Title: Why do 90% of the drugs fail in development? An analysis of 82+ submissions; and a case study of immunotherapy
Abstract:
Despite unprecedented advances in molecular biology, genomics, biomarker discovery, artificial intelligence and translational science, the probability that a drug entering clinical development will ultimately reach patients remains remarkably low. Why do apparently promising molecules, targets and development programs fail?
This keynote presents a systematic analysis of failure across the drug-development continuum, integrating more than four decades of experience in immunology and more than two decades in biologics and clinical drug development with a structured analysis of 200+ U.S. FDA Complete Response Letters (CRLs) and development submissions/programs. Rather than viewing failures as isolated events, the analysis seeks to identify recurring patterns, failure modes and points in the development pathway where risk could potentially have been recognized earlier. The presentation will examine failures arising from efficacy, safety, clinical-trial design, patient selection, biomarkers, pharmacology, manufacturing and product quality, immunogenicity, regulatory expectations and the translation of preclinical findings into clinical benefit. The objective is to move from a retrospective description of failure toward a predictive framework for failure prevention and risk mitigation.
A detailed case study of cancer immunotherapy illustrates this concept. The success of immune-checkpoint blockade targeting CTLA-4 and PD-1/PD-L1 created enormous expectations that other inhibitory pathways would produce similar therapeutic benefits. However, despite compelling biological rationale and extensive clinical investigation, programs targeting TIGIT, LAG-3 and TIM-3 have produced disappointing or inconsistent outcomes. Why did a successful therapeutic paradigm not generalize across apparently related checkpoint pathways?
By comparing successful and unsuccessful immunotherapy programs, the keynote will explore the distinction between biological plausibility and clinical validation, the importance of target biology, redundancy and context within the tumor microenvironment, patient selection, biomarker strategy, combination design and the limitations of extrapolating from one successful mechanism to another. The central message is that drug development failures are rarely random. Failure leaves patterns. Systematically capturing, classifying and modeling those patterns could enable developers to identify high-risk assumptions earlier, improve development decisions and ultimately increase the probability of successful translation from molecule to medicine.
