Predicting individual treatment response in alcohol use disorders: a reverse translational proof-of-concept study.

The development of medications for alcohol use disorders (AUD) faces stagnation, as promising drugs failed to translate in clinic. Screening on homogeneous groups of animals drugs later tested on heterogeneous clinical cohorts may contribute to the translational gap. We hypothesized that a preclinical model of AUD accounting for inter-individual heterogeneity would predict the lack of efficacy of a drug that failed clinical trials (Memantine) and the efficacy of an approved AUD medication (Naltrexone). Baseline alcohol drinking, motivation, and cued reinstatement were screened in NIH genetically heterogeneous-stock rats before testing the effect of Memantine and Naltrexone on alcohol (ASA) and saccharin self-administration (SSA). Based on the individual effect of Memantine and Naltrexone on ASA, rats were allocated into independent clusters of responders and non-responders to each drug. The same doses of Memantine reduced both ASA and SSA in both clusters, while Naltrexone selectively reduced ASA in responder rats. Naltrexone responders were in majority males, while non-responders were mostly females. Naltrexone responders and non-responders showed similar alcohol drinking and motivation, but non-responders did not show cued reinstatement of alcohol seeking. In line with clinical observations, in a model accounting for individual heterogeneity Memantine failed to selectively reduce ASA, the population could be unbiasedly clustered in responders and non-responders, and cued reactivity associated with Naltrexone response in males. These results advocate the use of inter-individual heterogeneity for preclinical prediction of drug efficacy in AUD before clinical trials. In addition, we observed sex differences in response to Naltrexone that can be back-translated in clinic.