Need for NAMs: A systematic evidence synthesis revealing over half a century of drug development failure.

Abstract

Background: Preclinical assessment of drug safety, efficacy and pharmacokinetics still relies largely on animal‑derived data, despite their limited ability to predict human outcomes. New Approach Methodologies (NAMs) are emerging as more human‑relevant preclinical tools. This systematic evidence synthesis examines (1) current levels and temporal trends in clinical drug attrition and (2) the proportion of failures attributed to biological factors such as safety, efficacy and pharmacokinetics that are routinely evaluated during preclinical testing, to identify priority areas for NAM implementation.

Methods and findings: We systematically searched PubMed on March 5, 2024, for peer-reviewed, English-language studies since 1963 reporting numerical primary data on clinical drug attrition rates and/or reasons for failure. The protocol was prospectively registered (osf.io/46mw5). We excluded non-original research and studies on specific cases, diseases, or single drug classes. Thirty-two publications met the criteria, yielding 60 attrition and 36 failure-reason datasets (1963-2017). Across six decades, mean clinical attrition ranged from ∼80 to ∼91 % with the highest rates within the past two decades. Overall, biological factors (safety, efficacy, pharmacokinetics) accounted for ∼77% of reported causes. Limitations include heterogeneous reporting, restriction to English, and exclusion of grey literature.

Conclusions: Clinical drug attrition has not improved in the last six decades, peaking at ∼91% in the 2000s-2010s, with most failures linked to biological factors. These findings suggest that prevailing preclinical approaches - many of which heavily rely on animal models - often have limited ability to predict clinical outcomes and highlight the urgent need for more predictive, human-relevant NAMs.