Abuse potential studies

Studies of abuse potential for drugs targeting the central nervous system have historically used the non-human primate as the ‘gold standard model’. Although this often involves naïve animals many studies use substitution paradigms where animals are trained to self-administer known substances of abuse such as cocaine, which are then removed to determine whether the animal self-administers the test drug.

Practice varies among companies with some using the rat and others the non-human primate. Working primarily with Pfizer, we have built a case to use the rat, supporting a change in ICH M3 to accept rodent instead of non-human primate data. We have analysed more than 500 papers reporting data on 71 drugs that were assessed in the rat self-administration model and the clinic. We found that overall there was 90% (64/71) concordance between the rat and human for a range of drug classes. For the drugs where non-human primate data were also available there was no statistical difference between the rat and non-human primate at predicting human abuse liability and scheduling status. We have also generated data for certain classes of substances such as opioids which shows that the rat has the same self-administration dose response as the non-human primate. 

Per cent of drugs for which rat self-administration studies are concordant with at least one clinical indicator of abuse liability.

We are also looking at specific aspects of study design to identify recommendations for refinement to the rodent studies. In collaboration with the CAMARADES group we are carrying out a systematic review and meta-analysis of opioid self-administration studies in the rat to investigate the impact of variables such as feeding restrictions, restraint, the type of training animals receive and the amount of time they are exposed to a drug. Understanding how these variables influence the response will enable us to formulate recommendations to improve the welfare of the animals used.


O'Connor E., Chapman K., et al. (2011). The predictive validity of the rat self-administration model for abuse liability. Neuroscience & Biobehavioral Reviews 35(3): 912-938. doi:10.1016/j.neubiorev.2010.10.012

Back to top