The FDA Roadmap to Reducing Animal Testing in Preclinical Safety Studies: Where Will It Lead Us?

For several years, the FDA, EMA, and other regulatory agencies have been concerned about the number of animals used in drug research, particularly regarding toxicology studies.1 This concern is not only based on animal welfare considerations; it is also based on the understanding that animal toxicology data does not always accurately predict human toxicity. Against this backdrop, in April 2025, the FDA published its Roadmap to reduce animal testing in preclinical safety studies.2 The Roadmap outlines a comprehensive long-term plan to reduce or possibly eliminate animal toxicology testing, starting with monoclonal antibodies (mAbs), by using what are termed “New Approach Methodologies” (NAMs), which include the use of human tissue-based systems, such as organs-on-chips (OOCs)—human tissue constructs maintained in a microfluidic environment—in silico modeling (physiologically based pharmacokinetics modeling [PBPK] and systems pharmacology modeling), and other innovative approaches, such as high-throughput cell-based screening. As stated in the Roadmap, the goal (3–5 years from the date of publication of this Roadmap) is to make animal studies the exception, rather than the norm, not only for mAbs, but for all therapeutics. We question whether the complete elimination of all animal toxicology studies is a reasonable goal. It is important to remember that, while animal studies might not be perfect, they have been used in science for hundreds of years to very good effect. For the most part, preclinical toxicology studies do a good job of predicting human safety (i.e., absence of toxicological response), and this is why they have been required for decades. The preclinical in vivo IND-enabling toxicity study program approach to assess the general toxicity of pharmaceuticals has been generally successful in allowing reasonably safe clinical trials to proceed.7, 8 From 2000 to 2010, approximately 18% of all clinical programs were discontinued due to either animal or human toxicity (vs 20%–30% for the 50 years prior). In addition, the general ability of animal toxicity data to identify potential adverse human effects is good,9, 10 and the absence of animal toxicity is particularly good at predicting an absence of toxicities in human Phase 1 clinical trials.11 That is, a greater than 86% effective rate for human safety of new chemical entities with unknown properties. However, the ability of animal studies to predict a specific toxicity incidence is about 50%.11 There are currently no alternative predictive toxicological methods that readily allow identification of appropriate dose ranges for clinical exploration and of potential adverse effects over time regarding systemic exposure. As such, there is a very low likelihood of replacing whole animal general toxicity studies in drug development. Scientifically and ethically, any alternative to these studies should probably be required to demonstrate, at a minimum, a success rate similar to animal studies. How predictive are OOCs with regard to determining drug safety? How consistent are the results across various manufacturers of OOCs? What is the ideal experimental design using OOCs? OOCs from different manufacturers may differ significantly in terms of cell types, genetics, and composition of the overall chip structure. It is not yet clear how comparable results are across different OOC platforms. How good will predictions from PBPK models be without preclinical data? Can systems pharmacology models eventually replace experimental animal models of disease or toxicology? The answer to this question, at least in the short term, is likely to be no, mostly because we continue to lack a fundamental understanding of what exactly causes many diseases, and we are not adept at predicting toxic effects without animal data. In vitro cell-based assays have their place in drug discovery but have not yet replaced animal studies in most therapeutic areas, and like OOCs, are a huge oversimplification of what is a very complex system. That is, it will be a challenge for a chronic multiple organ system disease such as diabetes to be replicated within a 96-well plate over a 12-h incubation, even if they are human cells. The lack of congruence between reductionist in vitro assays and human diseases has been pointed out previously.16 We commend the FDA for developing this Roadmap. It shows that the FDA is concerned about the cost and timelines of drug development and is willing to consider novel methods for obtaining toxicology and nonclinical efficacy data. The Roadmap will serve as a useful framework to begin simplifying some aspects of nonclinical toxicology studies. Any reduction in the number of animals used in drug development represents a significant advance, benefiting both industry and the FDA. However, in our view, it is not likely—given the current state of NAM technology and our incomplete understanding of human disease—that animals can ever be completely eliminated from drug research. Both authors consult for the pharmaceutical industry.