Experimental design and participants
In the current study, we analyzed plasma samples collected from participants in our previous multi-center, parallel-group, placebo-controlled, double-blind, confirmatory trial of intranasal oxytocin in adult males with high-functioning ASD. The trial sites were the University of Tokyo Hospital, Nagoya University Hospital, Kanazawa University Hospital, and University of Fukui Hospital in Japan (UMIN000015264) (14). The details of this trial are described elsewhere (11, 14). Briefly, the inclusion criteria of this trial were as follows: (1) 18–54 years of age; (2) male; (3) diagnosis of autistic disorder, Asperger’s disorder, or pervasive developmental disorders not otherwise specified (PDD-NOS) based on DSM-IV-TR; (4) score exceeding the cut-off value (i.e. 10) for qualitative abnormalities in social reciprocity on autism diagnostic interview revised (ADIR) (15); and (5) full IQ above 80 and verbal IQ above 85 based on WAIS-Third Edition (WAIS-III) (16). The exclusion criteria were: (1) primary psychiatric diagnosis other than ASD; (2) instable comorbid mental disorders (e.g. instable mood or anxiety disorder); (3) changes in medication or doses of psychotropics within 1 month before randomization; (4) current medication with more than two psychotropics; (5) current pharmacological treatment for comorbid attention-deficit/hyperactivity disorder; (6) history of repeated administrations of oxytocin; (7) history of hyper-sensitivity to oxytocin; (8) history of traumatic brain injury with loss of consciousness for longer than 5 min or seizures; or (9) history of alcohol-related disorders, substance abuse, or addiction. Open to the public recruitment and the processes testing eligibility are explained in detail elsewhere (14).
A total of 106 men with high-functioning ASD were recruited between January 2015 and March 2016. Among these participants, 94 were psychotropic-free other than oxytocin during the all trial period, while 12 continued their medications with psychotropic during the trial period (four antidepressants, four antipsychotics, two mood stabilizers, and two hypnotics). The diagnosis for subtypes of participants with ASD were autistic disorder (N = 83), Asperger’s disorder (N = 12), and PDD-NOS (N = 11).
The participants received administrations of oxytocin (48 IU/day) or placebo in the morning and afternoon during 6 weeks (14). The placebo contained all of the inactive ingredients in order to control for any effect of substances other than oxytocin. On the last day of the 6-week administration period, data, including peripheral blood and clinical evaluations including autism diagnostic observation schedule (ADOS) (17), were collected from the participants. These endpoint clinical assessments were started 15 min after the last administration of intranasal oxytocin or placebo. All participants were sufficiently trained with identical instructions for intranasal administration, and the procedure of intranasal administration was evaluated at each 2-week assessment point. A self-report daily record was utilized to record treatment adherence.
Randomization and masking
Drug administration was randomly assigned the participants to the oxytocin or placebo group in a one-to-one ratio by the manager of randomization and masking based on a computer-generated randomized order. The randomization was stratified based on the trial site and median score of ADIR (<18 or ≥18, defined based on the results from our preliminary trial (18)). Spray bottles with the same visual appearance were utilized to store both active drug and placebo (Victoria Pharmacy, Zurich, Switzerland). The manager covered the labels of spray bottle to keep oxytocin or placebo blind to all the clinicians, assessors, their families, and participants. Registration, allocation, and data management procedures were defined separately (14).
The main outcome of the current study
The main outcome of the current study was metabolite concentrations in plasma samples collected at baseline, immediately before the first administration of oxytocin or placebo, and at endpoint, 60 min after the last administration of oxytocin or placebo at 6 weeks from baseline. Peripheral blood samples were collected from the participants while they were fasting (>3 hours without any meal and/or nutritious drink). The blood sampling procedure was conducted by experienced physicians. Plasmas were isolated with centrifugation at 1,600 g for 15 min at 4 °C. Then, the plasmas were stored at −80 °C until assay. 450 μL of methanol containing 10 mM each of methionine sulfone and 10-camphorsulfonic acid were added to the plasma samples (100 μL), and mixed well. Then, 500 μL chloroform and 200 μL of Milli-Q deionized water (EMD Millipore, Billerica, MA, USA) were added. The solution was centrifuged at 2,300 g for 5 min at 4 °C. Then, to remove proteins, a 400-μL aliquot of the supernatant was centrifugally filtered with a 5-kDa cutoff filter (Human Metabolome Technologies Inc., Tsuruoka, Japan). The filtrate was centrifugally concentrated in a vacuum evaporator and dissolved in 50 μL of Milli-Q water containing reference compounds before mass spectrometry analyses.
Plasma samples were measured using a capillary electrophoresis system with an Agilent 6210 time-of-flight mass spectrometer (CE-TOFMS, Agilent Technologies, Santa Clara, CA, USA) (19). A customized proprietary software (MathDAMP) was utilized to process raw data files acquired from CE-TOFMS (20). To identify target metabolites, their mass-to-charge ratio (m/z) values and migration times were matched with the annotation table of the metabolomics library (The Basic Scan metabolomics service of Human Metabolome Technologies Inc.) (21). The relative area was defined by dividing all peak areas with the area of the internal standard. The definition of relative areas allowed avoidance of mass-spectrometry detector sensitivity bias and injection-volume bias across multiple measurements and normalization of the signal intensities. Based on the peak area of internal controls of each metabolite, the absolute quantities of 110 pre-determined major metabolites can be measured with analysis by CE-TOFMS in our system. We used the absolute quantities obtained with CE-TOFMS as metabolite concentrations in plasma samples.
Other outcome measures of oxytocin efficacy
To examine their relationship to metabolite concentrations, we also included six additional outcomes found to be significant effects of oxytocin in this trial (11, 14) as well as in previous trials (11, 18). The six clinical and behavioral indices of oxytocin efficacy were as follows: (i) ADOS repetitive behavior = changes in the ADOS repetitive score between baseline and 6-week endpoint of oxytocin administration (endpoint − baseline). ADOS is a standard diagnosis tool for ASD but recently has been increasingly adopted as a primary outcome in ASD-related trials (14, 18, 22-26). (ii) Gaze fixation time on socially relevant regions = changes in the percentage of gaze fixation time on the eye region of a talking face presented on a video monitor, between baseline and 6-week endpoint (endpoint − baseline), which were measured with Gazefinder using the standardized and validated method described details in elsewhere (14, 27-29) (JVC KENWOOD Corporation, Yokohama, Japan). (iii, iv, v, & vi) log-PDFmode of neutral facial expression during 0-6, 0-2, 2-4, and 4-6 weeks = changes in the natural logarithm of the mode of the probability density function of neutral facial expression intensity during a semi-structured situation conducting social interaction in “Cartoons” an activity of ADOS module 4, quantified with a dedicated software program (30-32)(FaceReader version 6·1, Noldus Information Technology Inc., Wageningen, The Netherlands) in the validated method described details in elsewhere (11, 33). In addition to baseline and the 6-week endpoint, facial expression was assessed every 2 weeks as changes in log-PDFmode of neutral facial expression between each assessment point (i.e., (iii) 6 weeks –baseline, (iv) 2 weeks − baseline, (v) 4 weeks − 2 weeks, and (vi) 6 weeks − 4 weeks). The log-PDFmode for neutral facial expression is considered to reflect variation in facial expression (33).
Classification of participants according to time-course change in the efficacy of oxytocin
To investigate the mechanism of the time-course change in the efficacy of oxytocin repeated administration, we defined a subgroup from the oxytocin-administered group comprised of participants exhibiting a prominent time-course change. This classification was based on our previous findings on the time course of oxytocin-induced quantitative changes in facial expression in ASD which showed maximum efficacy at 2 weeks and deterioration of efficacy from 2 weeks to 6 weeks (11). Individuals showing reduction of log-PDFmode of neutral facial expression (i.e., improvement in ASD core symptom) from baseline to 2 weeks, and increase of log-PDFmode neutral facial expression (i.e., deterioration in ASD core symptom) from 2 weeks to 6 weeks, were classified as participants exhibiting a time-course change (Figure 2c).
Demographic and clinical information was compared using independent t-tests between placebo and oxytocin-administered groups and between the placebo-administered group and the oxytocin-administered group exhibiting the time-course change.
We analyzed the effects of oxytocin on metabolite concentrations using independent t-tests for comparing changes in metabolite concentrations during the 6-week administration period between the oxytocin-administered group and the placebo-administered group. Furthermore, because the change in metabolite levels over the 6-week oxytocin administration period could be associated with both clinical improvement and potential attenuation of oxytocin effectiveness, differences in changes in metabolite levels were also examined between the oxytocin-administered group displaying the time-course change in efficacy and the placebo-administered group. The independent t-tests were conducted for each metabolite, with absolute quantities successfully measured by CE-TOFMS measurement in at least 80% of all subjects (≧67 subjects) (34). The Benjamini-Hochberg false discovery rate (FDR) correction for the number of metabolites tested was applied, and FDR-corrected p-values of <0.05 were considered statistically significant.
For the oxytocin-administered group, we calculated Pearson’s correlation coefficients for 6-week changes in outcomes versus changes in metabolite concentrations (identified as significant differences between the oxytocin and placebo-administered participants). The outcomes used in the correlation analysis were 6-week change in ADOS repetitive behavior, 6-week change in gaze fixation time on socially relevant regions, and log-PDFmode of neutral facial expression change from baseline to 6 weeks. Furthermore, to clarify the relationships between the detected metabolite change and the time-course change in efficacy, changes in log-PDFmode of neutral facial expression between each assessment point (i.e., 2 weeks − baseline, 4 weeks − 2 weeks, and 6 weeks − 4 weeks) were calculated and correlated with changes in metabolites using Pearson’s correlation coefficient. The Benjamini-Hochberg FDR correction for the number of outcomes tested was applied to adjust the results, and the statistical significance level was defined as FDR-corrected p-values of <0.05. STATA version 14.0 and GraphPad Prism 8.4.1 were employed to conduct all statistical analyses.