, EADs and RfDs were derived for BP and its active metabolite 4-hydroxybenzophenone. The RfDs from NAMs (e.g., 0.002–0.083 mg/kg-day) were within an order of magnitude of the traditional animal-derived RfD (0.02 mg/kg-day), supporting the protective value of the animal-based RfD for endocrine endpoints. (2) N-methylmorpholine N-oxide (NMMO): In vitro models of spermatogenesis in rat, monkey, and human showed that NMMO caused dose-dependent decreases in secondary spermatocytes in rats at oral equivalent doses (OEDs) ≥ 89 mg/kg/day, but no such effects in monkeys up to 1376 mg/kg/day. This species difference indicates NMMO is unlikely to be a spermatotoxin in humans. (3) XU-18840.00 (new cosmetic ingredient): A nontargeted NAM battery (in silico predictions, in vitro assays for skin irritation, genotoxicity, eye corrosion, phototoxicity, Safety47 panel, cell stress panel) identified the lowest point of departure (PoD) at 60 µM (decreased glycolysis). Forward dosimetry using dermal penetration (≤3.5%) and PBK modeling estimated an internal dose of 9.4 nM, yielding an MoS > 1000, supporting safe use at 0.1% in face cream. In vivo 90-day oral rat data (NOAEL 624 mg/kg/day) corroborated the NAM-based safety conclusion.
**Clinical Implications:** This framework provides a standardized, transparent process for integrating NAMs into chemical risk assessment, reducing reliance on animal testing while improving human relevance. It enables derivation of protective reference doses or margins of safety for chemicals with limited or no animal data, particularly relevant for cosmetics (where animal testing is banned in the EU), industrial chemicals, and food contact substances. The case studies demonstrate that NAMs can effectively address endocrine, reproductive, and systemic toxicity endpoints, and that QIVIVE can bridge in vitro bioactivity to human exposure levels. The framework also highlights key uncertainties (e.g., metabolic competence, method variability, population variability) and proposes approaches to manage them (e.g., using uncertainty factors, Monte Carlo analysis). As NAMs continue to advance, this framework supports their regulatory acceptance and broader adoption in public health protection, aligning with global initiatives for a toxic-free environment and One Health.