Economic choice between remifentanil and food in squirrel monkeys

Abstract

Traditional approaches for evaluating if compounds are reinforcing, and thus a risk for abuse, include preclinical self-administration procedures conducted in the absence of alternative reinforcers. While the track record of this approach for determining abuse potential is good, that for predicting efficacy of addiction treatments is not. An alternate approach would be economic choice between drug and nondrug rewards, with parametrically varied options from trial to trial. This would promote goal-directed decisions between reward modalities and should provide metrics that reflect changes in internal state that influence desirability of a given option. We report herein a high throughput economic choice procedure in which squirrel monkeys choose between a short-lived opiate, remifentanil, and a palatable food reward. Stimuli on touchscreens indicate the amount of each reward type offered by varying the number of reward-specific elements. The rapid clearance of remifentanil avoids accumulation of confounding levels of drug, and permits a large number of trials with a wide range of offers of each reward modality. The use of a single metric encompassing multiple values of each reward type within a session enables estimation of indifference values using logistic regression. This indifference value is sensitive to reward devaluation within each reward domain, and is therefore a useful metric for determining shifts in reward preference, as shown with satiation and pharmacological treatment approaches.

Fig. 1. Task design and choice allocation between remifentanil and milk reward.

A Visual of behavioral apparatus which sits in a sound attenuating chamber. B Animals performed a choice task between two stimuli on a touch screen, in which the number of red circles indicated the volume of milk reward and the number of green triangles indicated the amount of intravenous remifentanil reward obtainable by touching the stimulus. The ratio of drug to milk (DM) symbols was used to calculate a reward contrast value as indicated. DM ratios presented to animals and their corresponding reward contrast values are tabulated. C Group average of the last 15 baseline weeks for all subjects (n = 10). Data points represent the group averages (±SEM) for % drug choice at each reward contrast. The sigmoidal choice curve was fitted by logistic regression of % drug and reward contrast.

Fig. 2. Response latencies and stabilty of choice distribution.

A Group response latency (over last 15 baseline weeks) varied across different reward contrasts. Post-hoc analysis revealed significant decreases in response latency for all drug-dominant reward contrasts compared to the zero reward contrast (DM 1:1) using Holm Sidak’s multiple comparison correction. *p < 0.01. B There was no significant correlation of IndV with cumulative remifentanil exposure (session number). Each of 10 subjects is represented by a different color.

Fig. 3. The effects of milk satiation and saline substitution on drug-choice probability.

Data points are group means ± SEM. A Milk satiation increased drug-choice. B Saline substitution (n = 10) decreased drug-choice.

Fig. 4. The effects of μ-opioid agonists and antagonists on remifenanil choice.

A Naltrexone 0.1 mg/kg decreased drug choice, as did B Naltrexone 0.3 mg/kg. C Morphine 1.0 mg/kg did not affect choice allocation.

Fig. 5. Summary figure illustrating the treatment-induced indifference value change from baseline (∆IndV).

A ∆IndV of 0 would indicate no change in drug-choice behavior when compared to baseline. Positive ∆IndV are indicative of decreased drug choice (right curve shift) while negative ∆IndV are indicative of increased drug choice (left curve shift). Saline substitution, naltrexone 0.1 IV, and naltrexone 0.3 IV conditions resulted in a significant positive ∆IndV. Naltrexone effects were different from each other, and both differed from saline. Milk satiation was the only condition that resulted in a significant negative ∆IndV. Data show means ± SEM. *p < 0.05.