A New Paradigm Begins, Gradually Replacing After Careful Validation

　　The core components of the new toolbox include human organoids and organ chips. Compared to animals, these emerging technologies can better simulate human tissue structure and function, identifying human-specific toxic reactions that are undetectable or easily overlooked in traditional animal models. Particularly in tumor drug development, organoids and organ chips are expected to achieve more precise personalized drug screening, with the latter offering the added advantages of high efficiency and high throughput.

　　This time, the FDA has designated "drug-induced liver injury" as the first validation target for non-animal method safety evaluations. This is because the liver is the direct core organ for drug metabolism, with relatively sufficient accumulated data, making it possible to establish a referenceable and regulatable standard model first. Wang Mingwei predicts that if the validation of the drug-induced liver injury project progresses smoothly, subsequent validations for the heart, kidneys, lungs, digestive tract, and blood vessels will follow in sequence. "The scientific community has a deeper understanding of these organs, and their indicators are easier to quantify."

　　The advantages of non-animal methods are obvious. Organoids and organ chips can more accurately predict the actual effects of drugs on the human body, significantly shorten drug development cycles, allow patients to access innovative therapies sooner, and reduce drug development costs and prices. According to FDA estimates, the development cost of monoclonal antibody drugs typically ranges from $650 million to $750 million, requiring the use of 144 primates over a nine-year development cycle, with each primate costing up to $50,000. FDA Commissioner Marty Makary therefore refers to non-animal alternative methods as a "win-win for public health and ethics."

　　Of course, emerging technologies also have some shortcomings. Wang Mingwei notes that existing organoid models struggle to fully simulate the complex human system, especially the interconnected reactions involving multiple organs like the immune system, which are harder to replicate with a single model. Additionally, organoids currently cannot support long-term cultivation and observation, whereas experiments for chronic diseases like diabetes and hypertension require years of research.

　　Beyond organoids and organ chips, the FDA encourages drug developers to use AI-based computational models to predict drug efficacy. Wang Mingwei cautions that current AI models can only serve as auxiliary predictions, as the safety and efficacy of drugs still need to be tested through cells, organoids, or animal experiments, and ultimately validated clinically.

　　In response to concerns about "using humans as guinea pigs," the FDA also plans to issue guidelines for the use of alternative methods, continuously validating new models with clinical data and updating the guidelines. Wang Mingwei believes that during the transition phase, drug development will still require data from both alternative methods and animal experiments. Only after the comparability of new methods is validated will they be allowed to gradually replace traditional methods. "In the short term, the two methods will definitely be used in combination, with partial replacement, and the speed of gradual replacement will depend on the results from the models."