In this international pilot study we use data from four clinical trials of heroin dependence treatment to identify which patients will be better respondents to naltrexone treatment and among respondents which type of naltrexone formulation (oral, implantable, or injectable) leads to the best outcome. Although in the United States naltrexone has been FDA approved for treatment of opioid addiction it has been underutilized. In Russia (where substitution therapy is illegal), naltrexone is the only pharmacological treatment option. Naltrexone is available in three forms: oral and extended release, injectable, and implantable depot-formulations. A novel methodology (Model-based Random Forest, mobForest) (Garge et al. 2013) will be extended to heroin treatment and to the development of a single interpretable parsimonious model. These methods will be applied to combined data from past four naltrexone efficacy randomized double blind, double dummy, placebo- controlled clinical trials and identify which patients are likely to respond to different naltrexone formulations. These studies were conducted in St. Petersburg, Russia, have data on almost 1,000 patients with more than 400 individual-level variables spanning demographics, clinical, genetic, and psychometrics domains. We will develop validated predictive models that provide prognosis of patients' response. We will provide clinically- relevant interpretations to cross-cultural clinical practice, and assess potential use in a larger multi-country study.
In this international pilot study we use data from four clinical trials of heroin dependence treatment to identify which patients will be better respondents to naltrexone treatment; and among respondents which type of naltrexone formulation (oral, implantable, or injectable) leads to the best outcome. We will provide clinically- relevant interpretations to cross-cultural clinical practice, and assess potential use in a larger multi-country study.
