The 2021 ADAPT-2 trial found a higher response with extended-release naltrexone plus bupropion than with placebo. Its often quoted 13.6% response needs its original definition: a weighted result from two trial stages, requiring at least three methamphetamine-negative urine samples among four scheduled samples near a stage’s end. It was neither a cure rate nor the percentage continuously abstinent for twelve weeks.
Who received which treatment?
Trivedi and colleagues enrolled 403 adults with moderate or severe methamphetamine use disorder. Eligibility included reported use on at least 18 of the preceding 30 days, repeated positive urine results and being opioid-free at randomization. Participants reported approximately 27 use days in the previous month on average. This was a selected treatment-seeking population with frequent use.
The experimental regimen combined 380 mg of extended-release naltrexone by intramuscular injection every three weeks with oral extended-release bupropion targeted to 450 mg daily. Matching injections and tablets supplied the placebo comparison. These are research doses and schedules; the study is historical treatment evidence, not instructions for self-treatment.
Why were there two randomizations?
The first six-week stage assigned 109 participants to the combination and 294 to placebo. In the second stage, 225 eligible placebo nonresponders were randomized again: 114 to combination treatment and 111 to placebo. Those 225 people were drawn from the original cohort. Adding 403 and 225 would count them twice.
This sequential design deliberately changed the population entering the second comparison. Stage-two participants had already failed the trial’s response criterion while receiving placebo. Their results cannot simply be pooled as though they were an independent second trial of unselected new patients.
How was response calculated?
Response required at least three negative results among four urine samples in weeks 5–6 or weeks 11–12. Participants with at least two missing samples, or who discontinued, counted as nonresponders. The rule therefore incorporated both specimen results and availability.
Stage-one responses were 18/109 (16.5%) with combination treatment and 10/294 (3.4%) with placebo. Stage-two responses were 13/114 (11.4%) and 2/111 (1.8%), respectively. Applying prespecified weights of 0.43 and 0.57 produced 13.6% versus 2.5%, a difference of 11.1 percentage points. The reported 95% confidence interval had a lower boundary of 6.3 percentage points, and the prespecified one-sided primary test gave p<0.001.
What does that benefit establish?
Randomization supports a treatment effect on this defined endpoint under the study conditions. It does not establish that every negative specimen represents an entire drug-free interval, that responders remained abstinent after follow-up, or that nonresponders experienced no benefit whatsoever. A response threshold divides a continuum of possible outcomes into two categories.
The trial also evaluated craving, depressive symptoms and other outcomes. The authors cautioned that secondary-outcome confidence intervals were not adjusted for multiple comparisons. Those analyses should not acquire the same confirmatory status as the primary result.
Benefit and tolerability belong in the same account
Reported adverse events included nausea, vomiting, tremor, sweating and decreased appetite. Serious adverse events occurred in 8/223 participants exposed to combination treatment during the trial; that count does not mean the medication caused each event. Eligibility restrictions, clinical monitoring and imperfect oral adherence limit extrapolation.
A faithful explanation preserves both findings: the trial demonstrated an advantage over placebo, and the absolute weighted response remained low. Describing only the relative increase would obscure the endpoint’s modest absolute frequency and its limited observation window.
Source
Trivedi MH, Walker R, Ling W, et al. Bupropion and Naltrexone in Methamphetamine Use Disorder. New England Journal of Medicine. 2021;384:140–153. doi:10.1056/NEJMoa2020214. Original randomized human trial.
Leave a Reply