Methamphetamine use disorder remains a significant public health issue with high rates of morbidity and mortality and the absence of any United States Food and Drug Administration (FDA)-approved treatments. Therefore, the original report from the NIDA Clinical Trials Network (CTN) ADAPT-2 trial was an important milestone, demonstrating that extended-release naltrexone plus bupropion outperformed placebo in adults with moderate to severe methamphetamine use disorder. Yet, overall improvement remained modest, with only 16.5% of participants meeting the initial treatment response threshold. In a recent secondary analysis of data from the ADAPT-2 trial, we evaluated the factors associated with improvement (ie, attainment of negative urine drug screen [UDS] test) or return to use (ie, testing positive for methamphetamine after a negative UDS). We focused on two constructs that clinicians already recognize as central to maintaining the cycle of addiction, namely craving and impulsivity. Specifically, we examined these constructs dynamically, linking craving and impulsivity at one visit to UDS status at the next. We found that higher craving predicted a lower likelihood of transitioning from a positive UDS to a negative one, and a higher likelihood of transitioning from a negative UDS back to a positive one at the next visit.

Using a relatively simple craving measure of a visual analog scale, we found meaningful prognostic information over a short time period. However, the association between craving and likelihood of improvement with treatment varied based on the treatment and time. For transitions from positive to negative UDS, the association among craving, time, and treatment group was significant. In the naltrexone-bupropion group, lower craving consistently tracked a greater likelihood of improvement, and the separation between lower and higher craving appeared to widen over time. In the placebo group, by contrast, this effect was largely diminished by weeks 5 and 6.

We also found that measuring impulsivity provided additional clinically relevant information. Specifically, in this study, higher impulsivity independently predicted a lower probability of transitioning from positive to negative UDS, even after accounting for craving. The effect was additive, not interactive: individuals with lower craving but higher impulsivity still fared worse than those with lower craving and lower impulsivity. Impulsivity was not significantly associated with transition from negative to positive UDS. Craving, not impulsivity, appeared to be the more consistent signal of return to use once abstinence had been achieved. This difference in association may potentially indicate stage-specific mechanisms. Impulsivity may be especially relevant during the shift from ongoing use toward early abstinence, while craving may be more salient in destabilizing abstinence once achieved. This may suggest that relapse vulnerability is not governed by a single, uniform process and different mechanisms may be relevant during different phases of treatment.

From a clinical perspective, this study suggests the potential clinical utility of repeatedly assessing craving and impulsivity, where persistently elevated craving levels may signal lack of initial improvement with naltrexone-bupropion combination for methamphetamine use disorder. Furthermore, impulsivity may be an important symptom domain in the treatment planning process for individuals with methamphetamine use disorder. While a brief visual analog scale may not capture the full multidimensional nature of craving, and the two-item domain of impulsivity may not capture the broader trait-level features of irritability, this study shows that even simple measures can be clinically informative. That is especially relevant in real-world settings, where overly elaborate assessments may be difficult to implement in routine clinical practice. These findings should be useful in implementing a measurement-based care approach for the treatment of methamphetamine use disorder.

At the same time, the paper’s limitations should guide the field’s next steps. This was an unplanned secondary analysis, not a prospectively designed mechanistic study. The craving measure was retrospective and unidimensional. Missing urine drug screens were not imputed as positive, unlike the primary ADAPT-2 outcome approach. The subgroup contributing to negative-to-positive transitions was relatively small, given that all participants entered with documented recent methamphetamine-positive urine tests. And the use of urine drug screens, while clinically meaningful, does not capture finer-grained variation in amount or frequency of use.

Our findings further support the advancement of measures for craving based on the 2023 draft guidance from the FDA on development of medications for the treatment of stimulant (including methamphetamine) use disorder. Such development is particularly relevant for development of craving as an outcome measure in future FDA-regulated trials.