Oxytocin Administration Increases Proactive Control in Men with Overweight or Obesity: A Randomized, Double-Blind, Placebo-Controlled Crossover Study

Introduction

Recent advances in nutrition suggest that the hypothalamic neuropeptide oxytocin acts as a critical central-nervous-system factor in mediating food intake and weight. Oxytocin is produced in the paraventricular and supraoptic nuclei of the hypothalamus; decreases food intake in rodent, primate, and human obesity ((1-7)); and induces weight loss with minimal side effects ((2, 8-10)). As a consequence, oxytocin is under investigation as a potential new neurohormonal treatment for obesity. In the United States, 69% of adults have overweight, with 35% meeting criteria for obesity ((11)), and rates of associated health risks, such as diabetes, cardiovascular disease, and premature mortality, are rising. Although complications resulting from obesity may be reversible with successful weight reduction, achieving and maintaining meaningful weight loss is difficult with available pharmacological and lifestyle interventions, and current treatment options are associated with complications ((12)) and safety and tolerability problems ((13)). Thus, there is a clear need for effective, tolerable obesity therapies.

The mechanisms underlying the beneficial effects of oxytocin on food intake and weight in humans are not well understood but are important to unveil to determine the potential of oxytocin as a weight loss treatment. In humans, intranasal oxytocin reduces caloric consumption, particularly of palatable foods, without affecting subjective appetite ((3, 4)). In addition, when hunger- and reward-driven eating were contrasted, oxytocin affected hedonic eating (snacking in a satiety state) more than homeostatic food intake ((6)). These findings indicate that oxytocin might exert some of its effect on food intake in humans through altering eating behavior in response to hedonic urges to eat rather than in response to hunger signals.

Cognitive control, the group of mechanisms that regulate impulses, habits, decision-making, and personal autonomy ((14)), plays a key role in hedonic food intake, and human obesity is characterized by reduced cognitive control, which is linked to poor clinical outcomes. Individuals with obesity, compared with healthy control participants, show impaired cognitive control over impulsive actions, impaired cognitive flexibility, and more risky decision-making ((15-20)). Cognitive control was shown to predict food intake ((21)), and improvements in cognitive control following an 8-week weight loss intervention that included cognitive and behavioral strategies to achieve goals predicted weight loss in individuals with obesity ((16)). In addition, behavioral and neuroimaging studies demonstrate that individuals with obesity, compared with control participants with normal weight, show an increased responsiveness to reward attainment and a stronger bias for (instantly) rewarded behaviors and choices ((22-24)). Increased responsiveness to food reward amplifies the potential consequences of reduced cognitive control in the context of food intake, as it makes individuals with low cognitive control more susceptible to overeating, poor dietary choices, and ultimately weight gain ((25)). These data highlight cognitive control (and its interplay with reward- or habit-driven behavioral impulses) as a promising target in the search for novel approaches in obesity management.

Oxytocin could alter cognitive control. Oxytocin receptors are present in cognitive-control brain regions ((26)), and oxytocin promotes cortical control over subcortical brain structures ((27, 28)). In a randomized, double-blind, placebo-controlled crossover study in 10 men with overweight or obesity, using an exploratory whole-brain analysis, we have shown that a single dose of 24-IU intranasal oxytocin (the same dose that reduced food intake in previous studies ((3, 4, 6, 7))) increased functional magnetic resonance imaging (fMRI) activity in the anterior cingulate cortex and frontopolar prefrontal cortex (both of which have been associated with cognitive control ((29))) with a simultaneous attenuation of fMRI activity in brain areas processing food reward when participants viewed images of palatable food items compared with nonfood objects ((30)). Our findings are supported by Spetter et al. ((31)), who showed that in 15 men with normal weight, 24-IU single-dose intranasal oxytocin altered neural activation in reward-related brain regions and the cognitive-control network during a task requiring the engagement of both reward evaluation and cognitive-control processes. Although alterations of neural activity in the cognitive-control brain network are a promising first indication that oxytocin might affect cognitive control, it remains unclear whether those alterations translate into behavioral effects. In their neuroimaging study, Spetter et al. ((31)) showed that oxytocin also reduced food intake during a breakfast in the same sample by 12%. However, it is unclear whether this reduction in food intake is related to increases in cognitive control, reduced reward responsiveness, or other factors. To prove that the observed oxytocin-induced changes in neural activity within the cognitive-control network translate into behavioral advantages (i.e., reduced impulsive behavior), behavioral testing of cognitive control without involvement of reward processing and other factors that can affect food intake (e.g., homeostatic drive) is required.

We aimed to provide the first behavioral evidence for an effect of oxytocin on cognitive control over behavioral impulses in individuals with overweight or obesity. We asked the same group of men with overweight or obesity who underwent neuroimaging following oxytocin administration ((30)) to also complete a well-established paradigm to test the ability and strategy to control preexisting behavioral impulses: namely, a stop-signal task ((32)). In this paradigm, participants perform a simple choice-reaction task (go task). Occasionally, after the stimulus for the go task is provided, a signal indicates the need to suppress the response for the go task, thus stopping an already generated behavioral impulse from being executed (stop task). Performance on this task reflects an individual’s ability and strategy to control undesired behavioral impulses. First, the duration of the process of suppressing the behavioral impulse can be estimated in this task (stop-signal reaction time [SSRT]), and a shorter SSRT indicates a higher reactive ability to inhibit behavioral impulses ((33, 34)). Second, the stop-signal task is well suited to detect strategic shifts in cognitive control that introduce proactive tonic inhibition of behavioral impulses by increasing the response threshold for the go response ((35-38)). We hypothesized that in men with overweight or obesity, administration of a single dose of 24-IU intranasal oxytocin, compared with placebo, would alter cognitive processing, translating into improved performance in this task. Augmented performance would be revealed by improving reactive control over behavioral impulses (indicated by a reduced SSRT; hypothesis 1) and/or triggering an increase in meta-control capacities through upregulated proactive control by adapting a more cautious behavioral strategy (expressed as an increased response time [RT] in the go task and higher accuracy in stop trials; hypothesis 2).

Methods

Results

Performance in the stop-signal task

Participants displayed slower go RTs following oxytocin administration (M = 936 milliseconds, SD = 263 milliseconds) compared with following placebo administration (M = 833 milliseconds, SD = 278 milliseconds; t[9] = 3.13; P = 0.012; 95% CI: 29 to 178; d = 0.99). In addition, they made fewer stop errors following oxytocin administration (M = 36.41%, SD = 7.67%) compared with following placebo administration (M = 41.15%, SD = 8.86%; t[9] = −2.28; P = 0.049; 95% CI:−9.43% to −0.03%; d = −0.72) (Figure 1). Individual plots of differences in the go RT and stop-error rate between conditions are displayed in Figure 2. The error rate in go trials did not differ between oxytocin (median = 1.22%, interquartile range = 0.52%-2.43%) and the placebo condition (median = 1.74%, interquartile range = 0.52%-3.47%; z = −0.53; P = 0.599; r = −0.12). Similarly, the SSRT was not different between oxytocin (M = 147 milliseconds, SD = 110 milliseconds) and placebo visits (M = 140 milliseconds, SD = 154 milliseconds; t[9] = 0.21; P = 0.840; 95% CI: −70 to 84; d = 0.06).

As this was a pilot study investigating a novel research question, no a priori power analysis was performed. Achieved power was determined using G*Power (version 3.1.9.6; Heinrich Heine University, Düsseldorf, Germany) ((43)). With ten participants in this randomized, double-blind, placebo-controlled crossover study and an observed effect size of d = 0.99 for the difference in response speed in the go task, we had a power of 1 − β = 0.80 to detect a difference at a two-tailed significance level of α = 0.05.

Discussion

To our knowledge, this is the first study to investigate the effects of oxytocin on cognitive-control ability and strategy in individuals with overweight or obesity. In this randomized, double-blind, placebo-controlled crossover pilot investigation in men with overweight or obesity, we demonstrate that a single dose of 24-IU intranasal oxytocin increases proactive control over behavioral impulses in a stop-signal task. These data highlight the modulation of cognitive-control pathways with behavioral effects in the studied population.

After receiving oxytocin, men with overweight or obesity showed slower response times in the go task and a reduced frequency of executing preexisting impulses when prompted not to do so. This response pattern is consistent with a more cautious behavioral strategy (hypothesis 2). The strategic character of this increase in the go-response threshold is further highlighted by the simultaneous reduction of stop errors. As the stop-signal task adjusts to successful suppressions of responses in stop trials by presenting the stop signal 50 milliseconds later at the next occasion, a lower number of stop errors can only be achieved by a continuous slowing that closely follows the automatic calibration of the stop-signal onset that is part of the task ((36)). This response pattern of error aversion and a strategic shift in the go-response threshold has been theoretically derived (goal priority hypothesis ((37))) and repeatedly documented in the literature ((35, 36, 38)). It has been linked to personality traits triggering this response pattern in a subgroup of individuals but not in others in a standard stop-signal task, as applied here, and has also been reported as the result of explicit instructions encouraging participants to adopt a deliberate strategic control shift ((38)). We extend these findings by providing novel evidence that a change in endocrine state caused by intranasal oxytocin administration can induce a similar strategic shift. As participants could not reliably distinguish between oxytocin and placebo when prompted at the end of the study, it can be assumed that the observed behavioral change occurred without deliberation (e.g., via interoceptive signaling).

Although the sample size of this pilot investigation is small, the individual results show that the observed overall pattern is consistently present in eight out of ten participants, suggesting that the reported finding reflects a general effect of oxytocin. Furthermore, an a posteriori power analysis indicated an achieved power of 80% for the observed slowing in the go task. Although replications in larger sample sizes, including studying the role of BMI and sex, represent key next steps, this behavioral evidence for a beneficial effect of oxytocin on cognitive control to regulate behavioral impulses critically extends previous findings from neuroimaging studies showing that (the same dose of) intranasal oxytocin increases fMRI activation in the cognitive-control brain network ((30, 31)), proving that oxytocin can alter control over behavioral impulses in men with overweight or obesity.

The observed findings cannot be explained by a general oxytocin-induced cognitive slowing. In more detail, performance in stop trials is typically described by a horse-race model ((33)). As such, general cognitive slowing would reduce speed of both the go and stop processes, which would result in an unaltered difference in duration between the two. As this difference in durations determines the outcome in stop trials, a general slowing would result in an unchanged stop-error rate, whereas the current results show that the slowing in the go task under oxytocin was accompanied by a decrease in the stop-error rate in this condition.

The SSRT did not differ across conditions; thus, the present study provides no evidence that oxytocin alters the reactive ability to suppress impulses (hypothesis 1). However, the observed strategic trade-off biases the estimation of the SSRT (i.e., the SSRT is estimated on the basis of the distribution of the RTs in the go task together with the stop-error probability) ((36)). Therefore, the present findings cannot conclusively state that oxytocin does not alter reactive inhibition of behavioral impulses, and further studies that suppress strategic trade-offs (e.g., by setting stringent time-out constraints for the go task) will allow for determining the impact of oxytocin on the SSRT (with the limitation of not capturing potential strategic control shifts as detected here).

Oxytocin might trigger a proactive slowing strategy to reduce negative feedback. In the stop-signal task, failure to successfully inhibit a button press following the stop signal (tone) is experienced as frustrating by participants even when clear instructions are provided that occasional failures to suppress the response are part of the task. Avoiding the experience of committing a stop error can gain higher priority when the negative feedback of a failed response inhibition becomes more salient because of oxytocin’s effect on enhancing the salience of socially relevant stimuli (with an unwanted behavioral expression against a presented rule being a social event) ((44)). This idea of a strategic control shift toward more cautious behavior also aligns with the findings of a recent study reporting that oxytocin reduced risk-taking in decision-making during an Iowa Gambling Task in healthy men ((45)).

In summary, this pilot investigation provides the first evidence that oxytocin increases proactive control over behavioral impulses in men with overweight or obesity. The reported link between oxytocin and cognitive control over behavioral impulses opens exciting avenues for future research, as cognitive control can reduce behavioral engagement in urges to eat. This link could thus represent a potential mechanism for the previously reported effects of oxytocin in reducing food intake and weight in humans. Future studies are required to further characterize the effects of oxytocin on cognitive control of behavioral impulses in a larger sample including both men and women and to establish the predictive value and necessity of altered control over behavioral impulses regarding oxytocin’s effects on food intake and, ultimately, weight in human obesity. Furthermore, investigating whether such effects might be modulated by BMI, as observed for the inhibitory impact of oxytocin administration on caloric intake ((7)), and whether the effects might be influenced by metabolic state represent key follow-up research questions.

Funding agencies

This work was supported by the Boston Nutrition Obesity Research Center/NIH grant 5P30DK046200-20, Harvard Catalyst/NIH grant 1UL1TR001102, Nutrition Obesity Research Center at Harvard/NIH grant P30DK040561, and NIH grant K23MH092560. The funding sources had no involvement in the design of the study; collection, analysis, and interpretation of the data; or decision to publish this manuscript.

Disclosure

EAL is on the scientific advisory board and has a financial interest in OXT Therapeutics, a company developing intranasal oxytocin and long-acting analogs of oxytocin to treat obesity and metabolic disease. EAL’s interests were reviewed and are managed by Massachusetts General Hospital and Mass General Brigham in accordance with their conflict of interest policies. This company was not involved in any way in this research. The other authors declared no conflict of interest.

Clinical trial registration

ClinicalTrials.gov identifier NCT02276677.