Trajectories of Impulsivity in Opioid Use Disorder Treatment: A Longitudinal Study of Temporal Discounting and its Dynamic Relation to Drug Use and Treatment Efficacy

Background: Impulsivity is a core feature of substance use
disorders. Temporal discounting (TD) paradigms provide a
model-based approach to studying the dynamics of impulsive
decision-making as individuals with substance use
disorder undergo treatment. Here, we examine how TD
changes as opioid use disorder (OUD) subjects stabilize on
maintenance therapy and we assess how TD is predicted by
(or is predictive of) relevant clinical outcomes such as illicit
drug use, treatment adherence and clinical states like craving.
Methods: Individuals initiating medication-assisted treatment
for OUD were assessed weekly then bi-weekly (for up
to seven months) on a simple TD task. For each session, (1)
we derived a computational subject-specific parameter for
the TD rate as well as a model-free measure: the proportion
of immediate rewards chosen; (2) we monitored illicit drug
use through randomly administered weekly urine toxicology
and direct self-report; (3) we established their level of
adherence to their individual treatment plan as well as their
current medication dose; and (4) we scored their current
levels of craving, withdrawal symptoms and state anxiety.
A group of demographically matched drug-free community
controls (CC) were similarly assessed repeatedly in order to
establish the test-retest reliability of our measurement and
discard effects of practice and repetition.
In addition, eligible subjects from both groups completed the
tasks while we acquired functional magnetic resonance
imaging (MRI) data in two sessions: one at the beginning
of the study and the other 8-12 weeks later. During this
interval, subjects continued their regular assessments outside
of the scanner.
Results: As previously reported, OUD patients have
significantly higher discount rates compared to controls
but in our demographically matched groups the difference
appears to be smaller than previously reported. Our results
indicate that TD measurements have high test-retest
reliability. While stable in our control group, in OUD
patients the TD rates are a dynamic function of time in
treatment. Interestingly, TD rates also correlate with illicit
drug use events, peaking around the time that these occur.
Moreover, the individual trajectory of TD leading up to these
lapse events correlates with the degree of overall use during
our follow up, suggesting that the course of a patient's
impulsivity might be predictive of their relative success at
maintaining abstinence during treatment.
Conclusions: We conclude that TD, when assessed repeatedly
over the course of treatment, could be used as a
behavioral signature of a patient's clinical evolution and
potentially serve as a useful predictor of prognosis and
treatment adherence for OUD. Our TD task is easy to
automate and administer and therefore lends itself to use in
larger clinical studies and might be useful to incorporate into
the monitoring of these patients' progression. Our ongoing
efforts focus on the investigation of the neural substrate(s) of
the observed change in TD with treatment for OUD. We are
exploring how the activity of regions involved in the
computations necessary for impulsive decision-making (i.e.
the ventromedial prefrontal cortex, ventral striatum and
posterior cingulate cortex) contributes to treatment efficacy