Modernizing Preclinical Drug Development: The Role of New Approach Methodologies

Over 90% of investigational drugs fail during clinical development, largely due to poor translation of pharmacokinetic, efficacy, and toxicity data from preclinical to clinical settings. The high costs and ethical concerns associated with translational failures highlight the need for more efficient and reliable preclinical tools. Human-relevant new approach methodologies (NAMs), including advanced in vitro systems, in silico mechanistic models, and computational techniques like artificial intelligence and machine learning, can improve translational success, as evident by several literature examples. Case studies on physiologically based pharmacokinetic modeling and quantitative systems pharmacology applications demonstrate the potential of NAMs in improving translational accuracy, reducing reliance on animal studies. Additionally, mechanistic modeling approaches for drug-induced liver injury and tumor microenvironment models have provided critical insights into drug safety and efficacy. We propose a structured and iterative "a priori in silico" workflow that integrates NAM components to actively guide preclinical study designa step toward more predictive and resource-efficient drug development. The proposed workflow can enable in vivo predictions to guide the design of reduced and optimal preclinical studies. The findings from these preclinical studies can then be used to refine computational models to enhance the accuracy of human predictions or guide additional preclinical studies, as needed. To conclude, integrating computational and in vitro NAM approaches can optimize preclinical drug development, improving translational accuracy and reducing clinical trial failures. This paradigm shift is further supported by global regulations, such as the FDA Modernization Act 2.0 and EMA directive 2010/63/EU, underscoring the regulatory momentum toward adopting human-relevant NAMs as the new standard in preclinical drug development.