Building Robustness into Translational Research

Abstract

Nonclinical studies form the basis for the decision whether to take a therapeutic candidate into the clinic. These studies need to exhibit translational robustness for both ethical and economic reasons. Key findings confirmed in multiple species have a greater chance to also occur in humans. Given the heterogeneity of patient populations, preclinical studies or at least programs comprising multiple studies need to reflect such heterogeneity, e.g., regarding strains, sex, age, and comorbidities of experimental animals. However, introducing such heterogeneity requires larger studies/programs to maintain statistical power in the face of greater variability. In addition to classic sources of bias, e.g., related to lack of randomization and concealment, translational studies face specific sources of potential bias such as that introduced by a model that may not reflect the full spectrum of underlying pathophysiology in patients, that defined by timing of treatment, or that implied in dosing decisions and interspecies differences in pharmacokinetic profiles. The balance of all these factors needs to be considered carefully for each study and program.