Potential for limited reinforcing and abuse-related subjective effects of intranasal oxytocin

Abstract

There has been growing interest in the use of oxytocin as a pharmacotherapy for psychiatric disorders, including substance use disorder. Limited data exist regarding oxytocin’s reinforcing efficacy, which is a necessary consideration for novel pharmacotherapies, especially in substance-using populations. This study aimed to determine the potential reinforcing effects of intranasally-administered oxytocin versus placebo by assessing behavioral economic demand and subjective effects. Healthy adults (n = 23) participated in a double-blind, repeated-measures, laboratory study wherein they received intranasal oxytocin (40 IU) or placebo in a randomized order across two sessions. Participants completed hypothetical drug purchasing tasks at the conclusion of both sessions. Throughout both sessions, subjective and physiological effects were assessed. Demand curve analysis of purchasing tasks revealed moderately greater median purchasing for oxytocin relative to placebo at low prices. Physiological and subjective effects did not significantly differ between oxytocin and placebo. However, a non-significant trend was observed for moderately greater drug liking for oxytocin relative to placebo. There was a significant, positive correlation between the difference in drug liking (between oxytocin and placebo) and the difference in lowest-price purchasing (between oxytocin and placebo). These data suggest the potential for limited reinforcing and abuse-related subjective effects of intranasal oxytocin. Given the small sample, the moderately greater drug liking of oxytocin compared to placebo, and the positive relation between demand and drug liking, it is possible that oxytocin may produce reinforcing effects in a subset of participants. Therefore, additional studies of oxytocin reinforcement are warranted.

Introduction

Oxytocin has long been a mainstay pharmacological intervention in gynecological practice as a means for inducing labor or reducing post-partum hemorrhaging (WHO, 2011). In recent years, there has been growing interesting in using intranasally-administered oxytocin for use in treatment of psychiatric disorders. Clinical and preclinical research has demonstrated that oxytocin, whether endogenous or exogenous, can reduce anxiety (reviewed in Naja and Aoun, 2017) and increase prosocial behaviors (reviewed in Yamasue et al., 2012). These qualities suggest oxytocin may serve as a pharmacotherapeutic adjunct to facilitate standard psychiatric therapies for a range of psychiatric disorders. Early genetic data indicated a relation between oxytocin receptor polymorphisms and autism spectrum disorder phenotypes (Campbell et al., 2011), and intranasally-administered oxytocin has since been widely investigated as a potential therapeutic in autism spectrum disorder and has demonstrated improvement in core social and behavioral symptoms associated with the disorder (Anagnostou et al., 2012; Tachibana et al., 2013; Yamasue et al., 2018; Kruppa et al., 2018). Similarly, in individuals suffering from post-traumatic stress disorder, intranasally-administered oxytocin has been shown to attenuate progression of PTSD symptoms when administered early post-trauma (Frijling, 2017), reduce stress-reactivity (Flanagan et al., 2018a), and improve therapeutic alliance and depression symptoms when used as an adjunct to therapy (Flanagan et al., 2018b). In humans, low plasma oxytocin levels are associated with increased prevalence of chronic pain, while intranasal oxytocin administration decreases pain sensitivity experimentally (Rash and Campbell, 2014; Goodin et al., 2015; Anderberg and Uvnas-Moberg, 2000; Alfven, 2004), indicating oxytocin’s potential efficacy in pain management. Further, intranasal oxytocin has demonstrated effectiveness in reducing food-related stress (Russell et al., 2018) and attentional bias to body shape and food-related stimuli (Kim et al., 2014) in patients with anorexia nervosa. Moreover, oxytocin reduces caloric intake in patients with bulimia nervosa (Kim et al., 2015).

Oxytocin has also begun to emerge as a potential therapeutic for substance use disorder (Lee et al., 2016). In rats, systemically-administered oxytocin decreased self-administration of methamphetamine (Carson et al., 2010), cocaine (Kohtz et al., 2018), and heroin (Kovács and Van Ree 1985; Kovács, Borthaiser, and Telegdy, 1985), and reduced methamphetamine-primed reinstatement of methamphetamine self-administration (Carson et al., 2010). Similarly, in mice, intracerebroventricularly-administered oxytocin reduced methamphetamine conditioned-place preference and enhanced the extinction of methamphetamine-conditioned place preference, and both effects were reversed with administration of the OT1-receptor antagonist atosiban (Qi et al., 2009). Clinical studies with oxytocin in substance-using populations have yielded mixed results. In one study in cocaine-dependent individuals on methadone maintenance for opioid use disorder, twice-daily intranasal administration of 40 IU oxytocin over two weeks reduced cocaine craving relative to placebo and maintained heroin craving, as opposed to the increase seen in the placebo-treated group (Stauffer et al., 2016). Acute administration of 40 IU intranasal oxytocin, on the other hand, did not alter opioid craving in another sample of participants receiving opioid-replacement therapy (Woolley et al., 2016). In dependent cigarette smokers, 40 IU intranasal oxytocin reduced the desire to smoke and the total number of cigarettes smoked relative to placebo across two repeated laboratory sessions (Van Hedger et al., 2018a); however, the same dose did not alter stress-induced cigarette craving in another sample of smokers (Van Hedger et al., 2018b). In social drinkers, 40 IU oxytocin, relative to placebo, did not alter the subjective or physiological effects of alcohol (Vena et al., 2018). Similarly, 40 IU intranasal oxytocin did not alter stress-induced alcohol craving in patients with post-traumatic stress disorder with concurrent alcohol use disorder, but did inhibit cortisol reactivity to stress relative to placebo (Flanagan et al., 2018a). These preclinical and clinical data indicate that oxytocin may be an effective pharmacotherapy for treatment of some aspects of substance use disorder in certain populations.

With the emergence of oxytocin into psychiatric clinical practice comes the need for rigorous testing of multiple facets of oxytocin’s effects in order to fully understand potential untoward effects of its use. One important consideration for psychiatric drugs, especially those being used to treat substance use disorders, is the reinforcing aspects, or abuse liability, of the new treatment. One previous study evaluating the subjective effects of oxytocin indicated 40 IU intranasal oxytocin increased participant ratings of “would take again” relative to placebo (Kirkpatrick et al., 2014a), indicating potentially reinforcing effects of oxytocin at this dose. As such, there is a need for understanding oxytocin’s reinforcing effects before seriously considering its clinical utility. Although there are a number of ways to evaluate the abuse liability of novel therapeutics, two simple and quickly-implemented methods are hypothetical purchasing tasks and assessments of subjective effects.

Hypothetical purchasing tasks ask participants to identify how much of a commodity, usually a drug, they would purchase over a specified time frame across a variety of different prices. Resulting consumption can then be modeled using nonlinear regression techniques (Hursh and Silberberg, 2008) to provide various demand metrics related to reinforcement. Demand intensity, for instance, is a measure of consumption at the lowest price, whereas demand elasticity refers to the sensitivity of consumption to increases in unit price. Greater elasticity refers to proportionally greater decreases in consumption at higher prices compared to lower prices, and more inelasticity refers to lesser or no decreases in consumption at higher prices compared to lower prices. By evaluating the interaction between price and consumption, demand curve analysis allows for a multidimensional and robust assessment of reinforcement which conceptually integrates more traditional metrics of reinforcing efficacy (Bickel and Madden, 1999; Johnson and Bickel, 2006). Demand metrics obtained from unblinded drug purchasing tasks, where the participant completes the task about an explicitly-defined substance, have been demonstrated to relate to actual substance use and severity of use for alcohol (Murphy and MacKillop, 2006), cigarettes (MacKillop et al., 2008), cocaine (Bruner and Johnson, 2014), and cannabis (Aston, Metrik, and MacKillop, 2015; Strickland et al., 2017). Similarly, hypothetical purchase tasks can be completed under double-blind conditions to assess the abuse liability of experimentally-administered drugs. Hypothetical purchase tasks for reduced-nicotine content cigarettes administered under double-blind conditions found dose-dependent reductions in demand intensity and increases in elasticity (Smith et al., 2017; Higgins et al., 2017). Similarly, demand analysis of participants’ response patterns after 20 mg amphetamine and placebo in a double-blind, repeated-measures study demonstrated significantly greater demand intensity and reduced elasticity for amphetamine relative to placebo (MacKillop et al., 2018). Because hypothetical purchase tasks are easy to administer, relate to actual drug use patterns, and are sensitive to placebo and dose, they serve as a useful laboratory assessment of drug reinforcement in humans without the need for repeated drug consumption as in a self-administration model.

Although measures of drug consumption are necessary for understanding a drug’s reinforcing properties and predicting the potential for compulsive drug taking, acute subjective effects of novel drugs are important to consider, as these may also drive abuse or, conversely, medication-noncompliance in patients who find the effects unpleasant. Drug subjective effects are often measured in the laboratory with a drug effects questionnaire using a visual analog or Likert scale in response to questions such as “Are you high right now?” or “Do you like the drug effects?” (e.g. Morean et al., 2013). These drug effects questionnaires are highly modifiable and can be used to address a number of different aspects of the drug under investigation. Additionally, the questionnaire can be administered at different time points after drug administration to obtain a time course for various subjective effects of the drug (e.g. Carbonaro et al., 2018; Johnson et al., 2017; Vandrey et al., 2017). Oxytocin’s subjective effects have previously been evaluated using a drug effects questionnaire in conjunction with alcohol (Vena et al., 2018) or as part of a repeated-measures study also evaluating MDMA (Kirkpatrick et al., 2014a). In these studies, most subjective effects were not significantly different from placebo.

In an effort to probe the potential reinforcing effects of oxytocin in the absence of other pharmacological interventions, the current study evaluated the reinforcing and subjective effects of intranasally-administered oxytocin using a hypothetical purchase task and a drug subjective effects questionnaire.

Methods

Participants

All research methods were approved by the Johns Hopkins University School of Medicine Institutional Review Board (Protocol# IRB00133379). Participants were healthy adults (n = 23) from the Baltimore area. Participants were recruited via internet and word-of-mouth advertising. Participants were excluded from the study if they 1) had a serious illness (i.e. serious respiratory, cardiovascular, or neurologic disease), 2) demonstrated serious cognitive impairment (i.e. dementia, development delay disability), 3) were diagnosed with a serious psychiatric disorder, 4) provided a non-zero breath-alcohol concentration (BrAC) reading, 5) had a blood pressure reading greater than 140/90 mmHg, 6) tested positive for illicit substances other than THC, 7) demonstrated current dependence on any drug except caffeine or nicotine as determined using a DSM-V checklist (Hudziak et al., 1993; American Psychiatric Association, 2013), or 8) were pregnant or nursing. Participants received $30 for completing the screening, $50 for each session, and a $100 bonus for completing both sessions. Participants could also receive $100 for referring someone who completed the study.

Screening Procedure

Participants were initially screened via telephone for basic inclusion/exclusion criteria. Initially-qualified participants came to the laboratory for an in-person screening. Participants arrived for the screening at 8:00 AM. Participants were advised to go about their normal routine (i.e. eating breakfast or drinking coffee, if they normally did that in the morning) before the screening. Prior to providing informed consent, participants were administered a breath-alcohol test (Alco-Sensor IV, Intoximeters, Saint Louis, MO). A research assistant then read aloud the informed consent document, which detailed the nature of the study, to the participant and checked for comprehension throughout. Participants were made aware during consent that they would receive oxytocin (and no other drug) in one study session, and placebo in the other session, in a randomized order. Following provision of written informed consent, participants provided a urine sample to screen for drugs of abuse (STAT Dip, Micro Distributing, Belton, TX) and pregnancy in female participants (Instant-View Pregnancy Urine Dip-Strip, ALFA Scientific Designs, Poway, CA). Participants had their vitals (heart rate and blood pressure) taken prior to meeting with a medical practitioner for medical clearance. For safety reasons, participants were required to provide a systolic blood pressure of ≤ 140 mmHg and a diastolic blood pressure of ≤ 90 mmHg in order to qualify. Participants then completed questionnaires regarding their demographic information and a comprehensive history of their drug use. At the completion of the screening, qualified participants received a snack and then began the first administration session. The second session was scheduled at least 7 days after the first session. The timing of second session was scheduled so that dose administration occurred within the same 15-minute time of day as in the first session. On the second session, participants were administered a breath-alcohol test, provided a urine sample, and had their vital measures taken to ensure a zero BrAC, no recent, non-THC illicit drug use, and a safe blood pressure before receiving the second dose.

Drug Administration

Oxytocin and placebo were administered under double-blind conditions. The Behavioral Pharmacology Research Unit Pharmacy prepared oxytocin and placebo intranasal doses before each session. Oxytocin nasal spray solution (Pharmaworld, Zurich, Switzerland) contained 40 IU/mL oxytocin dissolved in 5 mL chlorobutanol hemihydrate with the antifungal preservatives E215 and E218. Placebo solution was Ocean® Saline Nasal Spray (0.65% Saline; Valeant Pharmaceuticals, Bridgewater, NJ). For both placebo and Oxytocin, 2.5 mL solution was transferred into identical 6-mL amber bottles with metered nasal spray pumps. Participants self-administered five 0.1-mL sprays in each nostril with each spray occurring 30 seconds apart over a five-minute period, resulting in a total of 0.5 mL in each nostril.

Hypothetical Drug Purchasing Task

Participants completed a computerized double-blind hypothetical drug purchasing task near the end of oxytocin’s expected time course (between 100 – 120 minutes after dose administration). Participants read the following instructional set prior to making any hypothetical purchases:

The following questions are hypothetical (pretend), but please answer as though the consequences were real. This means you should take into account your current financial situation and any other factors about your current life circumstances when answering.

Imagine that you have finished the study and will spend the next month in your usual home environment. Also imagine that you have the chance to buy today’s drug dose for your own personal use within the next month. You can buy as many doses as you like, but you cannot sell, trade, or give them away, and you cannot save them for more than a month. Other than the fact that the drug is for your own use within the next month, there is no limit to the number of doses you can buy. Please do not buy more than you will use.

The following questions will ask you how many of today’s drug doses you would buy if they were sold at various prices. For each question, enter the number of doses you would buy, and enter zero if you wouldn’t buy any at that price. Please consider each of the questions separately, meaning that if you buy doses in one question, pretend that you don’t have them when you answer the other questions. In other words, when you are answering each question, pretend that it is the only question being asked of you today. If you have any questions, please ask now.

Participants then entered how many doses of the treatment they received during that session they would purchase for use in the next month at the following per-dose prices: $0.01, $0.03, $0.10, $0.30, $1.00, $3.00, $10.00, $30.00, $100.00, $300.00, and $1,000.00. Each per-dose price was presented individually with a brief synopsis of the instructions above the question.

Subjective Effects Questionnaire

At the same time points as vitals were measured (15 minutes pre-administration, 15, 60, and 120 minutes post-administration), participants responded to 9 items measuring subjective drug effects using a computerized visual analog scale ranging from 0 to 100. There were six reference anchors on the visual analog scale: 0 = Not at all, 20 = Possibly mild, 40 = Definitely mild, 60 = Moderately, 80 = Strongly, and 100 = Extremely. Although these anchors were provided, responses were not limited to these anchor points, and could vary by increments of 1 on the scale. The following questions were asked: “Do you feel a rush?”, “Do you feel any drug effect?”, “Do you like the drug?”, “Does the drug have any good effects?”, “Does the drug have any bad effects?”, “How high are you?”, “How drowsy/sleepy are you?”, “Do you feel jittery?”, and “Do you feel stimulated?”.

Physiological Recordings and Safety Monitoring

In order to evaluate drug-induced changes in cardiovascular or respiratory function and monitor participant safety, participant vitals were measured throughout sessions. Blood pressure and heart rate were measured electronically (eQuality 506DN, Criticare Technologies, Waukesha, WI) 15 minutes before dose administration and 15, 60, and 120 minutes after dose administration. At the same time points, respiratory rate was measured by a trained research assistant who counted breaths for 15 seconds during the blood pressure reading, and the number of breaths was multiplied by 4 to obtain a per-minute respiratory rate. If a participant’s systolic blood pressure was < 90 mmHg or ≥ 180 mmHg and/or diastolic blood pressure was ≥ 120 mmHg at any point during the session, a study physician was immediately notified and emergency care was provided. Systolic and diastolic pressure were used for safety criteria, but Mean Arterial Pressure was the primary blood-pressure outcome measured. Similarly, if a participant’s heart rate exceeded their submaximal heart rate (calculated as: 220 – (participant’s age × 0.85)) or was less than 45 beats/minute at any point during the session, a study physician was immediately notified and emergency care was provided.

Data Analysis

Unless otherwise indicated, all statistical analyses were performed using IBM SPSS Statistics for Apple (Version 24, IBM Corp., Armonk, NY).

Median purchasing of oxytocin and placebo was modeled in GraphPad Prism (version 7.0c for Apple, Graphpad Software, La Jolla, California) using the exponential demand equation (Hursh and Silberberg, 2008):

$$\text{log}Q=\text{log}{Q}_0+k(e^{(-}{{}^{\alpha Q_{\mathit{0}}}}^C{}^{)}-1)$$ (Equation 1)

In equation 1, Q is consumption, Q₀ is a measure of demand intensity representing consumption at prices approaching 0, α is an elasticity-related rate constant describing the proportional change in consumption relative to proportional change in price, k is the range of consumption in logarithmic units, and C is price. A fixed value of 5.429 was used for k to describe median consumption for both commodities. In order to allow for modeling with log-transformed data, the first instance of 0 purchasing was transformed to 0.1 and subsequent responses were ignored. The number of doses purchased at $0.01 was used as the measure of demand intensity instead of the equation-derived Q₀ value. Breakpoint was defined as the price at which median purchasing fell to 0. Because no data currently exist for oxytocin self-administration or consumption to guide judgment for unrealistic responding on the purchase task, demand intensity values exceeding 4 standard deviations from the mean were excluded from further analysis. Demand intensity values for both oxytocin and placebo were square-root transformed and compared using a paired-samples t-test among the full sample. To assess potential differences in relative demand intensity between males and females, square-root-transformed demand intensity was compared in a two-way analysis of variance (ANOVA) with Drug (oxytocin vs placebo) as a repeated-measure and Gender as a between-subjects measure.

Responses on each item of the Subjective Effects Questionnaire were analyzed using separate two-way ANOVAs with Drug and Time as repeated measures. On subjective effects for which a main effect of time was detected, individual rating scores at each time point were transformed to the difference in rating relative to baseline for both placebo and oxytocin. For each participant, the greatest difference score was considered the peak-effect rating. In order to evaluate potential gender differences in subjective effects, peak-effect ratings were analyzed using a two-way ANOVA with Drug as a repeated measure and Gender as a between-subjects factor.

Change scores for demand intensity (square-root oxytocin intensity – square-root placebo intensity) and drug liking (oxytocin peak effect rating – placebo peak effect rating) were calculated to evaluate the relation between drug liking and demand using Pearson correlations. The transformed intensity values were selected for analysis because of the severe skew of the untransformed data.

Mean arterial blood pressure, heart rate, and respiration data were analyzed using separate two-way ANOVAs with time and drug treatment as repeated measures.

Results

Participants

Participant demographics are presented in Table 1. Participants were predominately college-educated, young, and White. Males (52%) and females (48%) were represented nearly evenly. All participants reported that their gender identity was the same as their biological sex. More than half of the sample (57%) endorsed illicit drug use within the past year, and 21 participants (91%) endorsed using alcohol in the past year.

Table 1.

Participant demographics

Characteristic	n (%)	Mean (SD)
Male	11 (48)
Female	12 (52)
Race/Ethnicity
White	12 (52)
African American	6 (26)
More than one race	4 (17)
Hispanic	1 (4)
Past-Year Illicit Drug Use	13 (57)
Past-Year Alcohol Use	21 (91)
Age in years		31.2 (12.5)
Education in years		16.1 (1.8)
Income per month in dollars		2080.9 (1032.2)

Hypothetical Purchase Task

Figure 1 shows demand curves based on median purchasing data points, and mean demand intensity generated from the hypothetical purchasing tasks for oxytocin and placebo. When considering the full sample, median purchasing for oxytocin was greater than placebo in the price range up to 30 cents per dose. Median purchasing for oxytocin among the full sample was 3.5 doses per month at $0.01/dose. With increased prices, median oxytocin purchasing conformed to the law of demand by decreasing in a monotonic function with increasing price. Median oxytocin purchasing was highly elastic (α = 0.31) and the breakpoint (price at which median purchasing reached zero) was $0.30. Median purchasing of placebo was 0 across prices, which resulted in an undefined demand curve and precluded calculation of demand elasticity. Forty-three percent of participants (10/23) indicated that they would not purchase even a single dose of oxytocin at any price, 52% (12/23) indicated zero purchasing for placebo at any price, and 35% (8/23) did not purchase either oxytocin or placebo at any price. Among the 15 participants who endorsed nonzero purchasing, 8 indicated greater purchasing of oxytocin relative to placebo, and 7 endorsed greater relative purchasing of placebo. Because there were so many instances of 0 purchasing at any price (12 placebo, 10 Oxytocin), curves were not fit to the individual data for analysis of within-subject changes in demand elasticity. Responses from one participant were omitted from demand analysis for endorsing purchasing 1000 doses of oxytocin at $0.01, which exceeded 4 standard deviations from mean oxytocin intensity (Z = 4.36). No significant difference in square-root-transformed demand intensity between oxytocin and placebo was determined among the full sample (t(21) = 0.018, p = .986, d = 0.004). Analysis with excluded data points did not alter the outcome when using either a paired-samples t-test (t(22) = 0.817, p = .423) nor a non-parametric Wilcoxon Signed Ranks test (Z = 0.455, p = .659). In the two-way ANOVA of square-root-transformed demand intensity by Drug and Gender, there were no main effects of Drug (F(1,20) = 0.000, p = .986, partial η² = 0.000), Gender (F(1,20) = 0.040, p = .844, partial η² = 0.002), or a Drug × Gender interaction (F(1,20) = 0.027, p = .870, partial η² = 0.001).

Figure 1.

Median purchasing and demand curves (left) and mean (±SEM) and individual square-root transformed demand intensity (right) for Oxytocin (filled circles, dashed line/black bars) and Placebo (open circles, solid line/grey bars).

Subjective Effects

Figure 2 shows time courses of the subjective effects of placebo and oxytocin treatment among the full sample. Table 2 shows results from the two-way ANOVAs with Time and Drug as within-subjects measures for each measured subjective effect. Although ratings for oxytocin generally appear higher than for placebo in Figure 1, no main effect of Drug was detected for any of the subjective effects among the full sample. A significant main effect of Time was detected for all subjective effects except for “Do you feel jittery?”. Although not statistically significant, medium effect sizes were observed on the drug-liking question for the Drug main effect (p = .061, partial η² = 0.151) and Drug × Time interaction (p = .072, partial η² = 0.100), with oxytocin showing a trend for greater liking than placebo at all post-administration time points.

Figure 2.

Subjective effect ratings on a visual analog scale (VAS) for Oxytocin (filled circles, dashed line) and placebo (open circles, solid line) over the study time course.

Table 2.

Results of Two-Way ANOVA testing differences in each subjective effect between oxytocin and placebo (Drug) across time points (Time). The first column lists the tested subjective effect and the second column describes the tested main effect or interaction with degrees of freedom (df) in parentheses. Bolded numbers indicate significant effects at α = .05.

Subjective Effect	Tested Effect (df)	F Value	P Value	Partial η2
	Drug (1, 22)	1.021	0.323	0.044
Any Drug Effects?	Time (3, 66)	14.183	<.001	0.392
	Drug × Time (3, 66)	1.510	0.220	0.064
	Drug (1, 22)	3.908	0.061	0.151
Do You Like the Drug?	Time (3, 66)	10.173	<.001	0.316
	Drug × Time (3, 66)	2.439	0.072	0.100
	Drug (1, 22)	2.287	0.145	0.094
Good Effects?	Time (3, 66)	9.858	<.001	0.309
	Drug × Time (3, 66)	1.673	0.181	0.071
	Drug (1, 22)	0.085	0.774	0.004
Any Bad Effects?	Time (3, 66)	4.424	0.007	0.167
	Drug × Time (3, 66)	0.593	0.622	0.026
	Drug (1, 22)	0.457	0.506	0.02
How High?	Time (3, 66)	8.083	<.001	0.269
	Drug × Time (3, 66)	0.494	0.688	0.022
	Drug (1, 22)	0.139	0.713	0.006
How Drowsy/Sleepy?	Time (3, 66)	4.588	0.006	0.173
	Drug × Time (3, 66)	0.487	0.692	0.022
	Drug (1, 22)	0.537	0.71	0.024
How Stimulated?	Time (3, 66)	5.707	0.002	0.206
	Drug × Time (3, 66)	0.909	0.441	0.04
	Drug (1, 22)	0.049	0.827	0.002
Feel A Rush?	Time (3, 66)	9.502	<.001	0.302
	Drug × Time (3, 66)	0.298	0.827	0.013
	Drug (1, 22)	0.648	0.43	0.029
Feel Jittery?	Time (3, 66)	0.729	0.538	0.032
	Drug × Time (3, 66)	0.291	0.832	0.013

Results from the two-way (Drug × Gender) ANOVA of peak subjective effect ratings (except for jittery, which did not show a significant main effect of Time) are presented in Table 3. Similar to the time course data, peak ratings were generally higher for oxytocin relative to placebo, but no significant main effects of Drug, Gender, or Gender by Drug interactions were detected (all ps > .05). However, a moderate effect size of Drug was detected for the drug liking measure (partial η² = 0.135), mirroring the aforementioned Drug by Time analysis, and suggesting a trend toward greater drug liking for oxytocin relative to placebo despite failing to reach significance (p = .084).

Table 3.

Results of Two-Way ANOVA testing differences in each peak subjective effect ratings between oxytocin and placebo (Drug) and males and females (Gender). The first column lists the tested subjective effect and the second column describes the tested main effect or interaction with degrees of freedom (df) in parentheses. Bolded numbers indicate significant effects at α = .05.

Subjective Effect	Tested Effect (df)	F Value	P Value	Partial η2
	Drug (1, 21)	2.030	0.169	0.088
Any Drug Effects?	Gender (1, 21)	0.683	0.418	0.031
	Drug × Gender (1, 21)	0.904	0.352	0.041
	Drug (1, 21)	3.280	0.084	0.135
Do You Like the Drug?	Gender (1, 21)	1.177	0.290	0.053
	Drug × Gender (1, 21)	0.039	0.845	0.002
	Drug (1, 21)	0.681	0.418	0.031
Good Effects?	Gender (1, 21)	0.526	0.476	0.024
	Drug × Gender (1, 21)	0.007	0.933	0.000
	Drug (1, 21)	0.499	0.488	0.023
Any Bad Effects?	Gender (1, 21)	1.130	0.300	0.051
	Drug × Gender (1, 21)	0.178	0.678	0.008
	Drug (1, 21)	0.547	0.468	0.025
How High?	Gender (1, 21)	0.421	0.524	0.020
	Drug × Gender (1, 21)	1.142	0.297	0.052
	Drug (1, 21)	1.583	0.222	0.070
How Drowsy/Sleepy?	Gender (1, 21)	0.781	0.387	0.036
	Drug × Gender (1, 21)	0.010	0.923	0.000
	Drug (1, 21)	1.019	0.324	0.046
How Stimulated?	Gender (1, 21)	1.228	0.280	0.055
	Drug × Gender (1, 21)	1.615	0.218	0.071
	Drug (1, 21)	0.345	0.563	0.016
Feel A Rush?	Gender (1, 21)	0.022	0.882	0.001
	Drug × Gender (1, 21)	0.000	0.984	0.000
	Drug (1, 21)	0.887	0.357	0.041
Feel Jittery?	Gender (1, 21)	2.118	0.160	0.092
	Drug × Gender (1, 21)	2.435	0.134	0.104

A significant positive correlation, illustrated in Figure 3, was determined between the change in transformed demand intensity and change in peak drug liking, r = .605, p = .003. Additional exploratory analyses of the relation between demand and drug liking indicated that among the 8 participants indicating greater purchasing of oxytocin relative to placebo in the purchase task, peak ratings of drug liking were generally, although not significantly, higher for oxytocin relative to placebo (t(7) = 2.031, p = .082, d = 0.96). A similar trend was not detected in the 7 participants endorsing greater relative placebo purchasing (t(6) = 0.484, p = .646, d = 0.21).

Figure 3.

Relation between the change in peak drug liking rating (oxytocin rating – placebo rating) and change in transformed demand intensity (square-root oxytocin intensity – square-root placebo intensity).

Physiology

Neither blood pressure nor heart rate fell outside of the prespecified safe ranges for any participant throughout the study. Time courses of the physiological effects of oxytocin and placebo administration are illustrated in Figure 4. There were no significant main effects of Drug (F(1,22) = 3.240, p = .086, partial η² = .128) or Time (F(3,66 )= 2.098, p = .109, partial η² = .087) or their interaction (F(3,66) = 0.198, p = .897, partial η² = .009) on mean arterial pressure. A significant main effect of time on heart rate was determined, showing a general decrease over time when averaging across drug conditions (F(3,66) = 8.484, p < .001, partial η² = .278), but there was neither a main effect of Drug (F(1,22) = 2.231, p = .150, partial η² = .092) nor a Drug × Time interaction (F(3,66) = 1.586, p = .201, partial η² = .067). Similarly, a significant main effect of Time on respiratory rate was determined, showing a general decrease from predrug to postdrug time points when averaging across drug conditions (F(3,66) = 4.657, p = .005, partial η² = .175). However, there was neither a main effect of Drug (F(1,22) = 0.020, p = .890, partial η² = .001) nor a Drug × Time interaction (F(3,66) = 0.516, p = .673, partial η² = .023).

Figure 4.

Physiological assessments of placebo (open circles, solid line) and oxytocin (filled circles, dashed line) over time. Data presented as mean (±SEM) heart rate (beats/minutes; left), blood pressure (mean arterial pressure; MAP; middle), and respiratory rate (breaths/minute; right).

Discussion

Understanding the reinforcing and abuse-related effects of intranasal oxytocin is essential, given the growing interest in oxytocin’s use in treatment of a broad array of psychiatric disorders. Although considerations of abuse liability are essential for any novel therapeutic agent, it is an especially-important concern in development and assessment of treatment for substance use disorders. Altogether, the current data suggest the possibility that intranasal oxytocin may have potential for limited reinforcing effects. These effects appear to occur without producing any changes in cardiac or respiratory function, and do not differ by gender. We will discuss each of these domains followed by a discussion of study limitations.

Abuse liability was determined using a double-blind hypothetical purchase task and a systematic evaluation of abuse-related subjective effects for both placebo and oxytocin. The data from the hypothetical purchasing task suggest the potential for limited reinforcing effects of oxytocin. In an effort to evaluate low rates of reinforcement, we assessed oxytocin purchasing over the course of a month instead of the commonly-used single-day assessment (e.g. Bruner and Johnson, 2014). Although single-day assessments are useful in populations who use a drug or drug class on a daily basis (e.g., tobacco use in daily cigarette smokers), use of a month-long period allows for detection of a broader range of consumption patterns, including low-frequency, sub-daily drug use. This expanded time frame allowed the present study to determine that median consumption of oxytocin at the lowest price evaluated was 3.5 doses over the course of a month. This level would likely have been missed if only daily use was assessed, as is typical in drug purchasing tasks. Hypothetical purchasing of oxytocin was highly elastic (i.e. price-sensitive), as median consumption rapidly decreased as a function of price and fell to 0 (breakpoint) at a per-unit price of $0.30. Although this pattern of demand is not comparable to highly-reinforcing drugs like nicotine, alcohol, or cocaine, that oxytocin produces a distinct, orderly pattern of hypothetical demand at a greater level than placebo was surprising and indicative of potentially weak reinforcing effects.

Similar to the limited demand, ratings of subjective-effect strength elicited by oxytocin administration were generally fairly low, and these ratings did not significantly differ from placebo; however, medium-sized trends toward greater ratings for oxytocin relative to placebo were encountered for drug liking and good drug effects, further indicating the potential for reinforcement or abuse liability. Given that there were significant main effects of time, but not drug, across most subjective effects, it is likely that many participants’ responses were influenced by expectancy, or placebo, effects, which mirrors some of the purchasing task responses in which a subset of participants endorsed purchasing of both placebo and oxytocin. These findings are generally in line with other reports evaluating the subjective effects of intranasal oxytocin administration (Kirkpatrick et al., 2014a; Vena et al., 2018). In one study, reinforcement-related subjective effects (i.e. “Do you feel high?”) did not differ between placebo and oxytocin at 20 or 40 IU; however, 40 IU oxytocin did produce significantly lower ratings of drug disliking relative to placebo (Kirkpatrick et al., 2014a; supplemental table 1). In another study, the subjective effects of 40 IU oxytocin did not significantly differ from placebo (Vena et al., 2018). Although the subjective ratings for oxytocin in the current and previous studies were generally low, it is worth noting that some medium-sized effects suggesting greater abuse-related subjective effects for oxytocin relative to placebo were encountered in the current study. These data suggest that 40 IU intranasal oxytocin may not alter all users’ subjective states in a manner that may promote compulsive abuse or decrease medication compliance; however, the potential reinforcing effects of oxytocin are not to be dismissed.

The positive correlation between relative drug liking and relative demand intensity provides further evidence that the trends suggesting oxytocin reinforcement were not random, and that greater purchasing of either treatment is associated with better relative subjective experiences. These results highlight that, for a subset of individuals, oxytocin may have some moderate subjective and reinforcing effects, and, for others, there is a pronounced placebo effect. This conclusion is supported by our exploratory analyses indicating greater oxytocin drug-liking in participants who purchased more oxytocin than placebo, which is bolstered by the lack of a similar effect in placebo-purchasing participants. Although relative drug liking and purchasing varied substantially within the sample, the individuals who indicated both a strong relative liking and a high degree of relative demand for oxytocin raise some concern over the potential for reinforcement in a small subset of individuals, especially when considering previous findings that participants in another study indicated a greater likelihood of using 40 IU oxytocin again relative to placebo (Kirkpatrick et al., 2014a). The possibility remains, however, given the comparable demand intensity but divergent subjective effects between placebo and oxytocin, that oxytocin may produce positive subjective effects without necessarily promoting compulsive drug seeking. Additional studies with larger samples and/or in participants with substance use disorder are warranted to fully elucidate potential reinforcing effects of oxytocin or risk factors associated with differential responding to oxytocin treatment and to determine whether the medium-sized, yet statistically nonsignificant effects in the current study are indicative of a Type II error. Future studies investigating oxytocin for psychiatric treatments, especially substance-use disorder, should be cautious and aware of these potential reinforcing effects.

In the current study, no effects of intranasal oxytocin on physiological vital signs were detected. These data replicate previous reports indicating no differences between intranasal oxytocin and placebo on physiological vital signs of blood pressure and heart rate (Norman et al., 2011; Kirkpatrick et al., 2014a; Van Hedger et al., 2018a; Vena et al., 2018). Bolus intravenous administration of oxytocin has previously been demonstrated to reduce blood pressure and increase heart rate (Weis et al., 1975; Langesæter, Rosseland, and Stubhaug, 2006; Rosseland et al., 2013); however, these effects are typically short-lived and return to baseline/placebo levels within a few minutes after administration. Because we did not continuously monitor cardiac activity, it is possible that we missed these immediate, short-lived changes under the current data-collection timeframe. When considering that no changes in cardiac functioning have been detected following prolonged administration of intranasal oxytocin (Tachibana et al., 2013; Busnelli et al., 2016), the current data further suggest the safety and tolerability of intranasal oxytocin for use in clinical psychiatry.

There were no differences in demand for or subjective response to placebo and oxytocin between males and females in the current study. Although sex differences in treatment response in animals and underlying oxytocinergic neurophysiology in both humans and animals and have previously been reported (reviewed in Dumais and Veenema, 2017; Borland et al., 2019), there have been mixed results for response to intranasal oxytocin administration between males and females in clinical studies. Oxytocin seems to generally increase responses to positive social experiences in women but not men, an effect that is, at least in-part, mediated by differential amygdala reactivity (e.g. Gao et al., 2016; Ma et al., 2018; Luo et al., 2017); however, this gender difference may be dose-dependent with males requiring a higher dose to achieve a comparable effect (Borland et al., 2019). Oxytocin also increases physiological and behavioral response to social stressors, but gender differences are less consistent with one study indicating more-pronounced stress reactivity in females relative to males (Reed et al., 2019) and another indicating no gender difference (Romney et al., 2018). Vena et al. (2018) demonstrated a three-way interaction between Drug, Time, and Gender, such that males’ ratings of alertness did not decrease after oxytocin administration but did after placebo, whereas females’ ratings decreased after both treatments. When considered with the larger literature, the current data suggest that gender differences in response to intranasal oxytocin administration may be context-specific, such that oxytocin modulates responses to social cues differently in females versus males, but the acute subjective and reinforcing effects are generally comparable between sexes.

There are a number of limitations to consider in the current study. First, as mentioned elsewhere, the sample size was fairly-limited with only 23 participants completing the study, which may have been underpowered for detection of significant, yet subtle, drug-related changes, especially on certain subjective-effects measures. The placebo (saline) used in this study was not identical to the oxytocin vehicle (chlorobutanol hemihydrate), which may have influenced subjective responses to each treatment; however, no participants mentioned differences in taste or sensations related to the sprays across sessions. Additionally, the primary outcomes in this study relied on self-report of subjective effects and purchasing under hypothetical conditions. Subjective effects, by nature, are dependent on self-report. Variations of the subjective effects questionnaire, using either Likert or Visual Analog Scales, have been used to reliably evaluate dose-dependent, abuse-related subjective effects of a variety of drugs such as cocaine (Johnson et al., 2017), psilocybin (Carbonaro et al., 2018), MDMA (Kirkpatrick et al., 2014a; 2014b; Bershad et al., 2016), THC/cannabis (Wardle, Marcus, and de Wit, 2015; Vandrey et al., 2017; Cone et al., 2015) and amphetamine (de Wit, Uhlenhuth, and Johanson, 1986; Wardle, Marcus, and de Wit, 2015). The purchase task is limited by its hypothetical nature, which makes it an indirect assessment of reinforcing value. Although a self-administration task would have provided a direct measure of reinforcement, allowing potentially unfettered access to a drug with fairly-limited behavioral and abuse-liability data poses a serious ethical quandary in healthy and substance-using participants alike. Previous studies using hypothetical purchase tasks have found strong correlations between demand metrics on the task and actual drug use (Aston, Metrik, and MacKillop, 2015; MacKillop et al., 2008; Bruner and Johnson, 2014), severity of drug use (Strickland et al., 2017; Murphy and MacKillop, 2006; MacKillop et al., 2008), and purchasing tasks for real and potentially-real outcomes (Wilson et al., 2016), indicating that hypothetical demand is a useful construct for approximating actual drug use. The study was also limited to healthy participants, so it is unknown whether subjective or reinforcing effects may differ in a population with substance use disorders. It is worth noting that 56.5% and 91.3% of our sample had used illicit substances and alcohol, respectively, within the past year, so these were not drug-naïve participants who would be entirely insensitive to reinforcing drug effects. Finally, the current study was limited to a single dose of oxytocin. We chose to examine 40 IU as this is the dose of oxytocin most-commonly used in previous studies evaluating its use in treatment of substance use (Van Hedger et al., 2018a; Vena et al., 2018; Flanagan et al., 2018a) or psychiatric disorders (e.g. Kim et al., 2015; Busnelli et al., 2016; Flanagan et al., 2018b); however, in order to fully understand the reinforcing nature of oxytocin, testing of a broader range of doses is necessary.

The limitations of the current data highlight the need for additional studies regarding the behavioral pharmacology of oxytocin. Ideal follow-up studies would replicate the current methods in a larger sample and expand upon them with the rigorous methods for drug abuse liability outlined in Griffiths, Bigelow, & Ator (2003). Additional assessments might include, but not be limited to, evaluating multiple doses of oxytocin (such as 20, 40, 80, and 160 IU) against placebo, utilizing a positive drug control with moderately reinforcing effects, such as caffeine or modafinil (Garrett & Griffiths, 2001; Dolder et al., 2018), and comparing the effects in multiple populations, such as healthy volunteers against clinical populations of interest in oxytocin research (i.e. patients with substance use disorder, PTSD, or anorexia). Although the abuse-related differences between oxytocin and placebo failed to reach significance in the current analysis, the trends therein suggested the potential for reinforcing effects, at least among a subset of individuals. Given the small sample size, the orderly demand patterns, the medium effect sizes of oxytocin’s greater abuse-related subjective effects relative to placebo, and the strong correlation between drug liking and demand, it is critical that additional studies evaluating oxytocin’s abuse-related aspects be conducted in larger samples and among different populations.