Model‐Informed Reverse and Forward Translation of Safety Risks in Drug Development
Historically, safety assessment has relied on toxicology testing in multiple preclinical models coupled with generous safety margins to both identify key risks and account for uncertainty in translation of these risks across species. Unfortunately, application of this approach alone has several limitations that contribute significantly to the safety attrition rate. First, in preclinical studies many toxicological findings are often observed, and identifying which of these is a true signal and will translate to the patient setting is not always obvious. Further, in non‐life‐threatening or chronic diseases, these margins must often be large to account for uncertainty or increase of severity over time, which can limit doses. Finally, in severe indications where narrow therapeutic margins are common, toxicology studies alone typically provide little information on dose and schedule, eventually leading to time‐consuming work in the clinic to optimize the therapeutic index empirically.
Better tools are therefore needed to translate between in vitro, preclinical and clinical results. Quantitative modeling and simulation approaches have promised to solve many of these challenges in the efficacy space by providing a quantitative framework for decision making which can aid translation by allowing one to:
While significant effort for translational modeling has gone to analyzing efficacy biomarkers and endpoints, a similar level of investment in translating safety signals would provide strong return on investment by strengthening the confidence in safety assessment and enhancing the interpretation of therapeutic index.