A treatment-effect difference between subgroups can generate an important research question without explaining why the difference occurred. A secondary ADAPT-2 analysis compared groups of men defined by reported sexual partners and found a larger medication effect in one group. It did not establish that sexual identity biologically determines response.
Which participants were included?
Kidd and colleagues used data from the existing naltrexone–bupropion trial, rather than recruiting a new independent cohort. Of 403 participants, 277 were cisgender men. After excluding 31 with missing sexual-partner information, the analysis included 246: 151 reporting male partners, with or without female partners, and 95 reporting only female partners.
The classification used partner reports at baseline, six weeks and twelve weeks. It measured behavior within specified windows, not a comprehensive assessment of identity or lifetime orientation. Only 14 men reported both male and female partners, so that group was combined with men reporting only male partners.
What treatment and response were compared?
The active regimen combined 380 mg extended-release naltrexone by intramuscular injection every three weeks with oral extended-release bupropion targeted to 450 mg daily. The parent study used two six-week stages, with eligible placebo nonresponders rerandomized for the second stage.
Response required at least three methamphetamine-negative urine specimens among four scheduled specimens at the end of a stage. Discontinuation or at least two missing results counted as nonresponse. The analysis therefore concerned that specific endpoint, not continuous abstinence, a diagnosis of recovery or the effect of either medicine alone.
Was there a direct test of subgroup differences?
Yes. The investigators tested a contrast of treatment effects, rather than relying only on a significant result in one group and a nonsignificant result in another. After adjustment for site and recorded demographic and clinical characteristics, the reported weighted medication-versus-placebo response difference was 0.1479 in the male-partner group and 0.0227 in the female-partner-only group: approximately 14.79 and 2.27 percentage points. The subgroup contrast had p = 0.04.
That interaction result is more informative than comparing two separate p-values. It nevertheless comes from a secondary analysis with limited subgroup sample sizes and needs prospective replication. A small estimated effect in one subgroup is not proof that no individual in it can benefit.
What explanations were established?
The groups differed in recorded characteristics, including HIV status, education and ethnicity. The models adjusted for several measured variables, but subgroup membership was not randomized. The study did not experimentally isolate a biological effect of orientation, the role of sexual context, stress or any specific mechanism.
The authors had originally hypothesized a smaller medication effect in the male-partner group; the observed direction was the opposite. They also found no significant group difference in reported medication adherence or baseline moderate-to-severe depression. Those findings limit simple explanations based on presumed adherence or symptom differences.
Who cannot be represented by this result?
The analysis included no transgender men and cannot establish effects in transgender or nonbinary people. Most participants were non-Hispanic White. Men with both male and female partners could not be evaluated separately with adequate numbers, and missing partner reports excluded others.
The authors emphasized limited measures of sexual identity, compulsivity and related factors. Behavior-based categories should not be silently relabeled as fixed identities, and a sample-level result should not become an assumption about a particular person.
What does the analysis add?
It identifies a hypothesis for trials designed to examine treatment-effect heterogeneity and the factors behind it. It is a separate question from the parent trial’s overall efficacy endpoint, but uses the same participants. It neither supplies an identity-based treatment rule nor constitutes an independent replication of ADAPT-2.
Source
Kidd JD et al. Sexual Orientation Differences among Men in a Randomized Clinical Trial of Extended-Release Naltrexone and Bupropion for Methamphetamine Use Disorder. Drug and Alcohol Dependence. 2023;250:110899. doi:10.1016/j.drugalcdep.2023.110899. Historical secondary human-trial subgroup analysis.
Leave a Reply