Baseline Moderators of Response to Naltrexone and Bupropion in the Treatment of Methamphetamine Use Disorder: An ADAPT-2 Study.

Abstract

Extended-release naltrexone and once-daily bupropion (NTX-BUP) reduced methamphetamine (MA) use in the Accelerated Development of Additive Pharmacotherapy Treatment for Methamphetamine Use Disorder (ADAPT-1, ADAPT-2) trials. Using a proof-of-concept machine learning technique, we created a classification rule using baseline characteristics to identify individuals who would benefit more from NTX-BUP than from placebo (NTX-BUP-preferred). ADAPT-2 (double-blind randomized-controlled sequential parallel comparison design trial, n=147) and ADAPT-1 (open-label pilot study, n=37) were used as the training and validation data sets, respectively. Baseline characteristics were combined to create the classification rule, which was trained to predict the number of negative urine drug screens (UDS) during weeks 5 and 6 of the ADAPT-2 trial and optimized using the ADAPT-1 data set. ADAPT-2 placebo nonresponders rerandomized to NTX-BUP (n=80) were used as the test data set. The performance of the classification rule was assessed by the number of negative UDS during weeks 11 and 12 and the number of responders. NTX-BUP-preferred participants (who would obtain greater benefit from NTX-BUP than from placebo), when compared with all NTX-BUP participants in the test data set, had higher response rates (0.26 vs. 0.16) and number of negative UDS (1.09 vs. 0.76). Placebo-preferred participants (would obtain greater benefit from placebo than from NTX-BUP) had lower response rates to NTX-BUP (0.07 vs. 0.16) and number of negative UDS (0.46 vs. 0.76) compared with all NTX-BUP participants. This is a proof-of-concept analysis that needs to establish generalizability. This classification rule could help improve treatment selection, clinicians' treatment decisions, and patient outcomes.