A development-focused report argues that improving preclinical-to-clinical translation requires more than candidate selection—it depends on how safety and predictability data are generated. It notes that more than 90% of drug candidates entering clinical trials fail to reach approval, citing inadequate efficacy or unexpected toxicity as leading causes. The update highlights limitations of animal models for predicting human outcomes, including differences in tissue biology and lack of genetic diversity. It also points to U.S. regulatory actions: the FDA Modernization Act 2.0 passed in 2022 and an FDA roadmap released in 2025 outlining plans to phase out animal testing for new drug evaluations. For biotech teams, the immediate takeaway is a strategic shift in how preclinical evidence may be expected to be built—pushing programs toward alternative models and better mechanistic relevance. The piece frames the clinical trial phases as the sequence where early translational evidence must hold up, reinforcing why better preclinical predictability is central to late-stage success rates.