Analysis of predictive value of animal repeated dose toxicity study results for clinical safety of US FDA-approved anticancer drugs between 2019 and 2023

Abstract

The translational relevance of nonclinical animal studies in predicting clinical safety outcomes remains poorly understood in oncology drug development. This study aimed to qualitatively assess the concordance between toxicological findings from animal repeat-dose toxicity studies and treatment-emergent adverse events (TEAEs) in clinical trials of 63 anticancer drugs approved by the U.S. FDA from 2019 to 2023. We extracted nonclinical and clinical safety data from FDA review documents and categorized observations across 20 organ systems using a Concordant Positive/Concordant Negative/Animal-only Positive/Clinical-only Positive classification framework. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated by organ class and animal species. Among 63 anti-cancer drugs analyzed, hematologic, hepatobiliary, and gastrointestinal systems exhibited high concordance (notably high PPV and sensitivity), while bone/tooth and lymphatic systems showed low translational predictivity. Predictive performance was similar across rats, dogs, and monkeys, with no statistically significant interspecies differences. These findings suggest that current animal models offer organ-specific, but not species-specific, translational value for anti-cancer drugs. A shift toward validated, human-relevant alternative models is recommended for organ classes with low predictive performance in animal studies, to enhance translational accuracy while reducing animal sacrifice. © 2026 Elsevier Inc.