Extracting Knowledge from Failed Development Programmes

Abstract

Drug development is a challenging business with high risks. Multiple factors can contribute to the failure of a programme. Lack of efficacy and unacceptable safety have been the major reasons for product failure since 2000. Even though the drug molecule and disease target are the most fundamental aspects for successful drug development, rational dose and dosing regimen selection also play an important role in the new drug development era and may determine the fate of a programme. Two case studies are discussed in detail to demonstrate the decision-making process at different stages of drug development that eventually led to the termination of the drug programmes. The two cases presented in this report both involve dose selection. The first programme may have failed as a result of too high doses, even though other factors could also contribute to the unacceptable safety profile. Early clinical studies or even late-phase clinical studies could not detect certain serious adverse events. Possible strategies to reduce this type of failure include application of quantitative structure-activity relationship (QSAR) models to select compounds with no potential serious toxicity, development of more sensitive and relevant animal models to detect drug toxicity, and targeting the minimum dose to achieve reasonable efficacy. The second programme failed during the registration stage. Lessons learned include the following: (i) modelling and simulation could have been conducted after the first pivotal trial to avoid the replication of the failed dose in subsequent trials; (ii) post-hoc subgroup analysis could be misleading, especially for innovative findings without extensive support from other sources; (iii) if the quantitative relationship between the biomarker response and the clinical endpoint is unknown, dose selection should not be guided by a statistically significant biomarker response; and (iv) a more mechanistic model could have been developed for the biomarker response based on the individual biomarker longitudinal response, which may have predicted potential tolerance for higher doses over time. Even though these programmes failed to yield safe and effective drug products, valuable knowledge should be extracted from these programmes as much as possible, in order to provide insight and guidance for all parties involved in the development of drugs, to avoid similar failures and to improve the efficiency of drug development.