Systematic evidence synthesis of clinical drug development failure rates
Not published yetThis systematic review examines the reliability of preclinical studies—many of which rely on animal testing—in predicting the safety and efficacy of new drug candidates in humans over a period spanning nearly six decades. The authors analyzed 60 datasets (1963–2017) on clinical failure rates, defined as the proportion of drugs that ultimately failed to reach the market despite positive results in preclinical studies, including animal experiments. Since the 1960s, approximately 80–90% of drug candidates that showed promising results in preclinical animal studies have failed during subsequent clinical trials in humans. Over the past two decades, this failure rate has remained at a record level of around 91%, exceeding the levels observed in the 1970s and 1980s. The analysis of 32 datasets further shows that, on average, 77% of these failures were attributable to biological factors, such as insufficient efficacy or unexpected toxicity in humans. Because these biological outcomes are precisely what preclinical testing is intended to predict, the findings point to substantial limitations in the translation of animal study results to human biology. Animal testing thus creates a false sense of confidence while contributing to major inefficiencies and costly failures in drug development. They therefore call for a fundamental shift toward human-relevant, animal-free New Approach Methodologies (NAMs), including organoids, organ-on-a-chip systems, and in silico models. According to the authors, wider adoption of these approaches could reduce the risk of biologically driven clinical failures while improving the efficiency and safety of pharmaceutical research.