An industry perspective on the role and utility of animal models of pain in drug discovery

• Evidence that new pain drug discovery failures are due to insufficient efficacy is poorly documented.
• Discovery of new drugs requires data from many assays in addition to behavioral models.
• Pain models are used in drug discovery to rank order compounds and focus resources.
• Use of new pain models/endpoints to improve translational success first requires their validation.
• Pain model data analysis using effect size and NNT may create better alignment with clinical data.

In recent years, animal behavioral models, particularly those used in pain research, have been increasingly scrutinized and criticized for their role in the poor translation of novel pharmacotherapies for chronic pain. This article addresses the use of animal models of pain from the perspective of industrial drug discovery research. It highlights how, when, and why animal models of pain are used as one of the many experimental tools used to gain better understanding of target mechanisms and rank-order compounds in the iterative process of establishing structure–activity relationships (SAR). Together, these models help create an ‘analgesic signature’ for a compound and inform the indications most likely to yield success in clinical trials. In addition, the authors discuss some often under-appreciated aspects of currently used (traditional) animal models of pain, including how industry balances efficacy with side effect measures as part of the overall conclusion of efficacy. This is provided to add perspective regarding current efforts to develop new models and endpoints both in rodents and larger animal species as well as assess cognitive and/or affective aspects of pain. Finally, the authors suggest ways in which efficacy evaluation in animal models of pain, whether traditional or new, might better align with clinical standards of analysis, citing examples where applying effect size and NNT estimations to animal model data suggest that the efficacy bar often may be set too low preclinically to allow successful translation to the clinical setting