Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Quantitative facial expression analysis revealed the efficacy and time course of oxytocin in autism

Quantitative facial expression analysis revealed the efficacy and time course of oxytocin in autism Abstract Discrepancies in efficacy between single-dose and repeated administration of oxytocin for autism spectrum disorder have led researchers to hypothesize that time-course changes in efficacy are induced by repeated administrations of the peptide hormone. However, repeatable, objective, and quantitative measurement of autism spectrum disorder’s core symptoms are lacking, making it difficult to examine potential time-course changes in efficacy. We tested this hypothesis using repeatable, objective, and quantitative measurement of the core symptoms of autism spectrum disorder. We examined videos recorded during semi-structured social interaction administered as the primary outcome in single-site exploratory (n = 18, crossover within-subjects design) and multisite confirmatory (n = 106, parallel-group design), double-blind, placebo-controlled 6-week trials of repeated intranasal administrations of oxytocin (48 IU/day) in adult males with autism spectrum disorder. The main outcomes were statistical representative values of objectively quantified facial expression intensity in a repeatable part of the Autism Diagnostic Observation Schedule: the maximum probability (i.e. mode) and the natural logarithm of mode on the probability density function of neutral facial expression and the natural logarithm of mode on the probability density function of happy expression. Our recent study revealed that increases in these indices characterize autistic facial expression, compared with neurotypical individuals. The current results revealed that oxytocin consistently and significantly decreased the increased natural logarithm of mode on the probability density function of neutral facial expression compared with placebo in exploratory (effect-size, −0.57; 95% CI, −1.27 to 0.13; P = 0.023) and confirmatory trials (−0.41; −0.62 to −0.20; P < 0.001). A significant interaction between time-course (at baseline, 2, 4, 6, and 8 weeks) and the efficacy of oxytocin on the natural logarithm of mode on the probability density function of neutral facial expression was found in confirmatory trial (P < 0.001). Post hoc analyses revealed maximum efficacy at 2 weeks (P < 0.001, Cohen’s d = −0.78; 95% CI, −1.21 to −0.35) and deterioration of efficacy at 4 weeks (P = 0.042, Cohen’s d = −0.46; 95% CI, −0.90 to −0.01) and 6 weeks (P = 0.10, Cohen’s d = −0.35; 95% CI, −0.77 to 0.08), while efficacy was preserved at 2 weeks post-treatment (i.e. 8 weeks) (P < 0.001, Cohen’s d = −1.24; 95% CI, −1.71 to −0.78). Quantitative facial expression analyses successfully verified the positive effects of repeated oxytocin on autistic individuals’ facial expressions and demonstrated a time-course change in efficacy. The current findings support further development of an optimized regimen of oxytocin treatment. Asperger syndrome, pervasive developmental disorder, neuroendocrinology, neuropeptide, surrogate marker Introduction Autism spectrum disorder (ASD) is a highly prevalent developmental disorder presenting core symptoms including deficits in social interaction and communication and restricted and repetitive behaviours (Baird et al., 2006; Xu et al., 2018). No approved medication is currently available to reduce the core symptoms of ASD, which are considered static traits rather than treatable symptoms (Lai et al., 2014). This situation may be related to a lack of research attention to the development of repeatable, objective, and quantitative longitudinal assessment tools for the core symptoms of ASD (Yamasue, 2016). Although oxytocin has recently been expected as a novel therapeutic for ASD core symptoms, both the previous findings and the outcomes assessing the efficacy have been controversial (Andari et al., 2010; Harris and Carter, 2013; Guastella et al., 2016; Hurlemann and Scheele, 2016; Leng and Ludwig, 2016; Walum et al., 2016; Yamasue and Domes, 2018). Previous studies have consistently demonstrated that single-dose administration of oxytocin significantly improves experimental measures associated with ASD core symptoms (Guastella et al., 2010; Gordon et al., 2013; Aoki et al., 2014, 2015; Domes et al., 2014; Watanabe et al., 2014). However, previous studies of repeated oxytocin treatment on ASD core symptoms have yielded inconsistent results (Guastella et al. 2015; Kosaka et al., 2016; Yatawara et al., 2016). Our exploratory trial revealed a significant improvement in Autism Diagnostic Observation Schedule (ADOS) reciprocity induced by 6 weeks of repeated administration of oxytocin (Watanabe et al., 2015), although the confirmatory trial did not reveal superiority in the efficacy of oxytocin compared with placebo using the same primary end-point in the similar demographic adult male individuals with high-functioning ASD (Yamasue et al., 2018). The discrepancy in efficacy could, at least partially, be derived from a significant placebo effect on ADOS reciprocity in the confirmatory trial (Yamasue et al., 2018). In contrast to the discrepant results regarding the primary end-point, repeated administration of oxytocin has shown significant effects on objectively quantified behavioural and neural measures during a social behaviour task and eye-gaze measurement on socially-relevant regions in both exploratory and confirmatory trials (Watanabe et al., 2015; Yamasue et al., 2018). Based on these findings, taken together with a previous exploratory trial from another research group (Anagnostou et al., 2012), it could reasonably be predicted that objective assessments of social behaviour are likely to reveal the efficacy of repeated administrations of oxytocin (Anagnostou et al., 2012; Watanabe et al., 2015; Yamasue et al., 2018). The discrepancy in efficacy between single and repeated administrations of oxytocin also indicates that repeated administration might induce a decline in efficacy (Watanabe et al., 2015; Yamasue, 2016). Our recent study supports this notion, consistently demonstrating differential neurochemical effects of single and repeated administrations of oxytocin on medial frontal cortices in human neuroimaging and examination of gene expressions in mice (Benner et al., 2018). Furthermore, a similar discrepancy has also been observed in animal studies: while acute oxytocin administration enhanced animal prosocial behaviour, repeated administration diminished these effects rather than increasing them, suggesting potential saturation, downregulation, or negative feedback of oxytocin receptors (Bales et al., 2013, 2014). However, previous human studies have not fully addressed this issue. The present study tested the effects of oxytocin on repeatable, objective, and quantitative measurement of facial expression, using a method recently developed in our laboratory (Owada et al., 2018), during social interactions in individuals with ASD. The first aim of the current study was to test and verify the effects of repeated administration of intranasal oxytocin on autistic social behaviour using objective quantitative analysis of facial expression. The second aim was to detect a potential time-course change in efficacy of oxytocin on autistic social deficits using the repeatability of quantitative face analysis. To explore these issues, we examined videos recorded during ADOS administered in two independent clinical trials (Watanabe et al., 2015; Yamasue et al., 2018). We first conducted an exploratory test of the study hypothesis in a small sample dataset of a single-site exploratory trial (Watanabe et al., 2015). We then conducted a confirmatory test to examine this effect, and explored potential time-course changes in efficacy in a larger sample dataset in our confirmatory multisite trial with longitudinal assessments (Yamasue et al., 2018). Materials and methods The ethics committee of University of Tokyo Hospital approved the current quantitative study of facial expression (#10245). After a complete explanation of the study, the mental capacity to consent was confirmed for each participant by a psychiatrist (H.Y. or K.O.), and written informed consent was obtained from all participants. The study protocols of the exploratory and confirmatory trials were approved by the institutional review boards at each site and registered in the University Hospital Medical Information Network Clinical Trials Registry (UMIN000007122 and UMIN000015264). Although the results on the primary and secondary outcomes, except for the quantitative measurement of facial expression in these trials, have been published previously (Watanabe et al., 2015; Yamasue et al., 2018), the quantitative facial expression measurement was an unpublished key secondary outcome of these trials and was newly examined in the current study. Study design and participants Exploratory trial The exploratory trial was a randomized, double-blind, placebo-controlled, crossover trial, which was primarily conducted in an outpatient clinic of The University of Tokyo Hospital (Watanabe et al., 2015). The inclusion criteria were: (i) age 18–54 years; (ii) male; (iii) diagnosis of autistic disorder, Asperger’s disorder, or pervasive developmental disorder-not otherwise specified (PDD-NOS) based on the Diagnostic and Statistical Manual-Revision IV-Text Revision (DSM-IV-TR) (American Psychiatric Association, 2000); and (iv) verbal intelligence quotient (IQ) above 85 and full IQ above 80 as measured using the Wechsler Adult Intelligent Scale (WAIS)-R (Wechsler, 1981). Exclusion criteria were: (i) instability of symptoms of comorbid mental disorders such as mood disorder or anxiety disorder; (ii) history of hypersensitivity to oxytocin; (iii) history of seizures or traumatic brain injury with loss of consciousness for longer than 5 min; and (iv) history of alcohol-related disorders, substance abuse, or addiction. Participants with contraindications to undergoing MRI scanning were also excluded in the exploratory trial, which involved functional MRI indices as a secondary outcome (Watanabe et al., 2015). An experienced psychiatrist (H.Y.) performed the diagnoses based on DSM-IV-TR (American Psychiatric Association, 2000) with more than 2 months of follow-up examinations. Another certified psychiatrist/psychologist confirmed the diagnoses using the Japanese version of the Autism Diagnostic Interview-Revised (ADIR) (Lord et al., 1994) (H.K.) and ADOS (Lord et al., 1989) (M.K.). Between March and December 2012, 20 high-functioning adult male individuals with ASD were enrolled in this trial, and final assessment at the 12-week end-point was performed between August 2012 and April 2013. Confirmatory trial This multicentre, parallel-group, placebo-controlled, double-blind, confirmatory trial of intranasal oxytocin in participants with ASD was conducted at the University of Tokyo Hospital, Nagoya University Hospital, Kanazawa University Hospital, and University of Fukui Hospital in Japan (Yamasue et al., 2018). The inclusion and exclusion criteria were similar to those in the exploratory trial. The inclusion criteria were: (i) age: 18–54 years; (ii) male; (iii) diagnosis of autistic disorder, Asperger’s disorder, or PDD-NOS based on DSM-IV-TR (American Psychiatric Association, 2000); (iv) score exceeding the cut-off value of 10 for qualitative abnormalities in social reciprocity on ADIR (Lord et al., 1994); and (v) verbal IQ above 85 and full IQ above 80 as measured using WAIS-III (Wechsler, 1981). Exclusion criteria were: (i) primary psychiatric diagnoses other than those listed in the inclusion criteria; (ii) instability of symptoms of comorbid mental disorders such as mood disorder or anxiety disorder; (iii) history of changes in medication or doses of psychotropics within 1 month before randomization; (iv) current treatment with more than two psychotropics; (v) current medication for comorbid attention-deficit/hyperactivity disorder (i.e. atomoxetine or methylphenidate, which were the only approved medications in Japan at the time of the study); (vi) history of continual oxytocin treatment; (vii) history of hypersensitivity to oxytocin; (viii) history of seizures or traumatic brain injury with loss of consciousness for longer than 5 min; and (ix) history of alcohol-related disorders, substance abuse, or addiction. The recruitment, which was open to the public, and the eligibility testing processes at the recruiting office are described in detail elsewhere (Yamasue et al., 2018) and in the Supplementary material. A psychiatrist (H.Y., T.O., Y.U., T.M., or H.Ko.) experienced in developmental disorders performed the diagnoses of ASD based on DSM-IV-TR. Another certified psychiatrist/psychologist confirmed the diagnoses using the Japanese version of the ADIR (Lord et al., 1994) (H.Ku., M.Ku., or T.F.) and ADOS (Lord et al., 1989) (M.Ku., Y.Y., or K.M.) (Table 1). The trial enrolled 106 males with high-functioning ASD between January 2015 and March 2016. End-point assessments were performed between March 2015 and April 2016. Table 1 Demographic data of the exploratory and confirmatory trial Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 aSocioeconomic status (SES) assessed using the Hollingshead Index; higher scores indicate lower status. bThe intelligence quotients were measured using the Wechsler Adult Intelligence Scale. IQ = intelligence quotient; OT = oxytocin; PL = placebo; SD = standard deviation. View Large Table 1 Demographic data of the exploratory and confirmatory trial Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 aSocioeconomic status (SES) assessed using the Hollingshead Index; higher scores indicate lower status. bThe intelligence quotients were measured using the Wechsler Adult Intelligence Scale. IQ = intelligence quotient; OT = oxytocin; PL = placebo; SD = standard deviation. View Large Intervention In the exploratory trial, participants received oxytocin (24 IU, Syntocinon-Spray; Novartis) in the morning and afternoon over six consecutive weeks (i.e. 48 IU/day) and placebo in the same way with cross-over administration (Fig. 1A). In the confirmatory trial, participants received oxytocin (48 IU/day) or placebo in the morning and afternoon over six consecutive weeks (Fig. 1B). The placebo contained all the inactive ingredients included in the oxytocin spray. On the last day of each 6-week administration, participants underwent end-point clinical assessments, including ADOS, 15 min after the last drug administration. All participants underwent sufficient training for intranasal administration with identical instructions (Supplementary material), and the effectiveness of training was confirmed at each 2-week assessment. Treatment adherence was assessed with a self-report daily record (Supplementary material). Figure 1 View largeDownload slide Study designs. (A) Exploratory trial in crossover design. (B) Confirmatory trial in parallel group design. Figure 1 View largeDownload slide Study designs. (A) Exploratory trial in crossover design. (B) Confirmatory trial in parallel group design. Randomization and masking Randomization and masking of drug administration was randomly assigned to participants in the two groups in a one-to-one ratio based on a computer-generated randomized order. In the exploratory crossover trial, this included a group in which oxytocin was administered first and a group in which placebo was given first, and, in the confirmatory trial, this included either the oxytocin or placebo group. In the confirmatory parallel-group trial, randomization was stratified according to study site and median ADIR score (<18, ≥18, based on the exploratory trial). Oxytocin and placebo were stored in visually-identical spray bottles (Victoria Pharmacy). Bottle labels were covered to keep drug types unknown to all the participants, their families, clinicians, and assessors including ADOS administrators and assessors. The registration, allocation, and data management procedures were defined separately (Supplementary material). Outcomes The outcome of the current study, facial expression intensity values, was common to the exploratory and confirmatory trials, and was described as a secondary outcome of these trials. Expression intensities during a semi-structured socially interactive situation extracted from ADOS module 4 (Lord et al., 1989) were quantified by a dedicated software program (FaceReader version 6·1; Noldus Information Technology Inc., Wageningen, The Netherlands), and were assessed at baseline, 6- and 12-week end-points in the exploratory trial, or every 2 weeks in the confirmatory trial (i.e. baseline; treatment Weeks 2, 4, and 6; and 2 weeks post-treatment). At the time of planning our study, we compared several available software programs dedicated to facial expression analysis. Among them, we found that FaceReader had the most credible references regarding the methodology and algorithms involved (Cohen et al., 2013; Lewinski et al., 2014; Fujiwara et al., 2015), in addition to its ability to output frame-by-frame emotional intensity values. Therefore, we used FaceReader in our study. To ensure a uniform and consistent setting for observing facial expressions, an activity (i.e. ‘Cartoons’) of ADOS was selected based on the degree of structure needed to orient a participant to a certain consistent response, repeatability to elicit improvised responses, and the feasibility of whether faces remained positioned such that the software could correctly recognize expressions and whether the activity length was short enough to observe situation-evoked facial expression with minimal contamination. The degrees of structure, repeatability, and feasibility were qualitatively and quantitatively assessed, and selection was performed based on the results from statistical comparisons between these parameters of ADOS activities. The process by which activity was selected from ADOS is described in detail elsewhere (Owada et al., 2018) and in the Supplementary material. The ADOS was administered by clinicians who were certificated or who completed a training course regarding the research use of ADOS and whose credentials had been validated by another certified administrator. The methods by which expression intensities were recorded and calculated are described in detail elsewhere (Owada et al., 2018) and in the Supplementary material. In brief, time-trend datasets from the selected activity of each participant yielded a pair of expression intensity variables. These represented the distribution of expression intensity values for each facial expression element: the value at which the probability was maximum (i.e. mode) and the natural logarithm of the maximum probability value on the probability density function (i.e. log-PDFmode), which was estimated by kernel density using Gaussian kernels. The mode gives a measure of the most likely intensity for the given time period, while the log-PDFmode gives a measure of the variability of intensity for the given time period. We used the increased mode and log-PDFmode of neutral facial expression and the log-PDFmode of happy facial expression as outcomes, as our previous study demonstrated that increases in these expression intensities can characterize facial expressions in individuals with ASD well, in comparison with neurotypical individuals (Owada et al., 2018). Statistical analysis The target number of participants was set as 20 in the exploratory trial based on previous studies (Andari et al., 2010; Guastella et al., 2010), while a target of 114 was calculated for the confirmatory trial based on power analyses using the results from the exploratory trial [see details in our previous reports (Watanabe et al., 2015; Yamasue et al., 2018) and Supplementary material]. Statistical analyses were performed using IBM SPSS 22.0 (IBM Corp., Armonk, NY), MATLAB R2015a (The MathWorks, Natick, MA), R 3.3.2 (http://www.R-project.org/), and Python libraries for scientific computation (NumPy, and SciPy). The outcomes were analysed using generalized estimating equations with an unstructured correlation and robust standard error estimates considering the time points as a factor (i.e. 0, 6, and 12 weeks in the exploratory trial; 0, 2, 4, 6, and 8 weeks in the confirmatory trial). Site was also considered as a factor in the multisite confirmatory trial. Main effects and interactions for the pharmacologically different conditions (oxytocin/placebo) and the factors were estimated with a significance level of P < 0.05. The least square mean between group differences and the associated 95% confidence intervals (CI) at the end-point were also estimated. The estimated marginal means (EM means) for levels of factors and factor interactions were also calculated. Data availability Individual de-identified participant data underlying the results reported in this study, as well as the study protocol, statistical analysis plan, and informed consent form are available on request from investigators providing a methodologically sound proposal and whose proposed use of the data has been approved by an independent review committee identified for this purpose during the period beginning 9 months and ending 5 years following article publication. Proposals should be directed to the corresponding author. Results Data and results from the exploratory trial The detailed participant flow in the exploratory trial has been reported elsewhere (Fig. 1A aaand Table 1) (Watanabe et al., 2015). Expression intensities were analysed for 18 individuals at three time points except for three (two in the oxytocin-first group at the end-point of placebo period and one in placebo-first group at the end-point of oxytocin period) for which data failed to be recorded. No major adverse effects were observed in either group (see details in Watanabe et al., 2015). The efficacy of oxytocin was analysed using generalized estimation equations with an unstructured correlation and robust standard error estimates considering the time points as a factor (i.e. 0, 6, and 12 weeks). We found that 6-week oxytocin administration significantly reduced the log-PDFmode of neutral face in the participants with ASD (P = 0.023, Cohen’s Cohen’s d = −0.57; 95% CI, −1.27 to 0.13; Fig. 2 and Table 2) with no significant main effect of time point and no significant interaction between drug type (oxytocin/placebo) and time point (P > 0.87). As the log-PDFmode of neutral facial expression was significantly increased in the participants with ASD compared with neurotypical controls before administration of oxytocin (Owada et al., 2018), the effect of oxytocin was observed in the direction of recovery. No significant improvement of the mode of neutral facial expression or the log-PDFmode of happy facial expression was found (P > 0.06). Table 2 Effect of oxytocin on facial expression intensities Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) aBy generalized estimating equations. bEstimated by EM means. Values are presented as mean (SD). EI = expression intensity; log-PDFmode = natural logarithm of maximum probability on the probability density function; OT = oxytocin; PL = placebo. View Large Table 2 Effect of oxytocin on facial expression intensities Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) aBy generalized estimating equations. bEstimated by EM means. Values are presented as mean (SD). EI = expression intensity; log-PDFmode = natural logarithm of maximum probability on the probability density function; OT = oxytocin; PL = placebo. View Large Figure 2 View largeDownload slide Effects of intranasal oxytocin on facial expressions in the exploratory trial. Effects of oxytocin or placebo administration on facial expression evaluated by the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (EIs). Estimated marginal means (EM means) at baseline, 6 and 12 weeks of placebo and oxytocin administrations are shown for placebo-first and oxytocin-first groups individually. The bars represent 95% CIs of EM means. The effect sizes between placebo and oxytocin administrations at 6 weeks (Cohen’s d = −0.36; 95% CI, −1.29 to 0.58) and 12 weeks (Cohen’s d = −0.76; 95% CI, −1.80 to 0.31) were also presented. Overall, oxytocin administration showed a significant reduction in log-PDFmode of neutral expression intensities compared with placebo (P = 0.023, Cohen’s d = −0.57; 95% CI, −1.27 to 0.13). Figure 2 View largeDownload slide Effects of intranasal oxytocin on facial expressions in the exploratory trial. Effects of oxytocin or placebo administration on facial expression evaluated by the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (EIs). Estimated marginal means (EM means) at baseline, 6 and 12 weeks of placebo and oxytocin administrations are shown for placebo-first and oxytocin-first groups individually. The bars represent 95% CIs of EM means. The effect sizes between placebo and oxytocin administrations at 6 weeks (Cohen’s d = −0.36; 95% CI, −1.29 to 0.58) and 12 weeks (Cohen’s d = −0.76; 95% CI, −1.80 to 0.31) were also presented. Overall, oxytocin administration showed a significant reduction in log-PDFmode of neutral expression intensities compared with placebo (P = 0.023, Cohen’s d = −0.57; 95% CI, −1.27 to 0.13). Although individuals with unstable comorbid mental disorders including moderate or severe anxiety were excluded, we additionally tested the possibility that anxiety or depressive tendencies of participants could confound the effects of oxytocin using analyses controlling for scores on the State-Trait Anxiety Inventory (STAI) (Spielberger et al., 1970) and The Center for Epidemiologic Studies Depression Scale (CESD) (Radloff, 1977). The results confirmed that the individual differences in baseline anxious or depressive tendency were not correlated with oxytocin’s efficacy regarding the log-PDFmode for neutral facial expression (P > 0.28). When baseline anxious or depressive tendencies were controlled as covariates, the efficacy of oxytocin on the variability of neutral facial expression was totally preserved (P < 0.025). Data and results from the confirmatory trial The detailed participant flow in the confirmatory trial has been reported elsewhere (Fig. 1B) (Yamasue et al., 2018). The current study analysed expression intensities for 103 individuals at five time points (515 data points) (Table 1), with the exception of 60 data points (25 in the oxytocin group and 35 in the placebo group), which were excluded because of recording failure, defocused video images, or poor facial recognition rate. Ninety-four of the participants were free from psychotropics other than oxytocin throughout the trial period, while 12 continued their psychotropic medications during the period [four antipsychotics, four antidepressants, two hypnotics, and two anticonvulsants (mood stabilizers)]. Eighty-three participants were diagnosed with high-functioning autistic disorder, 12 with Asperger’s disorder, and 11 with PDD-NOS. No significant differences between the oxytocin and placebo groups were observed in background information or prevalence of adverse events [see details elsewhere (Yamasue et al., 2018) and in Table 1]. Consistent with the exploratory trial, the effect of oxytocin was analysed using generalized estimating equations with an unstructured correlation and robust standard error estimates considering the time points as a factor (i.e. 0, 2, 4, 6, and 8 weeks). Site was also considered as a factor in the multisite confirmatory trial. Regarding the expression intensities, consistent with the exploratory trial, the log-PDFmode of neutral facial expression was significantly reduced in the oxytocin-administered group compared with the placebo-administered group across the assessment period (P < 0.001, Cohen’s d = −0.41; 95% CI, −0.62 to −0.20). A significant interaction between time point and treatment group was also found on the log-PDFmode of neutral facial expression (P < 0.001). As shown in Fig. 3, EM means and effect sizes at each time point revealed a time course change in the efficacy of oxytocin: a maximum effect was found at 2-weeks post-treatment (P < 0.001, Cohen’s d = −1.24; 95% CI, −1.71 to −0.78) compared with 6-week administration period such as 2 weeks (P = 0.014, Cohen’s d = −0.53; 95% CI, −0.94 to −0.10), 4 weeks (P = 0.068, Cohen’s d = −0.41; 95% CI, −0.85 to 0.03) and 6 weeks (P = 0.054, Cohen’s d = −0.41; 95% CI, −0.83 to 0.02). However, even after exclusion of the 2-week post-treatment data, post hoc analysis showed a significant interaction between time point and treatment (P = 0.02), indicating a significant time-course change within the 6-week administration period with a maximum effect at 2 weeks (P < 0.001, Cohen’s d = −0.78; 95% CI, −1.21 to −0.35) compared with 4 weeks (P = 0.042, Cohen’s d = − 0.46; 95% CI, −0.90 to −0.01) and 6 weeks (P = 0.10, Cohen’s d = −0.35; 95% CI, −0.77 to 0.08). No significant effect of oxytocin was found on the mode of neutral facial expression or on the log-PDFmode of happy facial expression (P > 0.07). Figure 3 View largeDownload slide Effect of intranasal oxytocin on autistic facial expression and its time-course change in the confirmatory trial. Estimated marginal means (EM means) of the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (expression intensities) at each evaluation point are shown for the placebo- and oxytocin-administered groups individually. The bars represent 95% CIs of EM means. Evaluations were performed every 2 weeks during the 6-week oxytocin or placebo administration and 2 weeks after ceasing administration. The effect sizes between placebo and oxytocin administrations at each assessment points were also presented. Overall, the difference in the changes of the log-PDFmode of neutral expression intensities between oxytocin- and placebo-administered groups was significant (P < 0.001, Cohen’s d = −0.41; 95% CI, −0.62 to −0.20). d = Cohen’s d. Figure 3 View largeDownload slide Effect of intranasal oxytocin on autistic facial expression and its time-course change in the confirmatory trial. Estimated marginal means (EM means) of the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (expression intensities) at each evaluation point are shown for the placebo- and oxytocin-administered groups individually. The bars represent 95% CIs of EM means. Evaluations were performed every 2 weeks during the 6-week oxytocin or placebo administration and 2 weeks after ceasing administration. The effect sizes between placebo and oxytocin administrations at each assessment points were also presented. Overall, the difference in the changes of the log-PDFmode of neutral expression intensities between oxytocin- and placebo-administered groups was significant (P < 0.001, Cohen’s d = −0.41; 95% CI, −0.62 to −0.20). d = Cohen’s d. Individual differences in baseline anxious or depressive tendencies did not relate to oxytocin’s efficacy on the log-PDFmode for neutral facial expression (P > 0.24). The efficacy of oxytocin on the variability of neutral facial expression was preserved even after controlling baseline anxious or depressive tendencies (P < 0.001). Discussion The crucial finding from the current study is that repeated administration of oxytocin significantly improved the characteristic reduced variability of neutral facial expression in individuals with ASD. As the log-PDFmode for neutral faces is considered to reflect variation in facial expression (Owada et al., 2018), oxytocin consistently increased variation in the neutral facial expressions of ASD participants in both the exploratory and confirmatory trials. As reduced variability of facial expressions in individuals with ASD is a well-known clinical feature, the present results suggest that oxytocin improves at least one component of the autism-related characteristics of facial expression during social interaction, indicating partial improvement of deficits in social interaction and communication. Furthermore, the confirmatory trial detected a significant time-course change in the efficacy of oxytocin. Although efficacy decreased during the 6 weeks of repeated administration, it was still preserved at 2 weeks post-treatment. Although we reported inconsistent results about the efficacy of oxytocin on scores in ADOS recently, a gold standard diagnostic tool based on assessments of ASD core symptoms (Watanabe et al., 2015; Yamasue et al., 2018), the current findings demonstrated that 6 weeks repeated administration of oxytocin significantly enhanced objectively quantified facial expressions in individuals with ASD consistently in the exploratory and confirmatory trials. When the full ADOS reciprocity assessment was administered, the primary outcome of our confirmatory trial, the significant superiority of oxytocin to placebo, was not apparent, as reported in our recent study (Yamasue et al., 2018). It should be noted that this might indicate that facial expression analysis using only a short component of the full ADOS was superior to using the ADOS as a whole, possibly due to objectivity and quantitative analysis of facial expression. In the confirmatory trial, the enhanced ability to detect efficacy might be attributable to the objectivity rather than sensitivity of facial expression analysis, because the effect size of oxytocin on facial expression (Cohen’s d = 0.27) is not larger than that for ADOS [Cohen’s d = 0.47, reported in our recent report (Yamasue et al., 2018)]. However, the effect size of placebo on facial expression (Cohen’s d = 0.05) is below that of ADOS [Cohen’s d = 0.56 (Yamasue et al., 2018)]. The objectivity of facial expression analysis suggested that oxytocin may have significant efficacy for improving one component of the facial expression deficit in ASD, which is related to social deficits. This notion is consistent with our recent paper reporting the superiority of measurement of eye gaze fixation on socially relevant regions over ADOS reciprocity in the confirmatory trial (Yamasue et al., 2018). Facial expression analysis is based on only a few minutes of the ADOS (Owada et al., 2018), whereas the entire ADOS requires 40–60 min to administer (Lord et al., 1989). Therefore, facial expression analysis is easily repeatable in longitudinal assessments. The features of repeatability and objective quantification successfully demonstrate for the first time, to our knowledge, a time-course change in the efficacy of oxytocin. Together with previous human (Watanabe et al., 2015; Yamasue, 2016; Benner et al., 2018; Yamasue and Domes, 2018) and animal studies (Bales et al., 2013, 2014; Benner et al., 2018) suggesting a discrepancy between single and repeated administrations of oxytocin, the current findings suggest a time-course change of efficacy of oxytocin on ASD social symptoms. The currently observed deterioration of oxytocin’s efficacy might support a potential underlying mechanism, such as a negative feedback system of peptide hormones, upregulation or downregulation of the closely related peptide arginine vasopressin, or downregulation of oxytocin receptor (Insel et al., 1992; Yamamoto et al., 2004). Further research is needed to address the issue. The current results further support the preservation of oxytocin’s effect even 2 weeks after the cessation of treatment. To our knowledge, no previous study has reported such an effect (Keech et al., 2018). Although speculative, potential disengagement from the mechanism of deterioration of oxytocin’s efficacy might preserve or even increase the effects of neuropeptide after discontinuation. However, as the exploratory trial did not support the post-treatment persistence of the effects of oxytocin, future studies should investigate this issue further. The current study involved several potential limitations and methodological considerations that should be considered. First, the participants in our trials were all adult, male, Japanese individuals with ASD. Therefore, although the uniformity in participants’ demographic characteristics enhanced the ability to detect scientifically sound evidence, it should be noted that the current results may not be generalizable to other clinical or non-clinical populations. Second, since the outcome measure of current study was developed in our research team, the findings should be replicated by other research groups. The majority of the quantification methods were automatically conducted and independent of human labour (Owada et al., 2018), facilitating the replication of the current findings. Third, as some autistic characteristics in facial expression at baseline, such as a high mode of neutral facial expression and log-PDFmode of happy facial expression were not significantly improved by oxytocin treatment, we were unable to conclude that oxytocin can treat or recover all characteristics of facial expression in individuals with ASD. Furthermore, we did not show an association between the effects of oxytocin on the quantified facial expression and those on beneficial therapeutic outcomes, such as Clinical Global Impressions (CGI) (Guy, 1976) or Global Assessment of Functioning (GAF) (Aas, 2011). Although the effects of oxytocin on the variability of neutral facial expression exhibited weak correlations with the neural effect of oxytocin on anterior cingulate activity during a social judgment task (ρ = −0.56, P = 0.028) and on resting state functional connectivity between anterior cingulate and dorsomedial prefrontal cortices (ρ = −0.60, P = 0.019) assessed with functional MRI in the exploratory trial (detailed in Watanabe et al., 2015), the current results did not necessarily support the clinically beneficial effects of oxytocin. Fourth, the possibility of further improving the method of quantifying facial expressions, the task used to induce facial expressions in individuals with autism, and the outcome variables, should be considered. FaceReader and ADOS are not optimized for longitudinal facial expression analyses in individuals with ASD. It could be useful to validate these tools (and other available software) in this methodological context using electromyography, and/or action unit processing. The outcome variables used were determined based on the results of our previous case-control comparison in a limited sample size (Owada et al., 2018). The current quantitative facial expression analyses in our exploratory and confirmatory clinical trials successfully detected and verified the therapeutic effect of repeated administrations of intranasal oxytocin on autistic features in facial expressions during social interaction. Furthermore, for the first time, the present study demonstrated a time-course change in efficacy, with deterioration during the repetitive administration phase and preservation during the post-treatment phase. The current findings support further development of optimization of objective, quantitative, and repeatable outcome measures for autistic social deficits and to establish optimized regimen of oxytocin treatment for ASD. Abbreviations Abbreviations ADOS Autism Diagnostic Observation Schedule ASD autism spectrum disorder PDF probability density function Acknowledgements The study design was reviewed and approved by the institutional review board of The University of Tokyo Hospital. Neither the funder nor sponsor, the Strategic Research Program for Brain Sciences from Japan Agency for Medical Research and Development (JP18dm0107134), had any involvement in the data collection, analyses, writing, or interpretation of the study. However, the sponsor participated in discussion regarding which sites should be included in the trial and how to best interpret the results. The corresponding author had full access to the data and held the final responsibility regarding the decision to submit for publication. Wendy Brooks, PhD, and Benjamin Knight, MSc, from Edanz Group (www.edanzediting.com/ac) provided English language editing services for a draft of this manuscript. Funding This study was partially supported by the Strategic Research Program for Brain Sciences from Japan Agency for Medical Research and Development (JP18dm0107134). Competing interests The authors report no competing interests. References Aas IM . Guidelines for rating Global Assessment of Functioning (GAF) . Ann Gen Psychiatry 2011 ; 10 : 1 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat American Psychiatric Association. Diagnostic and statistical manual of mental disorders . 4th edn . Text Revision (DSM-IV-TR). Arlington, VA : American Psychiatric Association ; 2000 . WorldCat Anagnostou E , Soorya L , Chaplin W , Bartz J , Halpern D , Wasserman S , et al. Intranasal oxytocin versus placebo in the treatment of adults with autism spectrum disorders: a randomized controlled trial . Mol Autism 2012 ; 3 : 16 . Google Scholar Crossref Search ADS PubMed WorldCat Andari E , Duhamel J-R , Zalla T , Herbrecht E , Leboyer M , Sirigu A . Promoting social behavior with oxytocin in high-functioning autism spectrum disorders . Proc Natl Acad Sci USA 2010 ; 107 : 4389 – 94 . Google Scholar Crossref Search ADS PubMed WorldCat Aoki Y , Watanabe T , Abe O , Kuwabara H , Yahata N , Takano Y , et al. Oxytocin’s neurochemical effects in the medial prefrontal cortex underlie recovery of task-specific brain activity in autism: a randomized controlled trial . Mol Psychiatry 2015 ; 20 : 447 – 53 . Google Scholar Crossref Search ADS PubMed WorldCat Aoki Y , Yahata N , Watanabe T , Takano Y , Kawakubo Y , Kuwabara H , et al. Oxytocin improves behavioural and neural deficits in inferring others’ social emotions in autism . Brain 2014 ; 137 : 3073 – 86 . Google Scholar Crossref Search ADS PubMed WorldCat Baird G , Simonoff E , Pickles A , Chandler S , Loucas T , Meldrum D , et al. Prevalence of disorders of the autism spectrum in a population cohort of children in South Thames: The Special Needs and Autism Project (SNAP) . Lancet 2006 ; 368 : 210 – 15 . Google Scholar Crossref Search ADS PubMed WorldCat Bales KL , Perkeybile AM , Conley OG , et al. Chronic intranasal oxytocin causes long-term impairments in partner preference formation in male prairie voles . Biol Psychiatry 2013 ; 74 : 180 – 88 . Google Scholar Crossref Search ADS PubMed WorldCat Bales KL , Solomon M , Jacob S , Crawley JN , Silverman JL , Larke RH , et al. Long-term exposure to intranasal oxytocin in a mouse autism model . Transl Psychiatry 2014 ; 4 : e480 . Google Scholar Crossref Search ADS PubMed WorldCat Benner S , Aoki Y , Watanabe T , Endo N , Abe O , Kuroda M , et al. Neurochemical evidence for differential effects of acute and repeated oxytocin administration . Mol Psychiatry 2018 . doi: 10.1038/s41380–018–0249–4 . WorldCat Cohen AS , Morrison SC , Callaway DA . Computerized facial analysis for understanding constricted/blunted affect: initial feasibility, reliability, and validity data . Schizophr Res 2013 ; 148 : 111 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Domes G , Kumbier E , Heinrichs M , Herpertz SC . Oxytocin promotes facial emotion recognition and amygdala reactivity in adults with asperger syndrome . Neuropsychopharmacology 2014 ; 39 : 698 – 706 . Google Scholar Crossref Search ADS PubMed WorldCat Fujiwara T , Mizuki R , Miki T , Chemtob C . Association between facial expression and PTSD symptoms among young children exposed to the Great East Japan Earthquake: a pilot study . Front Psychol 2015 ; 6 : 1534 . Google Scholar Crossref Search ADS PubMed WorldCat Gordon I , Vander Wyk BC , Bennett RH , Cordeaux C , Lucas MV , Eilbott JA , et al. Oxytocin enhances brain function in children with autism . Proc Natl Acad Sci USA 2013 ; 110 : 20953 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Guastella AJ , Einfeld SL , Gray KM , Rinehart NJ , Tonge BJ , Lambert TJ , et al. Intranasal oxytocin improves emotion recognition for youth with autism spectrum disorders . Biol Psychiatry 2010 ; 67 : 692 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat Guastella AJ , Gray KM , Rinehart NJ , et al. The effects of a course of intranasal oxytocin on social behaviors in youth diagnosed with autism spectrum disorders: a randomized controlled trial . J Child Psychol Psychiatry 2015 ; 56 : 444 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat Guastella AJ , Hickie IB . Oxytocin treatment, circuitry, and autism: a critical review of the literature placing oxytocin into the autism context . Biol Psychiatry 2016 ; 79 : 234 – 42 . Google Scholar Crossref Search ADS PubMed WorldCat Guy W . Assessment manual for psychopharmacology . Rockville, MD : National Institute of Mental Health ; 1976 . Google Preview WorldCat Harris JC , Carter CS . Therapeutic interventions with oxytocin: current status and concerns . J Am Acad Child Adolesc Psychiatry 2013 ; 52 : 998 – 1000 . Google Scholar Crossref Search ADS PubMed WorldCat Hurlemann R , Scheele D . Dissecting the Role of Oxytocin in the Formation and Loss of Social Relationships . Biol Psychiatry 2016 ; 79 : 185 – 93 . Google Scholar Crossref Search ADS PubMed WorldCat Insel TR , Winslow JT , Witt DM . Homologous regulation of oxytocin receptors . Endocrinology 1992 ; 130 : 2602 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Keech B , Crowe S , Hocking DR . Intranasal oxytocin, social cognition and neurodevelopmental disorders: A meta-analysis . Psychoneuroendocrinology 2018 ; 87 : 9 – 19 . Google Scholar Crossref Search ADS PubMed WorldCat Kosaka H , Okamoto Y , Munesue T , Yamasue H , Inohara K , Fujioka T , et al. Oxytocin efficacy is modulated by dosage and oxytocin receptor genotype in young adults with high-functioning autism: a 24-week randomized clinical trial . Transl Psychiatry 2016 ; 6 : e872 . Google Scholar Crossref Search ADS PubMed WorldCat Lai M-C , Lombardo M V , Baron-Cohen S . Autism . Lancet 2014 ; 383 : 896 – 910 . Google Scholar Crossref Search ADS PubMed WorldCat Leng G , Ludwig M . Intranasal oxytocin: myths and delusions . Biol Psychiatry 2016 ; 79 : 243 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat Lewinski P , Den Uyl TM , Butler C . Automated facial coding: validation of basic emotions and FACS AUs in facereader . J Neurosci Psychol Econ 2014 ; 7 : 227 – 36 . Google Scholar Crossref Search ADS WorldCat Lord C , Rutter M , Goode S , Heemsbergen J , Jordan H , Mawhood L , et al. Autism diagnostic observation schedule: a standardized observation of communicative and social behavior . J Autism Dev Disord 1989 ; 19 : 185 – 212 . Google Scholar Crossref Search ADS PubMed WorldCat Lord C , Rutter M , Le Couteur A . Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders . J Autism Dev Disord 1994 ; 24 : 659 – 85 . Google Scholar Crossref Search ADS PubMed WorldCat Owada K , Kojima M , Yassin W , Lee MH , Guoynes CD , Downing GM , et al. Computer-analyzed facial expression as a surrogate marker for autism spectrum social core symptoms . PLoS One 2018 ; 13 : e0190442 . Google Scholar Crossref Search ADS PubMed WorldCat Radloff LS . The CES-D Scale: a self-report depression scale for research in the general population . Appl Psychol Measur 1977 ; 1 : 385 – 401 . Google Scholar Crossref Search ADS WorldCat Spielberger CD , Gorsuch RL , Lushene RE . Manual for the State-Trait Anxiety Inventory . Palo Alto, CA : Consulting Psychologists Press ; 1970 . Google Preview WorldCat Walum H , Waldman ID , Young LJ . Statistical and methodological considerations for the interpretation of intranasal oxytocin studies . Biol Psychiatry 2016 ; 79 : 251 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat Watanabe T , Abe O , Kuwabara H , Yahata N , Takano Y , Iwashiro N , et al. Mitigation of sociocommunicational deficits of autism through oxytocin-induced recovery of medial prefrontal activity: a randomized trial . JAMA Psychiatry 2014 ; 71 : 166 – 75 . Google Scholar Crossref Search ADS PubMed WorldCat Watanabe T , Kuroda M , Kuwabara H , Aoki Y , Iwashiro N , Tatsunobu N , et al. Clinical and neural effects of six-week administration of oxytocin on core symptoms of autism . Brain 2015 ; 138 : 3400 – 12 . Google Scholar Crossref Search ADS PubMed WorldCat Wechsler D . The psychometric tradition: developing the Wechsler adult intelligence scale . Contemp Educ Psychol 1981 ; 6 : 82 – 5 . Google Scholar Crossref Search ADS WorldCat Xu G , Strathearn L , Liu B , Bao W . Prevalence of autism spectrum disorder among US children and adolescents, 2014–2016 . JAMA 2018 ; 319 : 81 – 2 . Google Scholar Crossref Search ADS PubMed WorldCat Yamamoto Y , Cushing BS , Kramer KM , Epperson PD , Hoffman GE , Carter CS . Neonatal manipulations of oxytocin alter expression of oxytocin and vasopressin immunoreactive cells in the paraventricular nucleus of the hypothalamus in a gender specific manner . Neuroscience 2004 ; 125 : 947 – 55 . Google Scholar Crossref Search ADS PubMed WorldCat Yamasue H , Domes G . Oxytocin and autism spectrum disorders . Curr Top Behav Neurosci 2018 ; 35 : 449 – 65 . Google Scholar Crossref Search ADS PubMed WorldCat Yamasue H , Okada T , Munesue T , Kuroda M , Fujioka T , Uno Y , et al. Effect of intranasal oxytocin on the core social symptoms of autism spectrum disorder: a randomized clinical trial . Mol Psychiatry 2018 . doi: 10.1038/s41380–018–0097–2 . WorldCat Yamasue H . Promising evidence and remaining issues regarding the clinical application of oxytocin in autism spectrum disorders . Psychiatry Clin Neurosci 2016 ; 70 : 89 – 99 . Google Scholar Crossref Search ADS PubMed WorldCat Yatawara CJ , Einfeld SL , Hickie IB , Alvares GA , Tonge BJ , Hickie IB , et al. The effect of oxytocin nasal spray on social interaction deficits observed in young children with autism: a randomized clinical crossover trial . Mol Psychiatry 2016 ; 21 : 1225 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) (2019). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Brain Oxford University Press

Loading next page...
 
/lp/oxford-university-press/quantitative-facial-expression-analysis-revealed-the-efficacy-and-time-OUGnxklw6I

References (52)

Publisher
Oxford University Press
Copyright
© The Author(s) (2019). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
0006-8950
eISSN
1460-2156
DOI
10.1093/brain/awz126
Publisher site
See Article on Publisher Site

Abstract

Abstract Discrepancies in efficacy between single-dose and repeated administration of oxytocin for autism spectrum disorder have led researchers to hypothesize that time-course changes in efficacy are induced by repeated administrations of the peptide hormone. However, repeatable, objective, and quantitative measurement of autism spectrum disorder’s core symptoms are lacking, making it difficult to examine potential time-course changes in efficacy. We tested this hypothesis using repeatable, objective, and quantitative measurement of the core symptoms of autism spectrum disorder. We examined videos recorded during semi-structured social interaction administered as the primary outcome in single-site exploratory (n = 18, crossover within-subjects design) and multisite confirmatory (n = 106, parallel-group design), double-blind, placebo-controlled 6-week trials of repeated intranasal administrations of oxytocin (48 IU/day) in adult males with autism spectrum disorder. The main outcomes were statistical representative values of objectively quantified facial expression intensity in a repeatable part of the Autism Diagnostic Observation Schedule: the maximum probability (i.e. mode) and the natural logarithm of mode on the probability density function of neutral facial expression and the natural logarithm of mode on the probability density function of happy expression. Our recent study revealed that increases in these indices characterize autistic facial expression, compared with neurotypical individuals. The current results revealed that oxytocin consistently and significantly decreased the increased natural logarithm of mode on the probability density function of neutral facial expression compared with placebo in exploratory (effect-size, −0.57; 95% CI, −1.27 to 0.13; P = 0.023) and confirmatory trials (−0.41; −0.62 to −0.20; P < 0.001). A significant interaction between time-course (at baseline, 2, 4, 6, and 8 weeks) and the efficacy of oxytocin on the natural logarithm of mode on the probability density function of neutral facial expression was found in confirmatory trial (P < 0.001). Post hoc analyses revealed maximum efficacy at 2 weeks (P < 0.001, Cohen’s d = −0.78; 95% CI, −1.21 to −0.35) and deterioration of efficacy at 4 weeks (P = 0.042, Cohen’s d = −0.46; 95% CI, −0.90 to −0.01) and 6 weeks (P = 0.10, Cohen’s d = −0.35; 95% CI, −0.77 to 0.08), while efficacy was preserved at 2 weeks post-treatment (i.e. 8 weeks) (P < 0.001, Cohen’s d = −1.24; 95% CI, −1.71 to −0.78). Quantitative facial expression analyses successfully verified the positive effects of repeated oxytocin on autistic individuals’ facial expressions and demonstrated a time-course change in efficacy. The current findings support further development of an optimized regimen of oxytocin treatment. Asperger syndrome, pervasive developmental disorder, neuroendocrinology, neuropeptide, surrogate marker Introduction Autism spectrum disorder (ASD) is a highly prevalent developmental disorder presenting core symptoms including deficits in social interaction and communication and restricted and repetitive behaviours (Baird et al., 2006; Xu et al., 2018). No approved medication is currently available to reduce the core symptoms of ASD, which are considered static traits rather than treatable symptoms (Lai et al., 2014). This situation may be related to a lack of research attention to the development of repeatable, objective, and quantitative longitudinal assessment tools for the core symptoms of ASD (Yamasue, 2016). Although oxytocin has recently been expected as a novel therapeutic for ASD core symptoms, both the previous findings and the outcomes assessing the efficacy have been controversial (Andari et al., 2010; Harris and Carter, 2013; Guastella et al., 2016; Hurlemann and Scheele, 2016; Leng and Ludwig, 2016; Walum et al., 2016; Yamasue and Domes, 2018). Previous studies have consistently demonstrated that single-dose administration of oxytocin significantly improves experimental measures associated with ASD core symptoms (Guastella et al., 2010; Gordon et al., 2013; Aoki et al., 2014, 2015; Domes et al., 2014; Watanabe et al., 2014). However, previous studies of repeated oxytocin treatment on ASD core symptoms have yielded inconsistent results (Guastella et al. 2015; Kosaka et al., 2016; Yatawara et al., 2016). Our exploratory trial revealed a significant improvement in Autism Diagnostic Observation Schedule (ADOS) reciprocity induced by 6 weeks of repeated administration of oxytocin (Watanabe et al., 2015), although the confirmatory trial did not reveal superiority in the efficacy of oxytocin compared with placebo using the same primary end-point in the similar demographic adult male individuals with high-functioning ASD (Yamasue et al., 2018). The discrepancy in efficacy could, at least partially, be derived from a significant placebo effect on ADOS reciprocity in the confirmatory trial (Yamasue et al., 2018). In contrast to the discrepant results regarding the primary end-point, repeated administration of oxytocin has shown significant effects on objectively quantified behavioural and neural measures during a social behaviour task and eye-gaze measurement on socially-relevant regions in both exploratory and confirmatory trials (Watanabe et al., 2015; Yamasue et al., 2018). Based on these findings, taken together with a previous exploratory trial from another research group (Anagnostou et al., 2012), it could reasonably be predicted that objective assessments of social behaviour are likely to reveal the efficacy of repeated administrations of oxytocin (Anagnostou et al., 2012; Watanabe et al., 2015; Yamasue et al., 2018). The discrepancy in efficacy between single and repeated administrations of oxytocin also indicates that repeated administration might induce a decline in efficacy (Watanabe et al., 2015; Yamasue, 2016). Our recent study supports this notion, consistently demonstrating differential neurochemical effects of single and repeated administrations of oxytocin on medial frontal cortices in human neuroimaging and examination of gene expressions in mice (Benner et al., 2018). Furthermore, a similar discrepancy has also been observed in animal studies: while acute oxytocin administration enhanced animal prosocial behaviour, repeated administration diminished these effects rather than increasing them, suggesting potential saturation, downregulation, or negative feedback of oxytocin receptors (Bales et al., 2013, 2014). However, previous human studies have not fully addressed this issue. The present study tested the effects of oxytocin on repeatable, objective, and quantitative measurement of facial expression, using a method recently developed in our laboratory (Owada et al., 2018), during social interactions in individuals with ASD. The first aim of the current study was to test and verify the effects of repeated administration of intranasal oxytocin on autistic social behaviour using objective quantitative analysis of facial expression. The second aim was to detect a potential time-course change in efficacy of oxytocin on autistic social deficits using the repeatability of quantitative face analysis. To explore these issues, we examined videos recorded during ADOS administered in two independent clinical trials (Watanabe et al., 2015; Yamasue et al., 2018). We first conducted an exploratory test of the study hypothesis in a small sample dataset of a single-site exploratory trial (Watanabe et al., 2015). We then conducted a confirmatory test to examine this effect, and explored potential time-course changes in efficacy in a larger sample dataset in our confirmatory multisite trial with longitudinal assessments (Yamasue et al., 2018). Materials and methods The ethics committee of University of Tokyo Hospital approved the current quantitative study of facial expression (#10245). After a complete explanation of the study, the mental capacity to consent was confirmed for each participant by a psychiatrist (H.Y. or K.O.), and written informed consent was obtained from all participants. The study protocols of the exploratory and confirmatory trials were approved by the institutional review boards at each site and registered in the University Hospital Medical Information Network Clinical Trials Registry (UMIN000007122 and UMIN000015264). Although the results on the primary and secondary outcomes, except for the quantitative measurement of facial expression in these trials, have been published previously (Watanabe et al., 2015; Yamasue et al., 2018), the quantitative facial expression measurement was an unpublished key secondary outcome of these trials and was newly examined in the current study. Study design and participants Exploratory trial The exploratory trial was a randomized, double-blind, placebo-controlled, crossover trial, which was primarily conducted in an outpatient clinic of The University of Tokyo Hospital (Watanabe et al., 2015). The inclusion criteria were: (i) age 18–54 years; (ii) male; (iii) diagnosis of autistic disorder, Asperger’s disorder, or pervasive developmental disorder-not otherwise specified (PDD-NOS) based on the Diagnostic and Statistical Manual-Revision IV-Text Revision (DSM-IV-TR) (American Psychiatric Association, 2000); and (iv) verbal intelligence quotient (IQ) above 85 and full IQ above 80 as measured using the Wechsler Adult Intelligent Scale (WAIS)-R (Wechsler, 1981). Exclusion criteria were: (i) instability of symptoms of comorbid mental disorders such as mood disorder or anxiety disorder; (ii) history of hypersensitivity to oxytocin; (iii) history of seizures or traumatic brain injury with loss of consciousness for longer than 5 min; and (iv) history of alcohol-related disorders, substance abuse, or addiction. Participants with contraindications to undergoing MRI scanning were also excluded in the exploratory trial, which involved functional MRI indices as a secondary outcome (Watanabe et al., 2015). An experienced psychiatrist (H.Y.) performed the diagnoses based on DSM-IV-TR (American Psychiatric Association, 2000) with more than 2 months of follow-up examinations. Another certified psychiatrist/psychologist confirmed the diagnoses using the Japanese version of the Autism Diagnostic Interview-Revised (ADIR) (Lord et al., 1994) (H.K.) and ADOS (Lord et al., 1989) (M.K.). Between March and December 2012, 20 high-functioning adult male individuals with ASD were enrolled in this trial, and final assessment at the 12-week end-point was performed between August 2012 and April 2013. Confirmatory trial This multicentre, parallel-group, placebo-controlled, double-blind, confirmatory trial of intranasal oxytocin in participants with ASD was conducted at the University of Tokyo Hospital, Nagoya University Hospital, Kanazawa University Hospital, and University of Fukui Hospital in Japan (Yamasue et al., 2018). The inclusion and exclusion criteria were similar to those in the exploratory trial. The inclusion criteria were: (i) age: 18–54 years; (ii) male; (iii) diagnosis of autistic disorder, Asperger’s disorder, or PDD-NOS based on DSM-IV-TR (American Psychiatric Association, 2000); (iv) score exceeding the cut-off value of 10 for qualitative abnormalities in social reciprocity on ADIR (Lord et al., 1994); and (v) verbal IQ above 85 and full IQ above 80 as measured using WAIS-III (Wechsler, 1981). Exclusion criteria were: (i) primary psychiatric diagnoses other than those listed in the inclusion criteria; (ii) instability of symptoms of comorbid mental disorders such as mood disorder or anxiety disorder; (iii) history of changes in medication or doses of psychotropics within 1 month before randomization; (iv) current treatment with more than two psychotropics; (v) current medication for comorbid attention-deficit/hyperactivity disorder (i.e. atomoxetine or methylphenidate, which were the only approved medications in Japan at the time of the study); (vi) history of continual oxytocin treatment; (vii) history of hypersensitivity to oxytocin; (viii) history of seizures or traumatic brain injury with loss of consciousness for longer than 5 min; and (ix) history of alcohol-related disorders, substance abuse, or addiction. The recruitment, which was open to the public, and the eligibility testing processes at the recruiting office are described in detail elsewhere (Yamasue et al., 2018) and in the Supplementary material. A psychiatrist (H.Y., T.O., Y.U., T.M., or H.Ko.) experienced in developmental disorders performed the diagnoses of ASD based on DSM-IV-TR. Another certified psychiatrist/psychologist confirmed the diagnoses using the Japanese version of the ADIR (Lord et al., 1994) (H.Ku., M.Ku., or T.F.) and ADOS (Lord et al., 1989) (M.Ku., Y.Y., or K.M.) (Table 1). The trial enrolled 106 males with high-functioning ASD between January 2015 and March 2016. End-point assessments were performed between March 2015 and April 2016. Table 1 Demographic data of the exploratory and confirmatory trial Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 aSocioeconomic status (SES) assessed using the Hollingshead Index; higher scores indicate lower status. bThe intelligence quotients were measured using the Wechsler Adult Intelligence Scale. IQ = intelligence quotient; OT = oxytocin; PL = placebo; SD = standard deviation. View Large Table 1 Demographic data of the exploratory and confirmatory trial Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 Exploratory crossover trial Confirmatory parallel group trial OT-PL group (n = 9) Mean (SD) PL-OT group (n = 9) Mean (SD) P-value OT-group (n = 51) Mean (SD) PL-group (n = 52) Mean (SD) P-value Age [range] 35.1 (7.6), [24–42] 29.3 (5.9) [24–43] 0.09 27.8 (7.6) [18–48] 26.4 (7.1) [18–48] 0.34 Height, cm 169.7 (3.8) 167.3 (4.9) 0.27 171.0 (6.3) 172.5 (5.8) 0.19 Body weight, kg 65.4 (12.6) 66.4 (18.0) 0.89 67.4 (12.3) 65.9 (11.1) 0.51 SESa 2.7 (0.97) 2.6 (1.0) 0.64 3.2 (1.0) 2.8 (1.2) 0.14 Parental SESa 2.0 (0.50) 2.4 (0.73) 0.15 2.2 (0.60) 2.2 (0.62) 0.85 Handedness: right/mixed/left 9/0/0 6/1/2 49/2 49/3 IQb Full IQ 109.3 (9.1) 101.8 (12.6) 0.17 105.7 (13.3) 108.6 (14.5) 0.29 Verbal IQ 119.6 (8.1) 104.5 (12.0) 0.01 111.7 (13.1) 114.3 (15.2) 0.35 Performance IQ 90.9 (12.6) 97.8 (18.9) 0.4 96.3 (15.5) 99.4 (15.0) 0.30 Autism Diagnostic Interview-Revised Social 14.0 (6.7) 15.6 (7.2) 0.67 21.2 (5.0) 22.2 (5.4) 0.33 Communication 13.0 (4.8) 11.3 (3.5) 0.44 15.6 (4.0) 16.4 (3.7) 0.29 Repetitive 4.1 (2.9) 4.6 (2.1) 0.74 5.2 (2.5) 5.8 (2.2) 0.18 aSocioeconomic status (SES) assessed using the Hollingshead Index; higher scores indicate lower status. bThe intelligence quotients were measured using the Wechsler Adult Intelligence Scale. IQ = intelligence quotient; OT = oxytocin; PL = placebo; SD = standard deviation. View Large Intervention In the exploratory trial, participants received oxytocin (24 IU, Syntocinon-Spray; Novartis) in the morning and afternoon over six consecutive weeks (i.e. 48 IU/day) and placebo in the same way with cross-over administration (Fig. 1A). In the confirmatory trial, participants received oxytocin (48 IU/day) or placebo in the morning and afternoon over six consecutive weeks (Fig. 1B). The placebo contained all the inactive ingredients included in the oxytocin spray. On the last day of each 6-week administration, participants underwent end-point clinical assessments, including ADOS, 15 min after the last drug administration. All participants underwent sufficient training for intranasal administration with identical instructions (Supplementary material), and the effectiveness of training was confirmed at each 2-week assessment. Treatment adherence was assessed with a self-report daily record (Supplementary material). Figure 1 View largeDownload slide Study designs. (A) Exploratory trial in crossover design. (B) Confirmatory trial in parallel group design. Figure 1 View largeDownload slide Study designs. (A) Exploratory trial in crossover design. (B) Confirmatory trial in parallel group design. Randomization and masking Randomization and masking of drug administration was randomly assigned to participants in the two groups in a one-to-one ratio based on a computer-generated randomized order. In the exploratory crossover trial, this included a group in which oxytocin was administered first and a group in which placebo was given first, and, in the confirmatory trial, this included either the oxytocin or placebo group. In the confirmatory parallel-group trial, randomization was stratified according to study site and median ADIR score (<18, ≥18, based on the exploratory trial). Oxytocin and placebo were stored in visually-identical spray bottles (Victoria Pharmacy). Bottle labels were covered to keep drug types unknown to all the participants, their families, clinicians, and assessors including ADOS administrators and assessors. The registration, allocation, and data management procedures were defined separately (Supplementary material). Outcomes The outcome of the current study, facial expression intensity values, was common to the exploratory and confirmatory trials, and was described as a secondary outcome of these trials. Expression intensities during a semi-structured socially interactive situation extracted from ADOS module 4 (Lord et al., 1989) were quantified by a dedicated software program (FaceReader version 6·1; Noldus Information Technology Inc., Wageningen, The Netherlands), and were assessed at baseline, 6- and 12-week end-points in the exploratory trial, or every 2 weeks in the confirmatory trial (i.e. baseline; treatment Weeks 2, 4, and 6; and 2 weeks post-treatment). At the time of planning our study, we compared several available software programs dedicated to facial expression analysis. Among them, we found that FaceReader had the most credible references regarding the methodology and algorithms involved (Cohen et al., 2013; Lewinski et al., 2014; Fujiwara et al., 2015), in addition to its ability to output frame-by-frame emotional intensity values. Therefore, we used FaceReader in our study. To ensure a uniform and consistent setting for observing facial expressions, an activity (i.e. ‘Cartoons’) of ADOS was selected based on the degree of structure needed to orient a participant to a certain consistent response, repeatability to elicit improvised responses, and the feasibility of whether faces remained positioned such that the software could correctly recognize expressions and whether the activity length was short enough to observe situation-evoked facial expression with minimal contamination. The degrees of structure, repeatability, and feasibility were qualitatively and quantitatively assessed, and selection was performed based on the results from statistical comparisons between these parameters of ADOS activities. The process by which activity was selected from ADOS is described in detail elsewhere (Owada et al., 2018) and in the Supplementary material. The ADOS was administered by clinicians who were certificated or who completed a training course regarding the research use of ADOS and whose credentials had been validated by another certified administrator. The methods by which expression intensities were recorded and calculated are described in detail elsewhere (Owada et al., 2018) and in the Supplementary material. In brief, time-trend datasets from the selected activity of each participant yielded a pair of expression intensity variables. These represented the distribution of expression intensity values for each facial expression element: the value at which the probability was maximum (i.e. mode) and the natural logarithm of the maximum probability value on the probability density function (i.e. log-PDFmode), which was estimated by kernel density using Gaussian kernels. The mode gives a measure of the most likely intensity for the given time period, while the log-PDFmode gives a measure of the variability of intensity for the given time period. We used the increased mode and log-PDFmode of neutral facial expression and the log-PDFmode of happy facial expression as outcomes, as our previous study demonstrated that increases in these expression intensities can characterize facial expressions in individuals with ASD well, in comparison with neurotypical individuals (Owada et al., 2018). Statistical analysis The target number of participants was set as 20 in the exploratory trial based on previous studies (Andari et al., 2010; Guastella et al., 2010), while a target of 114 was calculated for the confirmatory trial based on power analyses using the results from the exploratory trial [see details in our previous reports (Watanabe et al., 2015; Yamasue et al., 2018) and Supplementary material]. Statistical analyses were performed using IBM SPSS 22.0 (IBM Corp., Armonk, NY), MATLAB R2015a (The MathWorks, Natick, MA), R 3.3.2 (http://www.R-project.org/), and Python libraries for scientific computation (NumPy, and SciPy). The outcomes were analysed using generalized estimating equations with an unstructured correlation and robust standard error estimates considering the time points as a factor (i.e. 0, 6, and 12 weeks in the exploratory trial; 0, 2, 4, 6, and 8 weeks in the confirmatory trial). Site was also considered as a factor in the multisite confirmatory trial. Main effects and interactions for the pharmacologically different conditions (oxytocin/placebo) and the factors were estimated with a significance level of P < 0.05. The least square mean between group differences and the associated 95% confidence intervals (CI) at the end-point were also estimated. The estimated marginal means (EM means) for levels of factors and factor interactions were also calculated. Data availability Individual de-identified participant data underlying the results reported in this study, as well as the study protocol, statistical analysis plan, and informed consent form are available on request from investigators providing a methodologically sound proposal and whose proposed use of the data has been approved by an independent review committee identified for this purpose during the period beginning 9 months and ending 5 years following article publication. Proposals should be directed to the corresponding author. Results Data and results from the exploratory trial The detailed participant flow in the exploratory trial has been reported elsewhere (Fig. 1A aaand Table 1) (Watanabe et al., 2015). Expression intensities were analysed for 18 individuals at three time points except for three (two in the oxytocin-first group at the end-point of placebo period and one in placebo-first group at the end-point of oxytocin period) for which data failed to be recorded. No major adverse effects were observed in either group (see details in Watanabe et al., 2015). The efficacy of oxytocin was analysed using generalized estimation equations with an unstructured correlation and robust standard error estimates considering the time points as a factor (i.e. 0, 6, and 12 weeks). We found that 6-week oxytocin administration significantly reduced the log-PDFmode of neutral face in the participants with ASD (P = 0.023, Cohen’s Cohen’s d = −0.57; 95% CI, −1.27 to 0.13; Fig. 2 and Table 2) with no significant main effect of time point and no significant interaction between drug type (oxytocin/placebo) and time point (P > 0.87). As the log-PDFmode of neutral facial expression was significantly increased in the participants with ASD compared with neurotypical controls before administration of oxytocin (Owada et al., 2018), the effect of oxytocin was observed in the direction of recovery. No significant improvement of the mode of neutral facial expression or the log-PDFmode of happy facial expression was found (P > 0.06). Table 2 Effect of oxytocin on facial expression intensities Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) aBy generalized estimating equations. bEstimated by EM means. Values are presented as mean (SD). EI = expression intensity; log-PDFmode = natural logarithm of maximum probability on the probability density function; OT = oxytocin; PL = placebo. View Large Table 2 Effect of oxytocin on facial expression intensities Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Exploratory crossover trial EI Baseline 6 weeks 12 weeks OT-induced change (n = 17) PL-induced change (n = 16) P-valuea Cohen’s db Neutral, mode OT-PL (n = 9) 0.47 (0.04) 0.40 (0.05) 0.45 (0.03) −0.04 (0.09) −0.02 (0.07) 0.15 −0.26 PL-OT (n = 9) 0.46 (0.08) 0.44 (0.02) 0.45 (0.04) Neutral, log-PDFmode OT-PL (n = 9) 2.04 (0.26) 1.75 (0.30) 2.08 (0.46) −0.29 (0.53) −0.04 (0.41) 0.023 −0.57 PL-OT (n = 9) 2.04 (0.46) 1.89 (0.26) 1.63 (0.39) Happy, log-PDFmode OT-PL (n = 9) 5.43 (3.43) 4.91 (1.76) 5.46 (2.40) −0.59 (2.79) 1.27 (3.86) 0.064 −0.59 PL-OT (n = 9) 6.01 (3.53) 7.62 (3.96) 5.55 (2.38) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) Confirmatory parallel group trial EI Baseline 2 weeks 4 weeks 6 weeks (end) 8 weeks P-valuea Cohen’s db Neutral, mode OT (n = 49) 0.45 (0.11) 0.45 (0.16) 0.49 (0.15) 0.44 (0.15) 0.48 (0.14) 0.074 0.19 PL (n = 47) 0.44 (0.11) 0.46 (0.11) 0.46 (0.11) 0.45 (0.09) 0.45 (0.10) Neutral, log-PDFmode OT (n = 49) 1.87 (0.41) 1.70 (0.50) 1.81 (0.52) 1.76 (0.56) 1.74 (0.58) <0.001 −0.41 PL (n = 47) 1.63 (0.61) 1.68 (0.56) 1.76 (0.63) 1.66 (0.45) 1.70 (0.54) Happy, log-PDFmode OT (n = 49) 7.74 (3.93) 7.02 (3.73) 7.71 (4.08) 7.29 (4.08) 7.23 (4.10) 0.25 −0.12 PL (n = 47) 7.65 (4.03) 7.39 (3.28) 8.03 (3.26) 8.06 (3.71) 8.41 (3.18) aBy generalized estimating equations. bEstimated by EM means. Values are presented as mean (SD). EI = expression intensity; log-PDFmode = natural logarithm of maximum probability on the probability density function; OT = oxytocin; PL = placebo. View Large Figure 2 View largeDownload slide Effects of intranasal oxytocin on facial expressions in the exploratory trial. Effects of oxytocin or placebo administration on facial expression evaluated by the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (EIs). Estimated marginal means (EM means) at baseline, 6 and 12 weeks of placebo and oxytocin administrations are shown for placebo-first and oxytocin-first groups individually. The bars represent 95% CIs of EM means. The effect sizes between placebo and oxytocin administrations at 6 weeks (Cohen’s d = −0.36; 95% CI, −1.29 to 0.58) and 12 weeks (Cohen’s d = −0.76; 95% CI, −1.80 to 0.31) were also presented. Overall, oxytocin administration showed a significant reduction in log-PDFmode of neutral expression intensities compared with placebo (P = 0.023, Cohen’s d = −0.57; 95% CI, −1.27 to 0.13). Figure 2 View largeDownload slide Effects of intranasal oxytocin on facial expressions in the exploratory trial. Effects of oxytocin or placebo administration on facial expression evaluated by the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (EIs). Estimated marginal means (EM means) at baseline, 6 and 12 weeks of placebo and oxytocin administrations are shown for placebo-first and oxytocin-first groups individually. The bars represent 95% CIs of EM means. The effect sizes between placebo and oxytocin administrations at 6 weeks (Cohen’s d = −0.36; 95% CI, −1.29 to 0.58) and 12 weeks (Cohen’s d = −0.76; 95% CI, −1.80 to 0.31) were also presented. Overall, oxytocin administration showed a significant reduction in log-PDFmode of neutral expression intensities compared with placebo (P = 0.023, Cohen’s d = −0.57; 95% CI, −1.27 to 0.13). Although individuals with unstable comorbid mental disorders including moderate or severe anxiety were excluded, we additionally tested the possibility that anxiety or depressive tendencies of participants could confound the effects of oxytocin using analyses controlling for scores on the State-Trait Anxiety Inventory (STAI) (Spielberger et al., 1970) and The Center for Epidemiologic Studies Depression Scale (CESD) (Radloff, 1977). The results confirmed that the individual differences in baseline anxious or depressive tendency were not correlated with oxytocin’s efficacy regarding the log-PDFmode for neutral facial expression (P > 0.28). When baseline anxious or depressive tendencies were controlled as covariates, the efficacy of oxytocin on the variability of neutral facial expression was totally preserved (P < 0.025). Data and results from the confirmatory trial The detailed participant flow in the confirmatory trial has been reported elsewhere (Fig. 1B) (Yamasue et al., 2018). The current study analysed expression intensities for 103 individuals at five time points (515 data points) (Table 1), with the exception of 60 data points (25 in the oxytocin group and 35 in the placebo group), which were excluded because of recording failure, defocused video images, or poor facial recognition rate. Ninety-four of the participants were free from psychotropics other than oxytocin throughout the trial period, while 12 continued their psychotropic medications during the period [four antipsychotics, four antidepressants, two hypnotics, and two anticonvulsants (mood stabilizers)]. Eighty-three participants were diagnosed with high-functioning autistic disorder, 12 with Asperger’s disorder, and 11 with PDD-NOS. No significant differences between the oxytocin and placebo groups were observed in background information or prevalence of adverse events [see details elsewhere (Yamasue et al., 2018) and in Table 1]. Consistent with the exploratory trial, the effect of oxytocin was analysed using generalized estimating equations with an unstructured correlation and robust standard error estimates considering the time points as a factor (i.e. 0, 2, 4, 6, and 8 weeks). Site was also considered as a factor in the multisite confirmatory trial. Regarding the expression intensities, consistent with the exploratory trial, the log-PDFmode of neutral facial expression was significantly reduced in the oxytocin-administered group compared with the placebo-administered group across the assessment period (P < 0.001, Cohen’s d = −0.41; 95% CI, −0.62 to −0.20). A significant interaction between time point and treatment group was also found on the log-PDFmode of neutral facial expression (P < 0.001). As shown in Fig. 3, EM means and effect sizes at each time point revealed a time course change in the efficacy of oxytocin: a maximum effect was found at 2-weeks post-treatment (P < 0.001, Cohen’s d = −1.24; 95% CI, −1.71 to −0.78) compared with 6-week administration period such as 2 weeks (P = 0.014, Cohen’s d = −0.53; 95% CI, −0.94 to −0.10), 4 weeks (P = 0.068, Cohen’s d = −0.41; 95% CI, −0.85 to 0.03) and 6 weeks (P = 0.054, Cohen’s d = −0.41; 95% CI, −0.83 to 0.02). However, even after exclusion of the 2-week post-treatment data, post hoc analysis showed a significant interaction between time point and treatment (P = 0.02), indicating a significant time-course change within the 6-week administration period with a maximum effect at 2 weeks (P < 0.001, Cohen’s d = −0.78; 95% CI, −1.21 to −0.35) compared with 4 weeks (P = 0.042, Cohen’s d = − 0.46; 95% CI, −0.90 to −0.01) and 6 weeks (P = 0.10, Cohen’s d = −0.35; 95% CI, −0.77 to 0.08). No significant effect of oxytocin was found on the mode of neutral facial expression or on the log-PDFmode of happy facial expression (P > 0.07). Figure 3 View largeDownload slide Effect of intranasal oxytocin on autistic facial expression and its time-course change in the confirmatory trial. Estimated marginal means (EM means) of the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (expression intensities) at each evaluation point are shown for the placebo- and oxytocin-administered groups individually. The bars represent 95% CIs of EM means. Evaluations were performed every 2 weeks during the 6-week oxytocin or placebo administration and 2 weeks after ceasing administration. The effect sizes between placebo and oxytocin administrations at each assessment points were also presented. Overall, the difference in the changes of the log-PDFmode of neutral expression intensities between oxytocin- and placebo-administered groups was significant (P < 0.001, Cohen’s d = −0.41; 95% CI, −0.62 to −0.20). d = Cohen’s d. Figure 3 View largeDownload slide Effect of intranasal oxytocin on autistic facial expression and its time-course change in the confirmatory trial. Estimated marginal means (EM means) of the change from baseline in the natural logarithm of maximum probability on the probability density function (log-PDFmode) of neutral expression intensity (expression intensities) at each evaluation point are shown for the placebo- and oxytocin-administered groups individually. The bars represent 95% CIs of EM means. Evaluations were performed every 2 weeks during the 6-week oxytocin or placebo administration and 2 weeks after ceasing administration. The effect sizes between placebo and oxytocin administrations at each assessment points were also presented. Overall, the difference in the changes of the log-PDFmode of neutral expression intensities between oxytocin- and placebo-administered groups was significant (P < 0.001, Cohen’s d = −0.41; 95% CI, −0.62 to −0.20). d = Cohen’s d. Individual differences in baseline anxious or depressive tendencies did not relate to oxytocin’s efficacy on the log-PDFmode for neutral facial expression (P > 0.24). The efficacy of oxytocin on the variability of neutral facial expression was preserved even after controlling baseline anxious or depressive tendencies (P < 0.001). Discussion The crucial finding from the current study is that repeated administration of oxytocin significantly improved the characteristic reduced variability of neutral facial expression in individuals with ASD. As the log-PDFmode for neutral faces is considered to reflect variation in facial expression (Owada et al., 2018), oxytocin consistently increased variation in the neutral facial expressions of ASD participants in both the exploratory and confirmatory trials. As reduced variability of facial expressions in individuals with ASD is a well-known clinical feature, the present results suggest that oxytocin improves at least one component of the autism-related characteristics of facial expression during social interaction, indicating partial improvement of deficits in social interaction and communication. Furthermore, the confirmatory trial detected a significant time-course change in the efficacy of oxytocin. Although efficacy decreased during the 6 weeks of repeated administration, it was still preserved at 2 weeks post-treatment. Although we reported inconsistent results about the efficacy of oxytocin on scores in ADOS recently, a gold standard diagnostic tool based on assessments of ASD core symptoms (Watanabe et al., 2015; Yamasue et al., 2018), the current findings demonstrated that 6 weeks repeated administration of oxytocin significantly enhanced objectively quantified facial expressions in individuals with ASD consistently in the exploratory and confirmatory trials. When the full ADOS reciprocity assessment was administered, the primary outcome of our confirmatory trial, the significant superiority of oxytocin to placebo, was not apparent, as reported in our recent study (Yamasue et al., 2018). It should be noted that this might indicate that facial expression analysis using only a short component of the full ADOS was superior to using the ADOS as a whole, possibly due to objectivity and quantitative analysis of facial expression. In the confirmatory trial, the enhanced ability to detect efficacy might be attributable to the objectivity rather than sensitivity of facial expression analysis, because the effect size of oxytocin on facial expression (Cohen’s d = 0.27) is not larger than that for ADOS [Cohen’s d = 0.47, reported in our recent report (Yamasue et al., 2018)]. However, the effect size of placebo on facial expression (Cohen’s d = 0.05) is below that of ADOS [Cohen’s d = 0.56 (Yamasue et al., 2018)]. The objectivity of facial expression analysis suggested that oxytocin may have significant efficacy for improving one component of the facial expression deficit in ASD, which is related to social deficits. This notion is consistent with our recent paper reporting the superiority of measurement of eye gaze fixation on socially relevant regions over ADOS reciprocity in the confirmatory trial (Yamasue et al., 2018). Facial expression analysis is based on only a few minutes of the ADOS (Owada et al., 2018), whereas the entire ADOS requires 40–60 min to administer (Lord et al., 1989). Therefore, facial expression analysis is easily repeatable in longitudinal assessments. The features of repeatability and objective quantification successfully demonstrate for the first time, to our knowledge, a time-course change in the efficacy of oxytocin. Together with previous human (Watanabe et al., 2015; Yamasue, 2016; Benner et al., 2018; Yamasue and Domes, 2018) and animal studies (Bales et al., 2013, 2014; Benner et al., 2018) suggesting a discrepancy between single and repeated administrations of oxytocin, the current findings suggest a time-course change of efficacy of oxytocin on ASD social symptoms. The currently observed deterioration of oxytocin’s efficacy might support a potential underlying mechanism, such as a negative feedback system of peptide hormones, upregulation or downregulation of the closely related peptide arginine vasopressin, or downregulation of oxytocin receptor (Insel et al., 1992; Yamamoto et al., 2004). Further research is needed to address the issue. The current results further support the preservation of oxytocin’s effect even 2 weeks after the cessation of treatment. To our knowledge, no previous study has reported such an effect (Keech et al., 2018). Although speculative, potential disengagement from the mechanism of deterioration of oxytocin’s efficacy might preserve or even increase the effects of neuropeptide after discontinuation. However, as the exploratory trial did not support the post-treatment persistence of the effects of oxytocin, future studies should investigate this issue further. The current study involved several potential limitations and methodological considerations that should be considered. First, the participants in our trials were all adult, male, Japanese individuals with ASD. Therefore, although the uniformity in participants’ demographic characteristics enhanced the ability to detect scientifically sound evidence, it should be noted that the current results may not be generalizable to other clinical or non-clinical populations. Second, since the outcome measure of current study was developed in our research team, the findings should be replicated by other research groups. The majority of the quantification methods were automatically conducted and independent of human labour (Owada et al., 2018), facilitating the replication of the current findings. Third, as some autistic characteristics in facial expression at baseline, such as a high mode of neutral facial expression and log-PDFmode of happy facial expression were not significantly improved by oxytocin treatment, we were unable to conclude that oxytocin can treat or recover all characteristics of facial expression in individuals with ASD. Furthermore, we did not show an association between the effects of oxytocin on the quantified facial expression and those on beneficial therapeutic outcomes, such as Clinical Global Impressions (CGI) (Guy, 1976) or Global Assessment of Functioning (GAF) (Aas, 2011). Although the effects of oxytocin on the variability of neutral facial expression exhibited weak correlations with the neural effect of oxytocin on anterior cingulate activity during a social judgment task (ρ = −0.56, P = 0.028) and on resting state functional connectivity between anterior cingulate and dorsomedial prefrontal cortices (ρ = −0.60, P = 0.019) assessed with functional MRI in the exploratory trial (detailed in Watanabe et al., 2015), the current results did not necessarily support the clinically beneficial effects of oxytocin. Fourth, the possibility of further improving the method of quantifying facial expressions, the task used to induce facial expressions in individuals with autism, and the outcome variables, should be considered. FaceReader and ADOS are not optimized for longitudinal facial expression analyses in individuals with ASD. It could be useful to validate these tools (and other available software) in this methodological context using electromyography, and/or action unit processing. The outcome variables used were determined based on the results of our previous case-control comparison in a limited sample size (Owada et al., 2018). The current quantitative facial expression analyses in our exploratory and confirmatory clinical trials successfully detected and verified the therapeutic effect of repeated administrations of intranasal oxytocin on autistic features in facial expressions during social interaction. Furthermore, for the first time, the present study demonstrated a time-course change in efficacy, with deterioration during the repetitive administration phase and preservation during the post-treatment phase. The current findings support further development of optimization of objective, quantitative, and repeatable outcome measures for autistic social deficits and to establish optimized regimen of oxytocin treatment for ASD. Abbreviations Abbreviations ADOS Autism Diagnostic Observation Schedule ASD autism spectrum disorder PDF probability density function Acknowledgements The study design was reviewed and approved by the institutional review board of The University of Tokyo Hospital. Neither the funder nor sponsor, the Strategic Research Program for Brain Sciences from Japan Agency for Medical Research and Development (JP18dm0107134), had any involvement in the data collection, analyses, writing, or interpretation of the study. However, the sponsor participated in discussion regarding which sites should be included in the trial and how to best interpret the results. The corresponding author had full access to the data and held the final responsibility regarding the decision to submit for publication. Wendy Brooks, PhD, and Benjamin Knight, MSc, from Edanz Group (www.edanzediting.com/ac) provided English language editing services for a draft of this manuscript. Funding This study was partially supported by the Strategic Research Program for Brain Sciences from Japan Agency for Medical Research and Development (JP18dm0107134). Competing interests The authors report no competing interests. References Aas IM . Guidelines for rating Global Assessment of Functioning (GAF) . Ann Gen Psychiatry 2011 ; 10 : 1 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat American Psychiatric Association. Diagnostic and statistical manual of mental disorders . 4th edn . Text Revision (DSM-IV-TR). Arlington, VA : American Psychiatric Association ; 2000 . WorldCat Anagnostou E , Soorya L , Chaplin W , Bartz J , Halpern D , Wasserman S , et al. Intranasal oxytocin versus placebo in the treatment of adults with autism spectrum disorders: a randomized controlled trial . Mol Autism 2012 ; 3 : 16 . Google Scholar Crossref Search ADS PubMed WorldCat Andari E , Duhamel J-R , Zalla T , Herbrecht E , Leboyer M , Sirigu A . Promoting social behavior with oxytocin in high-functioning autism spectrum disorders . Proc Natl Acad Sci USA 2010 ; 107 : 4389 – 94 . Google Scholar Crossref Search ADS PubMed WorldCat Aoki Y , Watanabe T , Abe O , Kuwabara H , Yahata N , Takano Y , et al. Oxytocin’s neurochemical effects in the medial prefrontal cortex underlie recovery of task-specific brain activity in autism: a randomized controlled trial . Mol Psychiatry 2015 ; 20 : 447 – 53 . Google Scholar Crossref Search ADS PubMed WorldCat Aoki Y , Yahata N , Watanabe T , Takano Y , Kawakubo Y , Kuwabara H , et al. Oxytocin improves behavioural and neural deficits in inferring others’ social emotions in autism . Brain 2014 ; 137 : 3073 – 86 . Google Scholar Crossref Search ADS PubMed WorldCat Baird G , Simonoff E , Pickles A , Chandler S , Loucas T , Meldrum D , et al. Prevalence of disorders of the autism spectrum in a population cohort of children in South Thames: The Special Needs and Autism Project (SNAP) . Lancet 2006 ; 368 : 210 – 15 . Google Scholar Crossref Search ADS PubMed WorldCat Bales KL , Perkeybile AM , Conley OG , et al. Chronic intranasal oxytocin causes long-term impairments in partner preference formation in male prairie voles . Biol Psychiatry 2013 ; 74 : 180 – 88 . Google Scholar Crossref Search ADS PubMed WorldCat Bales KL , Solomon M , Jacob S , Crawley JN , Silverman JL , Larke RH , et al. Long-term exposure to intranasal oxytocin in a mouse autism model . Transl Psychiatry 2014 ; 4 : e480 . Google Scholar Crossref Search ADS PubMed WorldCat Benner S , Aoki Y , Watanabe T , Endo N , Abe O , Kuroda M , et al. Neurochemical evidence for differential effects of acute and repeated oxytocin administration . Mol Psychiatry 2018 . doi: 10.1038/s41380–018–0249–4 . WorldCat Cohen AS , Morrison SC , Callaway DA . Computerized facial analysis for understanding constricted/blunted affect: initial feasibility, reliability, and validity data . Schizophr Res 2013 ; 148 : 111 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Domes G , Kumbier E , Heinrichs M , Herpertz SC . Oxytocin promotes facial emotion recognition and amygdala reactivity in adults with asperger syndrome . Neuropsychopharmacology 2014 ; 39 : 698 – 706 . Google Scholar Crossref Search ADS PubMed WorldCat Fujiwara T , Mizuki R , Miki T , Chemtob C . Association between facial expression and PTSD symptoms among young children exposed to the Great East Japan Earthquake: a pilot study . Front Psychol 2015 ; 6 : 1534 . Google Scholar Crossref Search ADS PubMed WorldCat Gordon I , Vander Wyk BC , Bennett RH , Cordeaux C , Lucas MV , Eilbott JA , et al. Oxytocin enhances brain function in children with autism . Proc Natl Acad Sci USA 2013 ; 110 : 20953 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Guastella AJ , Einfeld SL , Gray KM , Rinehart NJ , Tonge BJ , Lambert TJ , et al. Intranasal oxytocin improves emotion recognition for youth with autism spectrum disorders . Biol Psychiatry 2010 ; 67 : 692 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat Guastella AJ , Gray KM , Rinehart NJ , et al. The effects of a course of intranasal oxytocin on social behaviors in youth diagnosed with autism spectrum disorders: a randomized controlled trial . J Child Psychol Psychiatry 2015 ; 56 : 444 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat Guastella AJ , Hickie IB . Oxytocin treatment, circuitry, and autism: a critical review of the literature placing oxytocin into the autism context . Biol Psychiatry 2016 ; 79 : 234 – 42 . Google Scholar Crossref Search ADS PubMed WorldCat Guy W . Assessment manual for psychopharmacology . Rockville, MD : National Institute of Mental Health ; 1976 . Google Preview WorldCat Harris JC , Carter CS . Therapeutic interventions with oxytocin: current status and concerns . J Am Acad Child Adolesc Psychiatry 2013 ; 52 : 998 – 1000 . Google Scholar Crossref Search ADS PubMed WorldCat Hurlemann R , Scheele D . Dissecting the Role of Oxytocin in the Formation and Loss of Social Relationships . Biol Psychiatry 2016 ; 79 : 185 – 93 . Google Scholar Crossref Search ADS PubMed WorldCat Insel TR , Winslow JT , Witt DM . Homologous regulation of oxytocin receptors . Endocrinology 1992 ; 130 : 2602 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Keech B , Crowe S , Hocking DR . Intranasal oxytocin, social cognition and neurodevelopmental disorders: A meta-analysis . Psychoneuroendocrinology 2018 ; 87 : 9 – 19 . Google Scholar Crossref Search ADS PubMed WorldCat Kosaka H , Okamoto Y , Munesue T , Yamasue H , Inohara K , Fujioka T , et al. Oxytocin efficacy is modulated by dosage and oxytocin receptor genotype in young adults with high-functioning autism: a 24-week randomized clinical trial . Transl Psychiatry 2016 ; 6 : e872 . Google Scholar Crossref Search ADS PubMed WorldCat Lai M-C , Lombardo M V , Baron-Cohen S . Autism . Lancet 2014 ; 383 : 896 – 910 . Google Scholar Crossref Search ADS PubMed WorldCat Leng G , Ludwig M . Intranasal oxytocin: myths and delusions . Biol Psychiatry 2016 ; 79 : 243 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat Lewinski P , Den Uyl TM , Butler C . Automated facial coding: validation of basic emotions and FACS AUs in facereader . J Neurosci Psychol Econ 2014 ; 7 : 227 – 36 . Google Scholar Crossref Search ADS WorldCat Lord C , Rutter M , Goode S , Heemsbergen J , Jordan H , Mawhood L , et al. Autism diagnostic observation schedule: a standardized observation of communicative and social behavior . J Autism Dev Disord 1989 ; 19 : 185 – 212 . Google Scholar Crossref Search ADS PubMed WorldCat Lord C , Rutter M , Le Couteur A . Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders . J Autism Dev Disord 1994 ; 24 : 659 – 85 . Google Scholar Crossref Search ADS PubMed WorldCat Owada K , Kojima M , Yassin W , Lee MH , Guoynes CD , Downing GM , et al. Computer-analyzed facial expression as a surrogate marker for autism spectrum social core symptoms . PLoS One 2018 ; 13 : e0190442 . Google Scholar Crossref Search ADS PubMed WorldCat Radloff LS . The CES-D Scale: a self-report depression scale for research in the general population . Appl Psychol Measur 1977 ; 1 : 385 – 401 . Google Scholar Crossref Search ADS WorldCat Spielberger CD , Gorsuch RL , Lushene RE . Manual for the State-Trait Anxiety Inventory . Palo Alto, CA : Consulting Psychologists Press ; 1970 . Google Preview WorldCat Walum H , Waldman ID , Young LJ . Statistical and methodological considerations for the interpretation of intranasal oxytocin studies . Biol Psychiatry 2016 ; 79 : 251 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat Watanabe T , Abe O , Kuwabara H , Yahata N , Takano Y , Iwashiro N , et al. Mitigation of sociocommunicational deficits of autism through oxytocin-induced recovery of medial prefrontal activity: a randomized trial . JAMA Psychiatry 2014 ; 71 : 166 – 75 . Google Scholar Crossref Search ADS PubMed WorldCat Watanabe T , Kuroda M , Kuwabara H , Aoki Y , Iwashiro N , Tatsunobu N , et al. Clinical and neural effects of six-week administration of oxytocin on core symptoms of autism . Brain 2015 ; 138 : 3400 – 12 . Google Scholar Crossref Search ADS PubMed WorldCat Wechsler D . The psychometric tradition: developing the Wechsler adult intelligence scale . Contemp Educ Psychol 1981 ; 6 : 82 – 5 . Google Scholar Crossref Search ADS WorldCat Xu G , Strathearn L , Liu B , Bao W . Prevalence of autism spectrum disorder among US children and adolescents, 2014–2016 . JAMA 2018 ; 319 : 81 – 2 . Google Scholar Crossref Search ADS PubMed WorldCat Yamamoto Y , Cushing BS , Kramer KM , Epperson PD , Hoffman GE , Carter CS . Neonatal manipulations of oxytocin alter expression of oxytocin and vasopressin immunoreactive cells in the paraventricular nucleus of the hypothalamus in a gender specific manner . Neuroscience 2004 ; 125 : 947 – 55 . Google Scholar Crossref Search ADS PubMed WorldCat Yamasue H , Domes G . Oxytocin and autism spectrum disorders . Curr Top Behav Neurosci 2018 ; 35 : 449 – 65 . Google Scholar Crossref Search ADS PubMed WorldCat Yamasue H , Okada T , Munesue T , Kuroda M , Fujioka T , Uno Y , et al. Effect of intranasal oxytocin on the core social symptoms of autism spectrum disorder: a randomized clinical trial . Mol Psychiatry 2018 . doi: 10.1038/s41380–018–0097–2 . WorldCat Yamasue H . Promising evidence and remaining issues regarding the clinical application of oxytocin in autism spectrum disorders . Psychiatry Clin Neurosci 2016 ; 70 : 89 – 99 . Google Scholar Crossref Search ADS PubMed WorldCat Yatawara CJ , Einfeld SL , Hickie IB , Alvares GA , Tonge BJ , Hickie IB , et al. The effect of oxytocin nasal spray on social interaction deficits observed in young children with autism: a randomized clinical crossover trial . Mol Psychiatry 2016 ; 21 : 1225 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) (2019). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

BrainOxford University Press

Published: Jul 1, 2019

There are no references for this article.