An exploratory ADAPT-2 analysis found additional improvement among participants continuing naltrexone–bupropion through twelve weeks. That finding is different from a randomized test of six weeks versus twelve weeks of treatment, and part of its comparison depended on imputed outcomes.
Why was a follow-up analysis needed?
The original ADAPT-2 trial used two six-week stages. At baseline, 109 participants were assigned to the medication combination and 294 to placebo. Eligible placebo nonresponders were then rerandomized: 114 switched to active medication and 111 remained on placebo. Participants originally receiving active medication continued it.
Li and colleagues used all 403 participants to examine changes in repeated urine-screen outcomes over the full trial. This was a secondary analysis of the same cohort, not a new trial or independent replication. The authors stated that this analysis was not preregistered and should be considered exploratory.
Which regimen was continued?
Extended-release naltrexone, 380 mg intramuscularly, was administered in weeks one, four, seven and ten. Oral extended-release bupropion was increased to 450 mg daily, with tapering after treatment. Participants supplied urine twice weekly during twelve treatment weeks, then at follow-up in weeks thirteen and sixteen.
No active-treatment group was randomly assigned to stop at week six. Consequently, continued improvement cannot by itself quantify what would have happened to the same participants if active treatment had ended then.
Why were placebo outcomes imputed?
Removing the 114 placebo nonresponders who switched treatment would change the composition of the remaining placebo group. To address this, the investigators used 1,000 imputations to simulate urine results for those participants as if they had remained on placebo during stage two.
The resulting comparison modeled all 294 original placebo participants as continuing placebo. Those counterfactual results were generated from a statistical submodel fitted to stage-two data; they were not observed specimens from people who actually stayed on placebo. Imputation addresses a design problem under assumptions, rather than creating additional measured evidence.
What improved during weeks seven to twelve?
The modeled probability of a methamphetamine-negative urine screen in the active group increased from 20.3% at week seven to 29.5% at week twelve. Table 1 and the results text report a 9.2-percentage-point increase, with a 95% confidence interval of 0.5–17.9 points and p = 0.038. The abstract prints a different lower confidence bound; the table and main results agree on 0.5.
The modeled placebo increase over the same period was 3.9 percentage points, with a 95% confidence interval of −0.3–8.1 and p = 0.072. Comparing significance within each group is not a direct test of the difference between those two stage-two changes.
What was the full-treatment comparison?
Across twelve weeks, the modeled increase was 27.1 percentage points with active medication and 11.4 with placebo. The reported difference in change was 15.8 percentage points, 95% confidence interval 4.5–27.0, p = 0.006. These are changes in modeled negative-screen probability, not the proportion continuously abstinent for twelve weeks or the original trial’s binary responder percentage.
Missed visits were treated as missing at random in the main model. Sensitivity analyses using last observation carried forward or counting missing results as positive retained a significant full-period treatment comparison. Those checks are reassuring for that comparison, but do not prove every modeling assumption or establish an optimal treatment duration.
What happened after treatment?
Neither group showed a significant change between the two post-treatment follow-up visits. With only those follow-up observations, a null change does not prove durable remission or show that stopping medication caused a plateau.
The paper supports investigating longer treatment and gradual use reduction. A prospective duration comparison and longer follow-up would be needed to establish how much additional treatment helps, for whom and for how long.
Source
Li MJ et al. Extended observation of reduced methamphetamine use with combined naltrexone plus bupropion in the ADAPT-2 trial. Addiction. 2024;119:1840–1845. doi:10.1111/add.16529. Exploratory secondary analysis of historical human-trial data.
Leave a Reply