Varenicline has shown promise for treating alcohol use disorder (AUD); however, not everyone will respond to varenicline. Machine-learning methods are well suited to identify treatment responders. In the present study, we examined data from the National Institute on Alcohol Abuse and Alcoholism Clinical Intervention Group multisite clinical trial of varenicline using two machine-learning methods. Baseline characteristics taken from a randomized clinical trial of varenicline were examined as potential moderators of treatment response using qualitative interaction trees ( N = 199) and group least absolute shrinkage and selection operator interaction nets ( N = 200). Results align with prior research, highlighting smoking status, AUD severity, medication adherence, and drinking goal as predictors of treatment response. Novel findings included the interaction between age and cardiovascular health in predicting clinical response and stronger medication effects among individuals with lower craving. With increased integration of machine-learning methods, studies that effectively integrate methods and medication development have high potential to inform clinical practice.

Background: Varenicline is FDA-approved for smoking cessation but has shown promise for the treatment of alcohol use disorder (AUD). However, not all individuals will respond to varenicline. Machine learning methods are well-suited to identify treatment responders using data-driven tools. Towards identifying responders to varenicline and improving the application of machine learning to the prediction of clinical response in AUD, the present study examines data from the NCIG multisite clinical trial of varenicline using two machine learning methods.Methods: Data for this secondary analysis were drawn from a 13-week, multisite Phase 2 double-blind, placebo-controlled, parallel group trial of varenicline. The current analysis used the primary and secondary drinking endpoints from the original clinical trial. Baseline characteristics were examined as potential moderators of treatment response. Qualitative INteraction Trees (QUINT) and group-lasso interaction-nets (Glinternet) models were used to identify treatment responders.Results: Both approaches indicated that older individuals who were current smokers responded better to treatment with varenicline on the primary study outcome. Within approaches, the Glinternet models highlighted race, higher medication dose, and less severe AUD symptom criteria as important features of treatment responders across primary and secondary drinking outcomes. The QUINT models were less consistent, but highlighted older age, heart rate, and more recent smoking as important features of treatment response.Conclusions: Results were consistent with theory-driven models by highlighting smoking status, severity of AUD, medication adherence, and drinking goal as predictors of treatment response. This study also produced novel findings, including the interaction between age and cardiovascular health in predicting clinical response and the stronger medication effect among individuals with lower craving scores. As the field continues to integrate machine learning methods in personalized medicine, studies that can provide a nuanced methodological examination and effectively integrate methods with the medication development literature have the highest potential to inform clinical practice.