ARTICLE

Neural Responses to Intranasal Oxytocin in Youths

With Severe Irritability
Soonjo Hwang, M.D., Ji-Woo Suk, Ph.D., Harma Meffert, Ph.D., Arica Lerdahl, B.A., William F. Garvey, M.S.,
Ryan Edwards, M.D., Alison Delizza, Ph.D., Brigette Soltis-Vaughan, A.P.R.N., Katrina Cordts, Ph.D., Ellen Leibenluft, M.D.,
R.J.R. Blair, Ph.D.

Objective: The authors investigated the neural impact of completed pre- and posttreatment functional MRI (fMRI)
intranasal oxytocin on emotion processing areas in youths scans with the affective Stroop task.
with severe irritability in the context of disruptive mood and
behavior disorders. Results: Youths who received oxytocin showed significant
improvement in CGI-S and CGI-I ratings compared with
Methods: Fifty-two participants with severe irritability, as those who received placebo. In the fMRI data, blood-
measured by a score ≥4 on the Affective Reactivity Index oxygen-level-dependent (BOLD) responses to emotional
(ARI), with diagnoses of disruptive behavior disorders (DBDs) stimuli in the dorsomedial prefrontal cortex and posterior
and/or disruptive mood dysregulation disorder (DMDD) were cingulate cortex were significantly reduced after oxytocin
randomly assigned to treatment with intranasal oxytocin or compared with placebo. These BOLD response changes
placebo daily for 3 weeks. Assessments were conducted at were correlated with improvement in clinical severity.
baseline and at the end of the trial; the primary outcomes
were measures of irritability on the ARI and ratings on the Conclusions: This study provides initial and preliminary evi­
Clinical Global Impressions severity scale (CGI-S) focusing dence that intranasal oxytocin may induce neural-level
on DBD and DMDD symptoms, and secondary outcomes changes in emotion processing in youths with irritability in
included the CGI improvement scale (CGI-I) and ratings of the context of DBDs and DMDD. This may lead to symptom
proactive and reactive aggressive behavior on the Reactive- and severity changes in irritability.
Proactive Aggression Questionnaire. Forty-three partici­
pants (22 in the oxytocin group and 21 in the placebo group) Am J Psychiatry 2024; 181:291–298; doi: 10.1176/appi.ajp.20230174

Irritability is a significant mental health issue in children and of disruptive behavior disorders (DBDs) (9). Decreased oxy­
adolescents (1) and is one of the most common reasons tocin levels have been reported in saliva or serum of children
children are referred for psychiatric treatment (2, 3). Children with DBDs (oppositional defiant disorder or conduct disorder)
with chronic irritability are at increased risk for long-lasting (10, 11). Notably, lower levels of serum oxytocin in children with
mental health problems, including anxiety and depressive attention deficit hyperactivity disorder (ADHD) correlate with
disorders, suicide, and substance use disorders (4, 5). their degree of aggressive behavior (12).
There are few evidence-based pharmacologic treatment In neuropsychiatric studies, emphasis is placed on the
options for irritability in children and adolescents. Fur­ ability to measure both symptom change and the engagement
thermore, very few studies have explored how any treat­ of the neural target of the intervention (13). However, rel­
ment modality impacts neural areas implicated in the atively little work has focused on the actual neural impact of
pathophysiology of irritability (6). Several studies have sug­ oxytocin, especially in youths with irritability (13, 14). No­
gested that the neuropeptide hormone oxytocin may be ef­ tably, one form of neurocognitive dysfunction in adolescents
fective in inducing neural changes in youths with irritability. with irritability is an atypically increased activation of neural
Oxytocin has been used in clinical trials of psychiatric diagnoses regions to emotional cues, including the medial prefrontal
associated with elevated risk for irritability, such as borderline cortex (mPFC), anterior cingulate cortex (ACC), posterior
personality disorder (7) and autism spectrum disorder (8). cingulate cortex (PCC), and amygdala (1, 15). Previous
Several studies implicate changes in oxytocin levels related to work with oxytocin (both single-dose administration and
irritability in children and adolescents, especially in the context clinical trials) has indicated that oxytocin reduces neural

See related features: Editorial by Dr. Stringaris and Ms. Silver (p. 259), AJP Audio (online), and Video by Dr. Pine (online)

Am J Psychiatry 181:4, April 2024 ajp.psychiatryonline.org 291

INTRANASAL OXYTOCIN FOR SEVERE IRRITABILITY IN YOUTHS

responsiveness to emotional cues in areas including amyg­ School-Age Children–Present and Lifetime Version, admin­
dala and mPFC in healthy individuals as well as in patients istered by a licensed and board-certified child and adolescent
with borderline personality disorder, anxiety disorders, psychiatrist and/or an advanced practice psychiatric nurse.
and posttraumatic stress disorder (16, 17). Hence, a po­ For inclusion and exclusion criteria, see the online supplement.
tential treatment target for irritability, atypically increased
neural responsiveness of areas including mPFC to emo­ Assessments
tional cues, may be modulated by oxytocin. The clinical severity of irritability was assessed with the ARI
Our goal in this study was to determine the impact at the screening and at the final assessment session after
of 3 weeks of daily intranasal oxytocin on 1) irritability 3 weeks of study treatment. Additional outcome measures,
symptom severity in youths with severe irritability in the including the Inventory of Callous-Unemotional Traits and
context of disruptive behavior (in DBDs) and mood disorders the Reactive-Proactive Aggression Questionnaire (RPAQ),
(in disruptive mood dysregulation disorder [DMDD]), since were completed at both time points. For descriptions of the
these disorders seem to share a common neurobiological symptom profile measurements, see the online supplement.
mechanism of irritability (compared to, for example, autism The Clinical Global Impressions severity and improvement
spectrum disorder) (1, 18); and 2) the neural profile relating scales (CGI-S and CGI-I) (19) as well as the Children’s Global
to emotional responding as indexed by a task (the affective Assessment of Severity (CGAS) (20) documented the clini­
Stroop task) previously demonstrated to identify dysfunction cians’ global assessment of the youths’ symptoms of DBDs
in neural responsiveness associated with irritability (mainly and DMDD, including irritability, and were completed at
increased response to emotional stimuli in the dorsomedial screening and at the final assessment session (the CGI-I was
prefrontal cortex [dmPFC] and PCC in youths with significant completed only at the final assessment session). The CGI was
levels of irritability) (15). The study used clinical mea­ administered by a psychiatrist with extensive experience in
surements of severity before and after oxytocin interven­ providing assessment and treatment in this population, blin­
tion in the context of randomized, double-blind administration ded to the participants’ treatment assignment. (For details on
for 3 weeks. Based on previous work, we hypothesized that the randomized double-blind administration of intranasal
intranasal oxytocin would reduce the severity of irrita­ oxytocin, see the online supplement.)
bility in youths with DBDs and DMDD and that the level of
symptom severity would be associated with the level of re­ Statistical Analysis of Clinical Trial Data
duction of neural responses in the mPFC and PCC to emo­ We first examined group differences in symptom profiles
tional stimuli under the affective Stroop task. (ARI, irritability, Inventory of Callous-Unemotional Traits,
proactive aggressive behavior, reactive aggressive behavior,
and total aggressive behavior in the RPAQ) and clinical se­
METHODS
verity (CGI-S and CGAS, focused on symptoms of DBDs and
Participants DMDD) at baseline by t test. We then conducted 2×2 repeated-
The study was conducted at an outpatient child and ado­ measures analyses of covariance (ANCOVAs) (group [oxytocin,
lescent psychiatry clinic at a large academic medical center placebo] by time [pre- and posttreatment]) on the symptom
in the Midwest from January 2016 until August 2021. The profile and clinical severity measures. T tests were conducted on
center’s institutional review board reviewed and approved the CGI-I scores at the final assessment, after 3 weeks of treat­
the study. The participants were between 10 and 18 years of ment, between participants in the oxytocin and placebo groups.
age, had diagnoses of ADHD, oppositional defiant disorder,
conduct disorder, and/or DMDD, and had significant levels Functional MRI Scans
of irritability, as indicated by a score ≥4 on the Affective Functional MRI (fMRI) sessions took place after the initial
Reactivity Index (ARI). All participants provided written assessment session but before initiation of intranasal oxy­
assent, and their parents written informed consent. Partic­ tocin or placebo (average time between first MRI scan and
ipants were recruited by referrals from outpatient clinic initiation of treatment, 1.4 days [SD=0.52]) and after the final
providers and the surrounding community by local adver­ postintervention assessment session 3 weeks later (average
tisements. The study was registered at ClinicalTrials.gov time between end of treatment and second MRI scan, 2.8
(NCT02824627) with two primary outcomes (the ARI and days [SD=0.74]). The research personnel who performed
the Clinical Global Impressions severity scale; see below). A the MRI procedure were also blinded to participants’
data safety monitoring board was formed at the medical center treatment assignment. For the youths with DBDs, psychi­
to monitor the integrity of the conduct of the study. Figure 1 atric treatment continued at the outpatient clinic or with
illustrates the flow of participant screening and recruitment. their local providers after study completion. For more details
on the fMRI scans, see the online supplement.
Screening Twenty-two participants in the oxytocin treatment group
Psychiatric diagnoses were confirmed using the semistruc­ and 21 in the placebo group completed both the pre- and
tured Schedule for Affective Disorders and Schizophrenia for posttreatment fMRI scans. Two participants in the oxytocin

292 ajp.psychiatryonline.org Am J Psychiatry 181:4, April 2024

 HWANG ET AL.

group and three in the pla­ FIGURE 1. Participant flow in a study of intranasal oxytocin treatment for youths with disruptive
cebo group completed only behavior disorders
one scan, as a result of
scheduling conflicts, ma­ Records identified through honest broker and provider referrals from 2017 to 2021 (N=2,069)

chine malfunctioning, or lack

Duplicate records excluded (N=1,288)
of interest. One participant
in the oxytocin group and Individuals screened for inclusion and exclusion criteria (N=781)
three in the placebo group
were excluded from the an­ Did not meet criteria (N=632)
alyses because of excessive
Individuals for whom permission to contact was requested of providers (N=149)
in-scanner motion. We in­
cluded only the data from Individuals not approved by provider (N=13)
these 39 participants (21 in
the oxytocin group and 18 in Individuals for whom providers approved contact (N=136)
the placebo group) in the fi­
Not interested or unable to contact (N=77)
nal MRI data analyses.
Enrolled (N=59)
Statistical Analysis of
fMRI Data Screening failure (low IQ) (N=1)
Behavioral data. We con­
Oxytocin (N=28) Placebo (N=30)
ducted a full 2×2×3×3
repeated-measures ANCOVA Withdrew (adverse Withdrew (adverse
(group [oxytocin, placebo] by event) (N=1), event) (N=2),
time [pre- and posttreatment] Lost to follow-up (N=2) Lost to follow-up (N=1)

by condition [incongruent,
Completed treatment on Completed treatment on
congruent, view] by emotion oxytocin (N=25) placebo (N=27)
[negative, positive, neutral]) on
the accuracy and reaction time
data, with sex as a covariate.

MRI data. We selected two approaches to the data analyses following a common approach (22). Similar to the method we
of blood-oxygen-level-dependent (BOLD) responses. First, implemented in a previous work (23), this approach would
we conducted a 2×2×3×3 ANCOVA (group [oxytocin, allow us to focus on the role of the ACC in the treatment
placebo; between-subject variable] by time [pretreatment response with oxytocin but be unbiased with respect to pre-
and follow-up; within-subject variable] by condition [in­ to-posttreatment differences among participants in the
congruent, congruent, and view] by emotion [negative, present study.
positive, and neutral]) on the whole-brain BOLD response
data, using the 3dMVM program in AFNI, with sex as a Relationship of treatment-related BOLD response changes to
covariate. Correction for multiple comparisons was per­ symptom-level changes. These were examined via correla­
formed using a spatial clustering operation in AFNI’s tional analyses. Differential BOLD responses (posttreatment
3dClustSim program, utilizing the auto-correction function minus pretreatment) in the neural areas showing a signifi­
(-acf ) with 10,000 Monte Carlo simulations for a whole-brain cant group-by-time interaction and in the ACC ROI were
gray matter mask. The initial significance threshold was set at correlated against changes in symptom levels (posttreatment
a p value of 0.001 (21). This procedure yielded an extent minus pretreatment).
threshold (k) of 25 voxels, which then resulted in a cluster-level
false positive probability of 0.05, corrected for multiple com­
RESULTS
parisons. To facilitate future meta-analytic work, effect sizes
(partial η2) for all clusters are reported. Clinical Characteristics
Second, we focused on the ACC region of interest (ROI). Twenty-eight participants were assigned to treatment
As noted, previous work with the same affective cognitive with intranasal oxytocin and 30 participants to placebo.
task (the affective Stroop task) has indicated atypical ACC Twenty-five participants in the oxytocin group completed
response to negative emotional stimuli in youths with severe the 3 weeks of study treatment. One participant withdrew
levels of irritability. The ACC ROI was generated from our because of an adverse event, and two were lost to follow-up.
previous study (15) of a large number of youths with severe Twenty-seven participants in the placebo group completed
levels of irritability (N=155, none of whom participated in the 3 weeks of study treatment. Two participants withdrew
the present study) (59 voxels; coordinates: −7.5, 28.5, 11.5), because of adverse events, and one was lost to follow-up.

Am J Psychiatry 181:4, April 2024 ajp.psychiatryonline.org 293

INTRANASAL OXYTOCIN FOR SEVERE IRRITABILITY IN YOUTHS

TABLE 1. Demographic and clinical characteristics of participants in a study of oxytocin treatment η2p =0.053), self-reported
for youths with disruptive behavior disordersa total aggression (F=2.13,
Oxytocin Treatment Placebo p=0.151; η2p =0.041), and
Group (N=25) Group (N=27) CGAS score (F=3.85, p=0.055;
Characteristic Mean SD Mean SD p η2p =0.072) showed reduc­
Age (years) 14.5 2.2 14.8 1.9 0.53 tions in the oxytocin group
IQ 100.7 17.3 99.8 14.5 0.85 compared with the placebo
N % N % p group, but the differences
were not statistically signif­
Sex 0.07
Male 17 68.0 24 88.9
icant. The number needed to
Female 8 32.0 3 11.1 treat, calculated from the
Primary diagnosis youth-reported ARI with the
ADHD 15 60.0 14 51.9 criterion of significant re­
DMDD 6 24.0 7 25.9 duction as ≤−4, was 5.7,
Conduct disorder 2 8.0 0 0.0 with an absolute risk reduc­
Otherb 2 8.0 6 22.2
tion of 17.5%. There was
Secondary diagnosis no significant difference
ADHD 6 24.0 6 22.2
Oppositional defiant disorder 13 52.0 15 55.6
in symptom reductions in
Conduct disorder 3 12.0 1 3.7 callous-unemotional traits
DMDD 5 20.0 7 25.9 or proactive aggression. For
Otherb 8 32.0 12 44.4 details, see Table 2.
a
ADHD=attention deficit hyperactivity disorder; DMDD=disruptive mood dysregulation disorder.
b
The “other” category includes mood disorder not otherwise specified, bipolar disorder not otherwise specified, anxiety
Adverse Events
disorder not otherwise specified, generalized anxiety disorder, social phobia, panic disorder, and social anxiety
disorder.
There were no significant
differences in types of adverse
events between participants
Table 1 summarizes the participants’ demographic and who received intranasal oxytocin and those who received
clinical characteristics, and Table 2 lists the treatment placebo (χ2=10.81, p=0.37) (for adverse event profiles, see
groups’ mean scores on assessment instruments at base­ Table S2 in the online supplement). All three participants
line and follow-up. None of the participants had any changes who discontinued the trial as a result of adverse events
in their psychiatric medication regimens or required higher experienced mood changes (two in the placebo group be­
levels of care (such as inpatient psychiatric hospitalization) came more irritable, and one in the oxytocin group became
during their participation in the study. (For a list of the more anxious), which were completely resolved when
psychiatric medications participants were taking during the assessed by telephone follow-up 2–3 weeks after their
study period, see Table S1 in the online supplement.) withdrawal. Of the adverse events more broadly de­
scribed, drowsiness was the most common (oxytocin
Symptom Profile Changes group, 40.0%; placebo group, 40.7%), followed by head­
Our repeated-measures ANCOVA revealed that there was ache (oxytocin group, 12.0%; placebo group, 18.5%) and
a significant group-by-time interaction in CGI-S scores mood changes (oxytocin group, 4.0%; placebo group,
(F=5.14, p=0.028; η2p =0.093) and that clinical severity was 22.2%).
significantly decreased for the oxytocin group compared
with the placebo group (4.44–3.40 and 4.59–4.04, respec­ Behavioral Data
tively; the scale ranges from 1 [normal, not at all ill] to 7 [the For the results of the behavioral data analyses, see the online
mostly extremely ill]). The t test for CGI-I scores also showed supplement.
significant improvement in the oxytocin group (mean=2.72,
SD=0.79) compared with the placebo group (mean=3.33, MRI Data
SD=1.07) (t=2.33, p=0.024) (the scale ranges from 1 [very Whole-brain analysis. Regions showing significant group-by-
much improved] to 4 [no change] to 7 [very much worse]). time-by-condition interactions included the left dmPFC, left
The number needed to treat, calculated from CGI-I scores PCC, right fusiform gyrus, left and right caudate, and left
with the criterion for significant change as a score of 1 (very thalamus/anterior nucleus (see Table 3). Within all these
much improved) or 2 (much improved), was 3.4, with an regions, participants in the oxytocin group showed a sig­
absolute risk reduction of 29.5%. nificantly greater decrease in activation in response to view
However, irritability as measured on the youth-reported trials after treatment relative to those in the placebo group
ARI did not reach statistical significance in the group-by- (t values, 2.20–3.71; p values, <0.001–0.02).There were no
time interaction (F=2.57, p=0.115; η2p =0.049). In addition, regions showing group-by-time-by-emotion or group-by-
self-reported reactive aggression (F=2.79, p=0.101; time-by-condition-by-emotion interactions.

294 ajp.psychiatryonline.org Am J Psychiatry 181:4, April 2024

 HWANG ET AL.

TABLE 2. Symptom profiles at baseline and follow-up among youths with disruptive behavior disorders in a study of intranasal oxytocin
treatmenta
Oxytocin Treatment Group (N=25) Placebo Group (N=27)
Baseline Follow-Up Baseline Follow-Up
Scale or Subscale Mean SD Mean SD Mean SD Mean SD p η2p
Affective Reactivity Index
Youth-reported 6.64 2.22 4.24 3.09 6.52 2.67 5.37 3.12 0.12 0.049
Parent-reported 7.52 2.73 4.32 2.67 8.67 1.98 6.07 3.44 0.53 0.008
Inventory of Callous-Unemotional Traits
Youth-reported 27.14 10.07 25.55 9.44 27.39 9.35 25.89 9.92 0.96 0.000
Parent-reported 35.13 12.25 32.24 11.67 35.14 8.93 32.85 12.52 0.80 0.001
Reactive-Proactive Aggression Questionnaire
Youth-reported 15.40 5.78 9.92 7.43 16.38 7.71 13.00 6.51 0.15 0.041
Parent-reported 20.64 7.09 12.08 6.95 22.33 6.45 15.87 10.25 0.37 0.016
Reactive aggression
Youth-reported 12.36 3.97 8.08 5.26 12.56 5.25 10.30 4.52 0.10 0.053
Parent-reported 15.25 3.76 9.52 4.45 16.04 3.71 11.78 6.46 0.29 0.023
Proactive aggression
Youth-reported 3.04 2.73 1.82 2.66 3.80 2.82 2.69 3.37 0.86 0.001
Parent-reported 5.40 3.82 2.60 2.96 6.30 3.11 4.11 4.41 0.59 0.006
Children’s Global Assessment 40.00 7.31 48.60 7.84 38.67 4.46 43.59 6.73 0.06 0.072
of Severity
CGI severity score 4.44 0.58 3.40 0.65 4.59 0.69 4.04 0.81 0.03 0.093
CGI improvement score 2.72 0.79 3.33 1.07 0.024
a
CGI=Clinical Global Impressions scale.

TABLE 3. Brain regions showing significant interactions

Coordinates of Peak Activationb
a
Interaction and Region Side BA x y z F Voxels η2
Group×time×condition
Dorsomedial frontal gyrus Left 8 −13.5 25.5 41.5 15.63 98 0.183
Posterior cingulate cortex Left 29 −1.5 −40.5 14.5 15.25 102 0.136
Fusiform gyrus Right 37 31.5 −46.5 −9.5 12.65 27 0.102
Caudate Right 16.5 19.5 8.5 12.65 38 0.100
Caudate Left −10.5 10.5 2.5 18.97 36 0.098
Thalamus/anterior nucleus Left −1.5 −1.5 11.5 18.00 46 0.172
Group×time×condition×emotion
Anterior cingulate cortex ROI Left 24 −7.5 28.5 11.5 6.68 59 0.164
a
Regions are based on the Talairach Daemon Atlas (http://www.nitrc.org/projects/tal-daemon/). BA=Brodmann area; ROI=region of interest.
b
Based on the Tournoux and Talairach standard brain template.

ROI analysis. Our ROI analysis of the ACC showed a sig­ associations between BOLD response changes in the
nificant time-by-group-by-condition-by-emotion interaction; dmPFC, PCC, and ACC and changes (reduction) in irri­
participants in the oxytocin group showed a significantly tability, reactive aggression, and total aggression (ρ values,
greater decrease in activation in response to negative view 0.32–0.64; p values, <0.001–0.038) (for details, see Table
trials after treatment relative to those in the placebo group S4 and Figure S3 in the online supplement).
(t=3.13, p=0.002) (Figure 2).
DISCUSSION
Correlation between BOLD response changes and symptom
improvement. We examined the correlation between BOLD To our knowledge, this is the first study reporting the neural
response changes (posttreatment vs. pretreatment) and impact of intranasal oxytocin intervention for youths with
symptom changes (posttreatment vs. pretreatment reduc­ significant levels of irritability in the context of disruptive
tion) among youths with DBDs, using the symptom profiles mood and behavior disorders. There were three main
and the signal in the neural areas that exhibited significant findings. First, of two primary outcomes (the ARI and the
group-by-time-by-condition interactions and in the ACC CGI), youths with significant levels of irritability who re­
ROI (Table 3). These revealed consistent significant ceived intranasal oxytocin showed significant improvement

Am J Psychiatry 181:4, April 2024 ajp.psychiatryonline.org 295

INTRANASAL OXYTOCIN FOR SEVERE IRRITABILITY IN YOUTHS

FIGURE 2. Neural level changes with oxytocin interventiona and symptom improvements
D. Dorsomedial Prefrontal Cortex ROI in irritability, reactive ag­
A 0.20 gression, and overall clinical
severity of illness.

Estimated Marginal Mean

0.10 Intranasal oxytocin ad­
ministration for 3 weeks
0.00
yielded significant reduc­
tions in overall clinical se­
verity of illness as measured
20 –0.10
Placebo by the CGI-S, and overall
Oxytocin improvement in functioning
F
–0.20 as measured by the CGI-I,
B compared with placebo, al­
E. Posterior Cingulate Cortex ROI
–20 though this finding could be a
0.10
Estimated Marginal Mean preliminary result given the
small sample size. Although
0.00 irritability (on the ARI)
showed only a nonsignificant
improvement, this may re­
–0.10
flect type II error, given the
relatively short period of
–0.20 Placebo administration (3 weeks) and
Oxytocin the relatively small number
of participants. Both the ARI
F. Anterior Cingulate Cortex ROI
C and the RPAQ, used to
0.10
measure irritability and ag­
Estimated Marginal Mean

gressive behavior, were

0.00 designed for trait measure­
ment rather than assessing
–0.10 changes by intervention,
whereas the CGI-S and CGI-
–0.20 I have been widely used for
Placebo changes or improvement in
Oxytocin clinical trials. Consequently,
–0.30
Baseline Follow-up the ARI or RPAQ may have
a
The regions of interest (ROIs) shown are as follows: in panel A, the dorsomedial prefrontal cortex (dmPFC)
not been as effective as the
(coordinates: −13.5, 25.5, 41.5); in panel B, the posterior cingulate cortex (PCC) (coordinates: −1.5, 40.5, 14.5); CGI-S or CGI-I in capturing
and in panel C, the anterior cingulate cortex (ACC) (coordinates: −7.5, 28.5, 11.5). The dmPFC and PCC showed the actual changes from the
significant reduction of blood-oxygen-level-dependent (BOLD) responses to emotional stimuli with intranasal
oxytocin compared with placebo, and the ACC showed a significant reduction of BOLD responses to negative
intranasal oxytocin inter­
emotional stimuli. Panels D–F show the BOLD response parameters. vention (for a similar argu­
ment, see reference 24). It is
in clinical severity, as measured by the CGI-S and the CGI-I. also possible that the aggregation of subthreshold im­
However, the reduction in irritability (on the ARI) in the provement in each disruptive mood and behavioral psy­
oxytocin group compared with the placebo group did not chopathology (especially irritability and aggressive behavior)
reach statistical significance (p=0.115). In addition, there led to overall statistically significant improvement in clinical
were greater improvements in reactive aggression, total level severity (25). Indeed, the aggregated standardized scores
of aggression, and level of daily functioning in the oxytocin (Z scores) of irritability and total aggression as well as irri­
group compared with the placebo group, although the dif­ tability and reactive aggression showed significant differences
ferences were not statistically significant (p values, between the two treatment arms (see the online supplement).
0.055–0.151). Second, youths in the oxytocin group, com­ The effect of intranasal oxytocin on aggressive behavior
pared with those in the placebo group, showed greater re­ has shown mixed results thus far. There have been reports
ductions in BOLD responses in the dmPFC, PCC, and that oxytocin increased aggression in healthy adults (26) but
fusiform gyrus in response to emotional stimuli, and in the decreased aggression in women with higher baseline state
ACC in response to negative emotional stimuli. Third, there anxiety (27). However, most previous work investigating this
were significant correlations among the reductions in BOLD treatment has indexed aggression via computer-based tasks
responses to emotional stimuli in the dmPFC, PCC, and ACC (28) rather than reported symptoms. It will be important for

296 ajp.psychiatryonline.org Am J Psychiatry 181:4, April 2024

 HWANG ET AL.

future work to consider aggression in contexts beyond the pharmacodynamic and pharmacokinetic model of oxytocin
laboratory. for behavioral effects (the latter is, however, under inves­
Studies of intranasal oxytocin in pediatric populations tigation in our ongoing study; ClinicalTrials.gov identifier,
with autism spectrum disorder (ASD) have also reported NCT03863288). Second, the small sample size was a limi­
increased irritability and/or aggressive behavior as adverse tation of the present study. The small number of participants
events (8). Irritability and emotional dysregulation in ASD in each categorical diagnosis meant that we could not
are a prominent clinical issue, but little is known of their conduct any separate analyses based on categorical di­
neurobiological mechanisms and etiology (29). It is possible agnoses. Although it has been suggested that the neuro­
that intranasal oxytocin affects children with ASD signifi­ biological mechanism of irritability is commonly shared
cantly differently than it does neurotypical children, which among the disruptive behavior and disruptive mood dis­
will need to be explored in future work. It is possible, too, that order diagnoses (1), it is possible that the effect of intra­
the impact of oxytocin differs depending on the target nasal oxytocin differs between youths with ADHD and
population and the type of dependent measure used to irritability and those with conduct disorder and irrita­
index aggression. bility, for example. Lastly, sex differences were not sta­
More importantly, intranasal oxytocin significantly mod­ tistically significant between the groups, but there was an
ulated BOLD responses within regions implicated in emo­ imbalance between the groups (p=0.07). To overcome
tional response and regulation. Specifically, the dmPFC and this, we used sex as a covariate in both symptom profile
PCC showed significantly greater reductions in BOLD re­ and BOLD response analyses. Clinical trials with larger
sponses to view trials after the intranasal oxytocin inter­ groups of youths with a greater variety of psychiatric
vention compared with placebo. In our previous study using diagnoses are warranted.
the same task (the affective Stroop task) in youths with ir­ In summary, we report the result of the impact of intranasal
ritability, we observed increased activation of a proximal oxytocin on youths with DBDs and high levels of irritability.
area of the mPFC (coordinates: 4.5, 17.5, 52.5 vs. −13.5, 25.5, We found that youths with DBDs and irritability showed
41.5 in this study) and PCC (coordinates: 4.5, −28.5, 33.5 vs. significant improvement in clinical severity of illness and
−1.5, −40.5, 14.5 in this study), and the degree of activation to subthreshold improvement of irritability symptoms and ag­
view trials in the mPFC area was correlated with level of gressive behavior. Moreover, we observed target engagement
irritability (15). Previous studies have also demonstrated of neural areas implicated in emotion processing (emotional
abnormal responsiveness within these areas related to ir­ responding and emotion regulation) with intranasal oxytocin.
ritability (30, 31) as well as reactive aggression (32). The ACC Notably, there were significant correlations between improve­
ROI, which showed decreased activation to negative stimuli ments in core DBD symptoms and neural-level changes. Al­
in our previous study (15), showed significantly greater re­ though this may be a preliminary result given the small sample
ductions in BOLD responses to negative view trials after size, it may provide guidance for the future direction of clinical
intranasal oxytocin intervention compared with placebo. We trials and treatment for this challenging population.
might have expected an increase in activation with oxytocin
in this area, given that it showed decreased baseline acti­
AUTHOR AND ARTICLE INFORMATION
vation in our previous study. However, it is also possible that
Department of Psychiatry (Hwang, Lerdahl, Edwards), Department of
the previous study’s finding reflected only a population with Psychology (Delizza), and Department of Neurological Sciences (Soltis-
high levels of both irritability and anxiety. In fact, our more Vaughan, Cordts), University of Nebraska Medical Center, Omaha; Digital
recent study showed that irritability was associated with Health Research Division, Korea Institute of Oriental Medicine, Daejeon,
increased activation of the ACC to threat cues (31). Many South Korea (Suk); Slimmer AI, Groningen, the Netherlands (Meffert);
Cognitive Ability and Plasticity Lab, University of Sheffield, Sheffield, U.K.
previous studies have suggested the role of the ACC in
(Garvey); Section on Mood Dysregulation and Neuroscience, NIMH,
emotional stimulus processing (e.g., 33–35), and the de­ Bethesda, Md. (Leibenluft); Child and Adolescent Mental Health Center,
gree of reduction of BOLD response in this area was Mental Health Services, Capital Region of Denmark, Copenhagen (Blair).
correlated with the degree of symptom improvement (see Send correspondence to Dr. Hwang ([email protected]).
Table S4 and Figure S2 in the online supplement), sup­ Supported by National Institute of General Medical Sciences grant
porting the notion that this was indeed a target engage­ 1U54GM115458-01 and NIMH grant U01MH120155-01A1.
ment of the intervention. ClinicalTrials.gov identifier: NCT02824627.
We have a few caveats to offer on interpretation of the The authors report no financial relationships with commercial interests.
results. First, the duration of treatment with oxytocin or
Received March 3, 2023; revisions received June 5 and August 31, 2023;
placebo was relatively short (3 weeks). This may have accepted September 8, 2023; published online February 29, 2024.
contributed to the subthreshold improvements in irritability
and aggressive behavior in the oxytocin group. However, the
REFERENCES
duration of the majority of previous oxytocin clinical trials
1. Leibenluft E: Pediatric irritability: a systems neuroscience ap­
for psychiatric disorders has been between 1 and 6 weeks proach. Trends Cogn Sci 2017; 21:277–289
(28). Currently, there is no clear consensus on the recom­ 2. Avenevoli S, Blader JC, Leibenluft E: Irritability in youth: an update.
mended duration of an oxytocin trial or determination of the J Am Acad Child Adolesc Psychiatry 2015; 54:881–883

Am J Psychiatry 181:4, April 2024 ajp.psychiatryonline.org 297

INTRANASAL OXYTOCIN FOR SEVERE IRRITABILITY IN YOUTHS

3. Collishaw S, Maughan B, Natarajan L, et al: Trends in adolescent 17. Leppanen J, Ng KW, Kim YR, et al: Meta-analytic review of the
emotional problems in England: a comparison of two national effects of a single dose of intranasal oxytocin on threat processing in
cohorts twenty years apart. J Child Psychol Psychiatry 2010; 51: humans. J Affect Disord 2018; 225:167–179
885–894 18. Cai RY, Richdale AL, Uljarevi�c M, et al: Emotion regulation in
4. Vidal-Ribas P, Stringaris A: How and why are irritability and de­ autism spectrum disorder: where we are and where we need to go.
pression linked? Child Adolesc Psychiatr Clin N Am 2021; 30: Autism Res 2018; 11:962–978
401–414 19. Spearing MK, Post RM, Leverich GS, et al: Modification of the
5. Jha MK, Minhajuddin A, Chin Fatt C, et al: Irritability as an in­ Clinical Global Impressions (CGI) Scale for use in bipolar illness
dependent predictor of concurrent and future suicidal ideation in (BP): the CGI-BP. Psychiatry Res 1997; 73:159–171
adults with stimulant use disorder: findings from the STRIDE study. 20. Shaffer D, Gould MS, Brasic J, et al: A Children’s Global Assessment
J Affect Disord 2021; 292:108–113 Scale (CGAS). Arch Gen Psychiatry 1983; 40:1228–1231
6. Leibenluft E, Kircanski K: Chronic irritability in youth: a reprise on 21. Cox RW, Chen G, Glen DR, et al: fMRI clustering in AFNI: false-
challenges and opportunities toward meeting unmet clinical needs. positive rates redux. Brain Connect 2017; 7:152–171
Child Adolesc Psychiatr Clin N Am 2021; 30:667–683 22. Poldrack RA: Region of interest analysis for fMRI. Soc Cogn Affect
7. Jawad MY, Ahmad B, Hashmi AM: Role of oxytocin in the path­ Neurosci 2007; 2:67–70
ogenesis and modulation of borderline personality disorder: a re­ 23. Hwang S, Chung U, Chang Y, et al: Neural responses to fluoxetine in
view. Cureus 2021; 13:e13190 youths with disruptive behavior and trauma exposure: a pilot study.
8. DeMayo MM, Song YJC, Hickie IB, et al: A review of the safety, J Child Adolesc Psychopharmacol 2021; 31:562–571
efficacy and mechanisms of delivery of nasal oxytocin in children: 24. Towbin K, Vidal-Ribas P, Brotman MA, et al: A double-blind
therapeutic potential for autism and Prader-Willi syndrome, and randomized placebo-controlled trial of citalopram adjunctive to
recommendations for future research. Paediatr Drugs 2017; 19: stimulant medication in youth with chronic severe irritability.
391–410 J Am Acad Child Adolesc Psychiatry 2020; 59:350–361
9. Macdonald K, Macdonald TM: The peptide that binds: a systematic 25. Fervaha G, Agid O, Takeuchi H, et al: Relationship between
review of oxytocin and its prosocial effects in humans. Harv Rev symptomatic improvement and overall illness severity in patients
Psychiatry 2010; 18:1–21 with schizophrenia. J Clin Psychopharmacol 2015; 35:128–133
10. Bakker-Huvenaars MJ, Greven CU, Herpers P, et al: Saliva oxytocin, 26. Ne’eman R, Perach-Barzilay N, Fischer-Shofty M, et al: Intranasal
cortisol, and testosterone levels in adolescent boys with autism administration of oxytocin increases human aggressive behavior.
spectrum disorder, oppositional defiant disorder/conduct disorder Horm Behav 2016; 80:125–131
and typically developing individuals. Eur Neuropsychopharmacol 27. Campbell A, Hausmann M: Effects of oxytocin on women’s ag­
2020; 30:87–101 gression depend on state anxiety. Aggress Behav 2013; 39:316–322
11. Ișık Ü, Bilgiç A, Toker A, et al: Serum levels of cortisol, dehy­ 28. Kendrick KM, Guastella AJ, Becker B: Overview of human oxytocin
droepiandrosterone, and oxytocin in children with attention- research. Curr Top Behav Neurosci 2018; 35:321–348
deficit/hyperactivity disorder combined presentation with and 29. Keluskar J, Reicher D, Gorecki A, et al: Understanding, assessing,
without comorbid conduct disorder. Psychiatry Res 2018; 261: and intervening with emotion dysregulation in autism spectrum
212–219 disorder: a developmental perspective. Child Adolesc Psychiatr
12. Demirci E, Ozmen S, Kilic E, et al: The relationship between ag­ Clin N Am 2021; 30:335–348
gression, empathy skills and serum oxytocin levels in male children 30. Tseng WL, Deveney CM, Stoddard J, et al: Brain mechanisms of
and adolescents with attention deficit and hyperactivity disorder. attention orienting following frustration: associations with irrita­
Behav Pharmacol 2016; 27:681–688 bility and age in youths. Am J Psychiatry 2019; 176:67–76
13. Insel TR: Translating oxytocin neuroscience to the clinic: a National 31. Zhang R, Bashford-Largo J, Lukoff J, et al: Callous-unemotional
Institute of Mental Health perspective. Biol Psychiatry 2016; 79: traits moderate the relationship between irritability and threat­
153–154 ening responding. Front Psychiatry 2021; 12:617052
14. Pacheco J, Garvey MA, Sarampote CS, et al: Annual Research 32. Lotze M, Veit R, Anders S, et al: Evidence for a different role of
Review: the contributions of the RDoC research framework on the ventral and dorsal medial prefrontal cortex for social re­
understanding the neurodevelopmental origins, progression and active aggression: an interactive fMRI study. Neuroimage 2007;
treatment of mental illnesses. J Child Psychol Psychiatry 2022; 63: 34:470–478
360–376 33. Bush G, Luu P, Posner MI: Cognitive and emotional influences in
15. Crum KI, Hwang S, Blair KS, et al: Interaction of irritability and anterior cingulate cortex. Trends Cogn Sci 2000; 4:215–222
anxiety on emotional responding and emotion regulation: a func­ 34. Chiu PH, Holmes AJ, Pizzagalli DA: Dissociable recruitment of
tional MRI study. Psychol Med 2020; 51:2778–2788 rostral anterior cingulate and inferior frontal cortex in emotional
16. Grace SA, Rossell SL, Heinrichs M, et al: Oxytocin and brain activity response inhibition. Neuroimage 2008; 42:988–997
in humans: a systematic review and coordinate-based meta-analysis 35. Feroz FS, Leicht G, Steinmann S, et al: The time course of activity
of functional MRI studies. Psychoneuroendocrinology 2018; 96: within the dorsal and rostral-ventral anterior cingulate cortex in the
6–24 emotional Stroop task. Brain Topogr 2017; 30:30–45

298 ajp.psychiatryonline.org Am J Psychiatry 181:4, April 2024