Changes in Computational Markers of Decision-Making are Proximally Tied to Heroin Use in Opioid Users Followed Longitudinally Through the First Months of Treatment

Anna Konova, Silvia Lopez-Guzman, Adelya Urmanche, Stephen Ross, Kenway Louie, John Rotrosen, Paul Glimcher

Research output: Contribution to journalMeeting Abstractpeer-review

2 Scopus citations


Background: Opioid overdose deaths have increased four-fold in the last decade and are now the leading cause of accidental death in the U.S. Relapse to opioid (heroin and painkiller) use during abstinence or opioid replacement treatment significantly increases the risk for overdose. Identifying predictors of relapse, and ultimately preventing relapse, in abstaining and/or treatment-seeking individuals is thus of critical importance. Here we tested the hypothesis that computational markers of decision-making are proximately tied to heroin use in individuals seeking treatment for their opioid addiction. We specifically hypothesized that the value of risky prospects is enhanced when these individuals are most vulnerable to return to heroin use, and that the neurobiological basis of this vulnerability would involve the brain’s valuation system comprised of the striatum and ventromedial prefrontal cortex (VMPFC). Methods: 79 chronic opioid users were serially studied over the first 7 months of initiating opioid replacement treatment (1-15 study sessions/subject; mean=6, SD=3.9). At each session, we characterized individual decision-making and documented any return to heroin use by self-report and random (weekly) urine toxicology tests. To derive computational markers of decision-making we used a validated task amenable to neuroeconomic modeling. In the task completed at each session, subjects repeatedly chose between certain gain of real money and a chance to play a lottery. Lotteries differed in possible monetary gain and in the probability of experiencing that gain. In some trials, the probability of winning the lottery was known (risky prospects), in others, the probability was only partially known (ambiguous prospects). This overall design allowed us to perform time-lagged analyses predicting heroin use next session from decision-making behavior on the current session via mixed-effects logistic regression. A subset of the larger cohort (n=12, study ongoing) also completed a multi-band fMRI session where they performed the same decision-making task and resting-state scans to isolate the neural mechanism of the decision process involved. Results: Of 605 total sessions, 288 (47%) were heroin positive. Only an increase in an ambiguity tolerance decision parameter was significantly predictive of prospective heroin use at the timescale examined [t(577)=3.39, p<0.001], supporting our hypothesis that the value of these (more risky) prospects is enhanced when individuals are most vulnerable. The preliminary fMRI data revealed activity in the striatum and VMPFC encoded the value of the ambiguous prospects on a trial-by-trial basis, increasing as their subjective value increased. Activity in these regions in turn correlated with connectivity between them at rest, together suggesting that coordinated activity in the brain’s valuation system might underlie both the observed decision-making behavior and heroin use vulnerability. Conclusions: Treatment monitoring and intervention efforts aimed at decision processes for risky prospects (and thus the valuation system) may help reduce incidence of relapse in a population at high risk for overdose death.
Original languageEnglish (US)
Pages (from-to)S279
Issue numberS1
StatePublished - Nov 1 2017


Dive into the research topics of 'Changes in Computational Markers of Decision-Making are Proximally Tied to Heroin Use in Opioid Users Followed Longitudinally Through the First Months of Treatment'. Together they form a unique fingerprint.

Cite this