Abstract
There are currently no approved drug interventions for social behavior dysfunction in autism spectrum disorder (ASD). Previous trials investigating effects of daily intranasal oxytocin treatment have reported inconsistent results and have not combined it with positive social interaction. However, In two preclinical studies we established that treatment every-other-day rather than daily is more efficacious in maintaining neural and behavioral effects by reducing receptor desensitization. We aimed to establish whether 6-weeks of intranasal oxytocin compared with placebo treatment, followed by a period of positive social interaction, would produce greater symptom improvements in children with ASD. A double-blind, randomized, cross over design trial was completed including 41 children with ASD aged 3-8 years. Primary outcomes were the Autism Diagnostic Observation Schedule-2 (ADOS-2) and social responsivity scale-2 (SRS-2). Secondary measures included other autism-related questionnaires and social attention assessed using two eye-tracking paradigms. A clinical reliable change index analysis revealed improvements in ADOS-2 total scores in 44% of children. Improvements in SRS-2 and behavioral adaptability scores were also found and correlated with increased basal saliva oxytocin concentrations. Additionally, oxytocin improved restrictive and repetitive behavior scores and increased time spent viewing dynamic social compared to geometric stimuli and the eye region of angry, happy and neutral expression faces. There were no adverse side-effects of oxytocin treatment. Overall, our results demonstrate that 6 weeks of intranasal oxytocin treatment administered every other day and followed by positive social interactions can improve clinical, eye-tracking and questionnaire-based assessments of symptoms in young autistic children.
Introduction
Autism spectrum disorder (ASD) is an early onset neurodevelopmental disorder with core symptoms including social communication and interaction problems and restricted and repetitive behavior patterns [1] and a prevalence worldwide of around 1% [2,3]. Currently there is no cure for ASD and no approved frontline medication for its core symptoms, especially in terms of social dysfunction, with early behavioral interventions currently being the most effective [2].
A promising therapeutic target to emerge over the last decade is the hypothalamic neuropeptide oxytocin (OXT). Single doses of intranasal OXT can enhance social cognition and motivation in both healthy and ASD adults as well as promoting extensive neural effects [4-8]. Furthermore, they can increase visual attention towards social relative to non-social stimuli, the eyes of emotional faces [9-11] and gaze-following [12] all of which are impaired in ASD [13,14]. Peripheral concentrations of OXT are lower in children with ASD compared to typically developing children [15], and polymorphisms of the OXT receptor (OXTR) are associated with ASD [16] as well as individual differences in autistic traits in both Caucasian and Asian cultures [17]. However, clinical trials using repeated daily doses of intranasal OXT over periods of 5 days up to 24 weeks have reported either modest beneficial effects on aspects of social responsivity [18-22] or on repetitive or adaptive behaviors [23-25] or no significant effects [26-30], although with no serious adverse effects. While a recent meta-analysis of 28 studies on single or repeated administration of OXT in autistic individuals concluded overall improvements in social functioning and with no evidence for publication bias it did not find evidence for significant advantages of chronic treatment [5]. A subsequent large trial conducted on 290 ASD children and adolescents aged 3 - 17 years and using an escalating daily dose strategy (from 8-80IU) over 24 weeks found no significant improvements [30]. Thus, despite numerous studies having reported positive neural and behavioral effects of single doses of intranasal OXT there is no clear evidence that repeated dosing produces more substantial improvements in social functioning. Indeed, many trials have now found no beneficial effects at all [26-30]. Indeed, only one study to date has reported that chronic but not acute dosing produces neurochemical changes in frontal cortex [31].
All previous trials investigating therapeutic effects of chronic intranasal OXT administration have adopted a once- or twice-daily dosing frequency with daily doses ranging from 8-80IU and with the general objective of raising basal concentrations of the peptide. However, there is no empirical support for this being an optimal chronic treatment strategy. It is well established that G protein-coupled receptors exhibit rapid desensitization, internalization and subsequent down-regulation following exogenous treatment with agonist ligands [32], including the OXTR [33-36]. Indeed, chronic/daily OXT treatment leads to extensive OXTR and vasopressin-1A receptor down regulation in the rat forebrain [34-36] and may even cause DNA damage in the hippocampus [37]. Additionally, when OXT binds to its receptor it can recruit different intracellular G protein-coupled pathways resulting in opposite effects on neuronal activity in a concentration-dependent manner, with increasing OXT bioavailability shifting coupling away from the excitatory Gq-protein to the inhibitory Gi-one [38-40]. This may explain evidence for inverted U-shaped dose-response curves reported for acute OXT effects in humans with higher doses having the opposite effects of lower ones [41,42]. Animal model studies demonstrated that chronic dosing with OXT can not only lead to receptor downregulation in the forebrain [34-36], but can also impair rather than enhance social behavior [43-45] and increase rather than decrease anxiety [46]. In humans, we have established in two preclinical trials that once daily treatment with intranasal OXT for 5 days leads to receptor desensitization leading to reduced effects on neural responses to emotional faces and anxiolytic responses to threatening stimuli. On the other hand, both neural and behavioral effects are maintained or enhanced by using a less frequent administration every other day to allow receptors more time to recover from desensitization [47,48]. Thus OXT given both at too great a frequency and/or too high a concentration can reduce or even reversing its potentially beneficial functional effects. In the current trial, we therefore incorporated a novel step of using an every-other-day rather than a daily dose administration strategy.
Simply increasing OXT concentrations alone in autistic individuals using exogenous supplements in order to render them equivalent to those in typically developing ones [15] may also not be the optimal strategy for promoting improved social behavior. The majority of studies using acute administration of intranasal OXT have demonstrated positive effects in subsequent tasks involving social attention and motivation, and in a number of previous studies we have reported that it can facilitate learning with positive social feedback [49-51]. A prevailing hypothesis concerning the functional effects of OXT is that it enhances the salience of social stimuli in whatever context it is administered [52]. As such, it is important for optimizing OXT’s potential therapeutic benefit that its administration occurs immediately prior to a positive social experience. The current trial therefore included a second novel step whereby caregivers were required to engage in positive social interactions with their children in the period immediately following each intranasal administration.
In this trial we therefore administered a standard 24IU dose of intranasal OXT every other day for 6 weeks in young children (3 - 8 years old) diagnosed with ASD in a placebo (PLC) controlled cross-over design where subjects receive both OXT and PLC treatments in a randomized order. Additionally, caregivers engaged in positive social interactions with their children following administration of each intranasal treatment. In view of a previous trial reporting some evidence for expectation (placebo) effects [22], we incorporated an additional component to assess this possibility whereby caregivers were informed that each treatment phase would last 8 weeks, although in fact the first 2 weeks were always PLC. Some caregiver-completed and eye-tracking outcome measures were taken at the beginning and end of this 2-week PLC lead-in to determine if significant improvements were reported. We chose to focus on young children with ASD given the general consensus that early interventions tend to lead to more positive outcomes [2,53,54].
In terms of outcome measures we chose both a gold-standard objective clinical measure (Autistic Diagnostic Observation Schedule – 2; ADOS-2 Comparison and Total scores) [55] and the caregiver-based social responsivity scale-2 (SRS-2 total score) [56,57] as primary outcomes for therapeutic efficacy. Additionally, we included a number of secondary outcome measures including, for the first time, eye-tracking paradigms as objective measures of improved attention towards social stimuli and which are becoming increasingly established as sensitive measures of social dysfunction in autism [13,14,58,59]. Other autism-related questionnaires completed by caregivers, including an assessment of behavioral adaptability (Adaptive behavior assessment system-II – ABAS-II) [60], repetitive behavior scale revised (RBS-R)[61] and social communication questionaire (SCQ) [62]. All the chosen caregiver completed assessments have good validity in terms of reliably distinguishing between children with ASD and typically developing children, are sensitive to severity of clinical symptoms, having good internal consistency and have been used in previous clinical trials. As such they adhere to the recent general recommendations for using clinimetric criteria for patient-reported outcome measures (CLIPRO) in clinical trials although since the subjects were young children their caretakers were completing the assessments or clinicians [63]. Additionally, as a measure of caregiver well-being during the trial, we incuded the caregiver strain questionaire (CSQ) [64]. Given the likelihood of some heterogeneity in responses we further investigated whether any improvements in symptoms were associated with basal or altered peripheral concentrations of OXT, age, autism subtype (using the Beijing Autism Subtype Questionnaire – BASQ) [65] or OXT receptor genotype (5 receptor genotype single nucleotide polymorphisms SNPs, either associated with clinical ASD or autistic traits) [16,17]. Finally, we also included a range of safety measures to monitor any adverse effects.
We hypothesized that the novel infrequent dose regime of intranasal OXT administration for 6 weeks in conjunction with positive social interactions following each dose would produce more reliable improvements in both objective and subjective outcome measures of autistic symptoms.
Materials and Methods
Trial design and participants
A total of 46 eligible autistic children were enrolled in the trial to permit a 1:1 computer generated randomization strategy for two groups receiving different treatment orders in a cross-over design. An a priori power analysis revealed that 46 participants would achieve > 90% power for a medium effect size of 0.5. The study used a computer randomized, double-blind, placebo-controlled, cross-over design with participants randomly allocated to two study arms, one receiving intranasal OXT first followed by intranasal PLC second and vice-versa for the other. This design allowed responses to OXT and PLC to be compared within each participant thereby reducing the influence of individual variability and potential confounding variables. Additionally, in both arms of the study participants received a 2-week initial period of PLC intranasal sprays prior to their 6-week treatment period with OXT or continued PLC. This additional control was administered blind to participants/caretakers and experimenters and was included to allow an assessment of possible expectation (placebo) effects reported in a previous cross-over trial with caregiver completed measures [22]. The study was conducted in the Chengdu Maternal and Children’s Central hospital (CMCCH) between June 2019 and July 2021 in accordance with the guidelines for Good Clinical Practice and the principles of the Declaration of Helsinki and was reviewed and approved by the Ethics Committee of CMCCH affiliated to the University of Electronic Science and Technology of China (number 201983) as well as the general ethics committee of the University (number 1420190601). Parents or legal guardians of all participants provided written informed consent and the trial was pre-registered (Chinese Clinical Trial Registry: ChiCTR1900023774 in Chinese and in English).
Eligible children were recruited from outpatient clinics at CMCCH and from other children’s hospitals in Chengdu and Sichuan Province via specialist networks as well as in response to the trial registration. Participants were considered eligible if they were diagnosed with ASD according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) (1); aged 3–8 years and their diagnosis was confirmed with the Autism Diagnostic Observation Schedule - 2 (ADOS-2) [55] administered by two certificated research reliable authors who were blind to the treatment received (JL and JK). None of the participants were receiving any psychotropic medications during the course of the trial. We excluded individuals with any identified genetic or chromosomal abnormalities or diagnosed with a neurological disease (e.g., epilepsy, Rett syndrome), or psychiatric disorder other than ASD or had severe respiratory, hearing or visual impairments. Out of 85 children screened, 54 were considered eligible and underwent ADOS-2 assessment although the caretakers of 8 children declined to participate leaving a total of 46 (see Fig. 1).
CONSORT flow diagram for trial
Assessment and medication schedules
For the trial protocol see Figure 2. Following informed consent by a caregiver, eligible children visited CHCCM in week 1 for all clinical (ADOS-2 modules 1 or 2), behavioral (SRS-2, RBS-R, SCQ, ABAS-II, CSQ) and completed the two eye-tracking paradigms (dynamic social versus non-social [58] and emotional faces) before the caregiver administered the first nasal spray to their child under supervision. Caregivers were then given their first two bottles of PLC nasal spray (labelled week 1 and week 2) to administer in the early morning every other day for the first two weeks. Importantly, caregivers were additionally instructed to always engage their child in a positive social interaction and play session in the first hour or so after intranasal administration and encouraged to comply with this by being informed that the treatment was expected to have better effects if they did so. Caregivers then returned to CHCCM with their children in accordance with the protocol schedule shown in Fig. 2 and were given intranasal treatment bottles on each occasion and returned used ones for weighing. Finally, all caregivers were contacted for a 6-month follow-up and asked to complete the ABAS-II, SRS-2, SCQ, RBS-R and CSQ questionnaires.
Plan of the trial. The schematic shows the organization of the two phases of the cross-over trial and intranasal treatments. Subjects received either oxytocin (OXT) or placebo (PLC) in each phase for 6 weeks and received PLC in the 2 weeks prior to this. There was a 2 week wash out period between phases. Outcome measures were taken at 8 different time-points during the trial (T1 – T8) and at a 6-month follow-up. Outcome measures taken at each point are shown. Autism Diagnostic Observation Schedule – 2 (ADOS-2); Social Responsivity Scale-2 (SRS-2); Adaptive behavior assessment system-II (ABAS-II); Social communication quotient (SCQ); Repetitive behavior scale - revised (RBS-R); caregiver strain questionnaire (CSQ); Eye-tracking, Tasks 1 and 2. Points where blood or saliva samples were taken are also indicated.
Participants were randomly assigned to receive 24IU intranasal OXT or PLC treatments in the first phase of the trial and then the other treatment during the second phase. Treatments were administered by caregivers every other day following initial training and with children and caregivers, investigators, psychiatrists and data analysts blind to the treatment allocation until the study database was unlocked. Both intranasal OXT and PLC sprays were supplied sterile by a pharmaceutical company (Sichuan Meike Pharmaceutical Co, Sichuan, China) and contained the same components other than OXT (i.e. glycerol, sodium chloride and water). Intranasal spray bottles were labelled and distributed to caretakers by an individual not involved in any other aspect of the trial and who was responsible for finally unmasking the treatment details at the end of the trial. Intranasal spay bottles for OXT or PLC were identical in appearance and labelling with each having a unique code. To help monitor compliance, all old bottles were returned for weighing which confirmed appropriate use. The volumes consumed in the spray bottles did not differ significantly during the 6-week OXT (mean ± SD = 19.0 ± 3.6 ml) and PLC (18.9 ± 3.7 ml, t = 0.013, p = 0.897) treatment phases and were in line with the expected volume (21 doses during 6 weeks (i.e. 6 sprays each = 0.6 ml x 21 = 12.6 ml) + 1-3 priming sprays 0.1-0.3 ml each time; total volume = 14.7-18.9 ml).
Primary outcome measures
The primary outcome measures were a clinical assessment using ADOS-2 scores (the Comparison and Total scores – ADOS-2 assessments were administered blind by two certificated research reliable authors – JL and JK) and the SRS-2 questionnaire completed by caregivers in line with previous trials [19,22,30]. The comparison ADOS-2 score was used in addition to the total score due to the fact that both modules 1 and 2 were used and the comparison score takes account of this as well as the age of the child [55]. However, individual participants were always assessed using the same ADOS-2 module in each of the two treatment phases.
Secondary outcome measures
Secondary outcome measures included a number of autism-related questionnaires completed by caregivers at various time points during the trial (see Fig. 2) and included ABAS-II (either for children below 6 years old or from 6-18 years old) [60], SCQ, RBS-R and CSQ questionnaires. Additionally, all children completed two eye-tracking tasks on 8 occasions (at the beginning and end of each treatment phase in the study and additionally after the initial 2-week PLC phase and then in the middle of each treatment phase – i.e. 3 weeks after the start of a treatment). Eye movement data were recorded using an eye tracker (Tobii TX300, Tobii, Danderyd, Sweden). All gaze data were recorded at 300 Hz sampling rate with a gaze accuracy of 0.4°. Recording and stimulus presentation were conducted using Tobii Pro Studio, E-prime 2.0 software, and E-Prime Extensions for Tobii (Psychology Software Tools, Pittsburgh, PA). All raw data with at least 50% gaze weight were analyzed using Tobii studio. The primary measure collected in the current study was total fixation duration with the proportion of time spent viewing the social (dancing children) stimulus in the first paradigm and the proportion of time spent viewing the eye, nose or mouth regions of the different emotional faces in the second paradigm. In both eye tracking tasks trials were excluded from analysis where children spent < 5% of the time viewing stimuli as in [58] (for Task 1 no trials were deleted and in Task 2, 1.6% of trials were deleted).
All children were either seated alone or on a caregiver’s lap in front of a display screen on which all stimuli were presented. Initially they were required to pass a five-point calibration using dynamic animations before being presented with the two experimental paradigms. Children were told to view the screen freely and caregivers instructed not to give any instructions to them about what to watch or interrupt them during the presentation period. Eye tracking measures were taken at all test time-points during the trial (see Fig. 2). In the first Social Attention Task (Task 1) dynamic dancing Chinese children versus dynamic geometric patterns were displayed side by side similar to previous publications [58,59]. In the static emotional face task (Task 2) different individual emotional (angry, fear, happy and neutral) faces of either adults or children were presented. The whole eye tracking task lasted 114 seconds. Task 1 was similar to our previous study comparing autistic and typically children [58] and designed to measure differential attention to pairs of dynamic social (dancing individuals) and non-social (geometric patterns) images (in color; 545 × 430-pixel resolution). The dynamic social stimuli involved one, two or three children dancing, while the geometric patterns involved different moving shapes. The total presentation included 20 × 2s video clips prepared as two continuous videos. The stimuli (social vs. geometric) were presented simultaneously with one category on the left and the other on the right in one video and vice versa in the other, with the order being counter balanced across sessions. There was a 1500ms blank interval between the 2 videos. To reduce familiarity effects, four different sets of stimuli were presented once during each treatment phase although in a different order for each child. The static emotional face task (Task 2) contained 4 children’s faces (boys vs. girls) and 2 adult faces (female vs. male) with angry, happy, fear and neutral expressions (in color and with a 680 × 845-pixel resolution and white background). Each face was presented for 2 seconds and followed by a 500ms interval where a black cross was displayed in the middle of a light grey background screen. In total, the face task lasted 72 seconds and the order of presented faces for each subject was random. As in Task 1, four different sets of faces were presented to subjects at different time points within each treatment phase. In total, there were 16 children’s faces and 8 adult faces used (i.e. 6 faces for each of the 4 emotions).
Blood (6 ml – collected into EDTA tubes) and saliva sample (1 ml using salivettes, passive drool method) [66] were taken from each participant on the first day of the trial for measurement of basal OXT concentrations. Subsequently, saliva samples were taken again at the end of the first treatment phase of the trial and before and after the second phase of the trial (see Fig. 2). Samples were taken 24-48h after the end of treatment to avoid contamination by the final intranasal dose of OXT. Samples were collected onto ice and subsequently both plasma (after centrifugation) and saliva were stored at -80 □C until assayed for OXT within 6 months. Oxytocin concentrations were analyzed in duplicate with a widely used enzyme linked immunoassay (ELISA - ENZO Life Sciences, USA) [47,66]. A standard prior extraction step was performed in accordance with the manufacturers recommended protocol. Detection sensitivity was 3pg/ml, inter- and intra-assay variation was < 9%.
Safety and other caretaker assessed behavioral observations
Caregivers recorded adverse events daily during treatment phases using an 18-item checklist (see Table 3 for items) and could also discuss any issues in person with medical staff or investigators when they visited for assessments and to receive/return intranasal treatment bottles (i.e. every 2-3 weeks during the trial). Caregivers were also asked to complete a daily diary noting any specific examples of positive or negative behavioral changes.
Autism sub-type and oxytocin receptor genotype
On their first visit to take part in the trial the autism social sub-type of children was assessed using the BASQ (aloof, passive or active but odd) [64]. Buccal cell samples were also taken for subsequent analysis of OXT receptor genotype [17]. Genotyping of 5 OXTR SNPs (rs53576, rs2254298, rs2268491, rs2268498, rs237887) was conducted on a Cobas Z 480 Light Cycler (Roche Diagnostics, Mannheim, Germany). A prior DNA extraction step was performed on a MagNA Pure 96 robot (Roche Diagnostics, Mannheim, Germany) and using probes designed by TIBMolBiol (Berlin, Germany).
Statistical analysis
All statistical analyses were performed using SPSS version 22.0 (SPSS Inc., Chicago, IL, USA). Given the within-subjects design, measures within each treatment phase were expressed as differences from baseline (i.e. a difference-in-difference approach) and treatment effects tested using paired t-tests (or Wilcoxon when normal distribution was not achieved) and effect sizes calculated using Cohen’s d for repeated measures. Correlations between measures were assessed with Pearson tests. Bonferroni corrections were applied for multiple secondary outcome questionnaire measures and for sub-scales. For the face emotion eye-tracking assessment repeated measures ANOVAs were performed with treatment, face region and face emotion as factors. Missing data was only imputed (using average differences between values of other subjects) for saliva OXT concentrations under PLC where 3 subjects had one missing sample.
The wash-out period between the two treatment phases was 4 weeks (2 weeks with no treatment and 2-weeks with PLC treatment), similar to a previous study [21]. To assess the effectiveness of the wash out period baseline measures were compared before each treatment phase using paired t-tests. Treatment effects (differences) were also calculated separately for the two cohorts receiving the different orders of treatments. The difference between measures at the end of phase 1 and beginning of phase 2 was also analyzed to determine if participants receiving OXT first showed a significant post-treatment return to baseline.
To assess the potential long-term impact of the intervention, caregivers were asked to complete some of the questionnaire-based measures 6 months after the end of the treatment phases of the trial. For those who complied paired t-tests were then used to compare scores with those at the end of the OXT and PLC phases of the trial.
An additional more robust clinical assessment of improvement was performed on primary outcome and eye-tracking measures by calculating a Reliable Change Index (RCI) [67] which indicates whether the change in an individual’s outcome measure score from the pre-to post-intervention assessment is significantly greater than a difference that could have occurred due to random measurement error alone.
Daily frequencies of urination and defecation were compared between treatments using Wilcoxon tests. Numbers of participants exhibiting non-serious adverse events in the OXT and PLC phases of the trial were compared statistically using chi-square tests for paired categorical data. In all cases p < 0.05 two-tailed was considered significant.
Results
Participants, baselines and compliance
Two subjects in the PLC treatment first group were excluded, one due to a severe cold and the other unable to travel to the clinical site due to COVID restrictions. For the OXT treatment first group one subject was excluded due to a severe cold and another due to a family issue unrelated to the trial. Thus, 41 participants receiving both OXT and PLC treatments (21 in the OXT-first arm of the trial and 20 in the PLC-first)(see Fig. 1). Demographic details and baseline measures for primary and secondary outcome measures and plasma and saliva OXT concentrations for the two treatment order groups are provided in Table 1 and there were no significant differences between them. Baselines for SRS-2; RBS-R, SCQ; ABAS-II and CSQ taken both immediately before and at 2 weeks into each treatment phase (where only PLC nasal spray was administered) assessed if there were any significant expectation effects on caregiver completed measures (i.e. placebo effect). However, comparisons of the two baselines in both treatment phases revealed no significant differences (for PLC followed by OXT all ps > 0.11 and OXT followed by PLC all ps > 0.12 see Fig. 3B for SRS-2 scores over time) and therefore having demonstrated no evidence for a placebo effect the two baselines were combined as an overall mean for each phase in all subsequent analyses.
Changes in primary outcome measures over time and in the reliable change index (RCI). Mean ± standard deviation scores for (A) Autism Diagnostic Observation Schedule – 2 (ADOS-2) total and (B) Social Responsivity Scale-2 (SRS-2) total at different time points during the trial for the subjects receiving oxytocin (OXT) first (n = 21) or placebo (PLC) first (n = 20). The SRS-2 was measured on two occasions at the beginning of each phase (T1 and T2 and T5 and T6) when all subjects were receiving PLC. No treatment occurred between time points 4 and 5 (wash out). * p < 0.05 or ** p < 0.01 paired t-test difference between 2 time points. (C) RCI analysis showing changes in ADOS-2 and SRS-2 total scores pre-versus post-PLC) or OXT treatment. The number of children showing reliable change (on or below lower dotted line) or deterioration (on or above upper dotted line) are given (see Table S2 for full details). Note: where two subjects have identical scores the size of the data dot is increased to indicate this.
Analysis of primary outcome data
Statistical analysis of difference scores revealed that OXT-treatment produced significantly greater improvements compared with PLC in all primary outcome measures (ADOS-2 comparison score, p = 0.005; ADOS-2 total score, p = 0.0003 and SRS-2 Total score, p = 0.0005 – all medium effect sizes, see Table 2). An exploratory sub-scale analysis revealed that difference scores on ADOS-2 subscales were significant for social affect (p = 0.0136) but not restrictive and repetitive behavior, p = 0.0625) and for the SRS-2 subscales difference scores for social awareness, cognition, communication and motivation were all significant (all ps < 0.022) but not for the restrictive and repetitive behavior subscale (p = 0.1042) (Table 2). An analysis of effects separately for the groups receiving the two different treatment orders revealed similar results of OXT treatment although generally the greatest changes were in the PLC-first group (see Table S1). Figures 3A and B show details of ADOS-2 total and SRS-2 total scores at different time points in the two groups receiving different treatment orders.
An RCI analysis showed that 44% (ADOS-2 total) and 27% (SRS-2 total) of individuals showed reliable improvement under OXT and none showed deterioration. Under PLC, 15% (ADOS-2 total) and 0% (SRS-2 total) showed reliable improvement and 10% (ADOS-2 total) and 5% (SRS-2 total) showed deterioration (see Fig. 3C and Table S2).
Analysis of secondary outcomes
Table 2 shows treatment outcome results for secondary outcome questionnaires and also changes in baseline saliva OXT concentrations. Comparisons of difference scores revealed that OXT relative to PLC treatment increased ABAS-II global adaptive composite (GAC) (p = 0.012) scores and decreased those for RBS-R (p = 0.0331) although the latter would not survive multiple correction for the number of different questionnaires (n = 4, p < 0.0125). There was a significantly greater proportionate increase in basal saliva OXT concentrations following OXT compared with PLC treatment, the post-treatment sample being taken 24-48h after the final intranasal administration and thus unlikely to have been influenced by the OXT in the final intranasal spray (p < 0.0001 – Table 2).
Eye-tracking assessments of the effects of OXT on social attention
For the eye-tracking paradigms post-treatment tests showed that OXT compared with PLC administration significantly increased the difference score for the proportion of time spent viewing dynamic social (dancing children) compared to geometric patterns in Task 1 (p = 0.006 see Fig. 4A and Table S3). At the interim 3-week assessment point after the start of OXT treatment the difference score was not significant (p = 0.233, n = 36, see Fig. 4A and Table S4). Note that 5 children were unable to attend this interim assessment due to COVID travel restrictions and so had no data. For the Task 2 face emotion paradigm a repeated measures ANOVA revealed a significant treatment x face region x face emotion interaction (F(6,228) = 4.113, p = 0.001, partial η2 = 0.098; n = 39). Post-hoc Bonferroni corrected tests revealed that OXT significantly increased the proportion of time viewing the eye region of all faces with increased time spent viewing angry (p <0.001), happy (p = 0.020) and neutral (p = 0.032) expression faces but decreased time spent looking at the eyes of fearful (p = 0.008) ones (see Fig. 4B and Table S3). Similar to Task 1, at the interim point test after 3 weeks of OXT treatment an ANOVA revealed no significant main or interaction effects involving treatment (all ps > 0.434, n = 34) (see Fig. 4B and Table S4). There were no treatment effects on the absolute amount of time spent viewing stimuli in either task (Task 1 mean ± SD = 31.03 ± 3.91s and 30.48 ± 5.66s out of 40s in PLC and OXT treatment phases respectively, p = 0.504 t-test; Task 2 mean ± SD = 5.71 ± 2.02s and 5.45 ± 2.77s out of 12s across the four face emotions in the PLC and OXT treatment phases respectively; p = 0.4416. There were no effects of OXT on total time viewing specific face emotions; all ps > 0.42).
Treatment effects on two eye-tracking paradigms (A) Task 1 - mean ± SD treatment (oxytocin – OXT and placebo PLC) differences in % total duration viewing the dynamic social compared to the dynamic geometric stimuli and (B) Task 2 the eye region of all, angry, happy, fearful and neutral faces. Differences are shown for both during treatment (after 3-weeks) and post-treatment (after 6-weeks). **p<0.01. *p<0.05 post-hoc Bonferroni corrected tests for OXT vs PLC.
An RCI analysis revealed that 22-24% of participants showed reliable improvements and 0-5% deterioration under OXT compared to 5-15% reliable improvement and 10-22% deterioration under PLC (see Table S2).
Analysis of the effectiveness of the wash-out period between treatment phases
While a cross-over design has the advantage of reducing individual variability through a within-subject analysis it also has the potential drawback of carry-over effects for individuals treated first with the experimental intervention prior to PLC. A comparison of treatment effects for the groups receiving OXT or PLC first suggests that improvements were generally greater in the latter (see Table S2). To further assess the efficacy of the wash-out period between treatment phases we also compared both the baselines at the beginning of each treatment phase and measures at the end of phase 1 compared to the beginning of phase 2. Results showed that baseline ADOS-2 total scores were significantly lower at the beginning of phase 2 compared with the beginning of phase 1 in the group receiving OXT first (p = 0.041, see Table S5) and also in this group overall improvements following OXT had not reduced significantly by the beginning of the PLC phase (see Table S6 and Figs. 3A and B). Thus, there is some evidence that the wash-out period was insufficient for measures to return to baseline for the PLC phase following the OXT phase, which may have acted to reduce the magnitude of difference scores between OXT and PLC in this sub-group.
Potential factors influencing oxytocin-induced improvements
There were no significant associations between age and OXT-induced improvements (difference of difference between OXT and PLC) in primary outcome measures (ADOS-2 comparison score, r = 0.12, p = 0.455; ADOS-2 Total score, r = -0.07, p = 0.664; SRS-2 total score, r = 0.26, p = 0.10). There were too few female participants (n = 3) to perform a meaningful comparison with males.
Proportionate increases in saliva OXT concentrations after the OXT treatment phase were significantly correlated with the improvement of both total SRS-2 (r = 0.401, p = 0.0126) and ABAS-II GAC (r = 0.604, p < 0.001) scores (see Figs. 5A and B) but not ADOS-2 scores (ps > 0.2). The magnitude of the change in viewing the eyes of fearful faces was also correlated with change in basal OXT concentrations (r = -0.327, p = 0.045, Fig. 5C). There were no significant correlations between initial baseline OXT concentrations in plasma (all ps > 0.397) or saliva (all ps > 0.293) and treatment outcomes (difference OT-PLC).
Correlations between increased saliva oxytocin (OXT) concentrations and improved outcome measures. Regression graphs plot improvements in (A) Social responsivity scale (SRS-2) total scores and (B) Adaptive behavior assessment system-II, Global Adaptive Composite (ABAS-II GAC) scores positively associated with % change in basal saliva OXT concentrations at the end of the OXT treatment phase together with r and p-values (Pearson) (C) shows the same but for a negative association between time spent looking at the eyes of fearful faces and % change in basal OXT.
Improvements in primary outcome measures were similar across the different autism sub-types measured using the BASQ (i.e. aloof, passive and active but odd - see Table S7). For the eye-tracking paradigms the aloof sub-type showed a greater increase in the proportion of time spent viewing the eyes of angry faces and the reduction in time spent viewing the eyes of fear faces was significantly stronger in both aloof and passive sub-types (see Table S8). Of the 5 different OXTR SNPS analyzed only rs2268491 showed some evidence for genotype-dependent effects of OXT with greater SRS-2 improvements and OXT concentration changes in CC-relative to T+-carriers (see Table S9).
Caregiver feedback on positive and negative treatment effects on social behavior
Although anecdotal, and not a planned outcome measure, caregivers completed a weekly diary of observed positive and negative social behavior changes during the OXT and PLC treatment phases. Caregivers of 17/41 children reported incidences (1-6 occasions) of improved social behaviors during OXT treatment compared with 7/41 (1-2 occasions) during PLC (chi-square = 4.167, p = 0.041). Under OXT, caregivers mostly reported examples of improved social communication (n = 8) and playing with others (n = 6), although other examples included increased eye-contact/attention/smiling towards others, imitation, sharing and social hugging. In terms of examples of unusual negative anti-social behavior (other than irascibility and aggression reported in Table 3) only 3 caregivers reported this during PLC and 1 during OXT, each on a single occasion.
Effects of intranasal oxytocin on safety and tolerability
No severe adverse effects were reported under either OXT or PLC treatments, although one participant withdrew during the PLC treatment phase due to being unable to tolerate the nasal spray (see Fig. 1). For non-serious adverse events there were no significant differences between proportions of participants experiencing physical or behavioral symptoms during OXT compared to PLC treatment. In line with findings from a previous trial [22] we did observe a small increase in the frequency of daily urination under OXT compared with PLC (see Table 3).
6-month follow up
Caregivers of 27 children completed SRS-2, SCQ, RBS-R and CSQ and 30 completed the ABAS-II questionnaires remotely 6 months after treatment phases of the trial to assess any long-term consequences of OXT treatment. SRS-2 scores were not significantly different from those recorded at the end of the OXT treatment phase of the trial but were still improved compared to the PLC phase (p = 0.002 see Table S10) suggesting the possibility of some long term positive effect of OXT.
Discussion/Conclusion
Overall, the findings of this cross-over trial demonstrate that intranasal OXT treatment can significantly improve symptoms in young children with autism when given infrequently as opposed to every day and followed by a period of positive social interaction. Importantly, the current trial not only achieved greater improvements in the same subjective measure (SRS-2) as reported in two previous studies in young children using daily treatment [19,22], but additionally for the first time found a significant reduction in an objective gold standard clinical measure, ADOS-2 scores (both the comparison and total scores). Furthermore, the trial provided the first evidence for improved social functioning using two eye-tracking paradigms. In agreement with previous trials on children [19,22,30], no serious adverse effects were found.
No previous PLC-controlled study in children has shown ADOS-2 changes in response to intranasal OXT [19,22] or to other pharmacological interventions such as the related peptide vasopressin [68] or bumetanide [69]. While the significant improvement in ADOS-2 scores in the current trial was relatively small the RCI analysis indicated that it was reliable in 44% of children and none showed deterioration. Similarly, while previous trials of intranasal OXT and vasopressin reported around a 6 to 8-point reduction in SRS-2 scores relative to PLC we found a 12-point reduction. An exploratory analysis of ADOS-2 and SRS-2 subscales support the conclusion that OXT mainly improved social symptoms although there was some indication of a reduction in restrictive and repetitive behaviors, particularly in the RBS-R.
The two key innovations in the current trial were firstly to reduce the frequency of OXT intranasal doses to every other day compared to the once or twice daily used in all previous studies and that caregivers were instructed to engage in positive social interaction/play with their child in the period immediately following intranasal administration. These innovations in combination may have contributed to greater improvements in autistic symptoms even though the average dose given every two days was reduced to between a quarter or a half of that in previous studies [19,22]. Indeed, the lower overall amount of OXT given may have additionally increased the possibility that it primarily influenced its receptor’s excitatory Gq-coupled intracellular cascade more than the inhibitory Gi-coupled one [38-40]. The slightly longer duration of treatment in our current study (6 weeks compared to 4-5 weeks) may have also contributed to differences in primary outcome measures. However, our preclinical evidence in healthy control adults that a 24IU dose of OXT given daily blunts both its behavioral and neural effects after 5 days, whereas dosing every other day does not [47,48], suggests that the infrequent dose regime is likely to have been the most influential factor in maintaining optimal functional effects by reducing the impact of repeated dosing on receptor desensitization and down-regulation.
In terms of secondary outcome measures OXT significantly increased behavioral adaptability scores (ABAS-II GAC) which may reflect a more general positive impact on adaptive behavior as a result of reduced autistic symptoms. As further potential objective measures of improved symptoms we included two eye tracking paradigms for the first time which have previously been shown to discriminate autistic from typically developing children. For the dynamic social (dancing children) paired with dynamic geometric stimuli autistic children spend proportionately less time viewing the social stimuli and correspondingly more for the geometric ones [58,59]. Our finding that after OXT treatment the autistic children spent proportionately more time viewing the social stimuli is therefore suggestive of an increased interest in them. In regard to viewing patterns for faces, previous findings have generally reported that autistic individuals show reduced time viewing the eye region [13,14]. Interestingly, we found that while OXT increased the proportion of time viewing the eye-region of happy, angry and neutral faces, it decreased that towards the eyes of fearful faces and so effects in children may be expression dependent. Notably, significant effects of OXT in these two paradigms was only found post-treatment (i.e. 6 weeks) and not after 3 weeks, possibly indicating that greater effects might have occurred following a longer treatment period. Indeed, the RCI analysis of the eye-tracking findings only indicated a reliable OXT effect in 22-24% of the children. It should be noted that in adult autistic and healthy subjects increased time viewing dynamic social stimuli [11] and gaze towards the eyes [9-11] has been reported after only a single dose of intranasal OXT. This may reflect a highly transient effect immediately following a single administration.
Given individual variation in the magnitude of improved symptoms following intranasal OXT treatment we additionally investigated factors which might potentially contributions to treatment outcome. There were no associations with age and with only 3 female subjects included in the trial it was not possible to perform a meaningful comparison with males. The magnitude of baseline OXT increase in saliva significantly correlated with improved total SRS-2 and ABAS-II GAC scores, suggesting that the ability of treatment to generally increase baseline peripheral OXT concentrations was a factor in determining improvement in social symptoms and behavioral adaptability. While it should be noted that saliva OXT concentration changes do not necessarily reflect similar changes in the brain, a recent study has reported stronger associations between saliva concentrations and those in cerebrospinal fluid compared to those in blood [70]. It is unclear however, whether individual differences in increased baseline OXT concentrations were due to efficiency of intranasal administration, size of nasal passages or other physical or behavioral related factors. Indeed, it is possible that increased OXT concentrations might not have been due only to the course of exogenous intranasal treatment per se but also to increased endogenous concentrations resulting from greater social interactions. The lack of a simple association between basal concentrations of OXT per se and symptom improvements is supported by the fact that, in contrast to a previous trial [19], pre-treatment basal blood and saliva OXT concentrations were not associated with subsequent ADOS-2 or SRS-2 or ABAS-II GAC score improvements.
There is some preliminary evidence that OXT receptor genotype might have influenced symptom improvements with increases in OXT concentrations and improved SRS-2 scores being less in T+ compared with CC carriers of rs2268491, although this would not survive correction for multiple SNPs. A meta-analysis has identified rs2268491 as being associated with autism [16] and T+ carriers with autism show a reduced neural response during an emotion recognition task compared to controls [71]. In terms of autism social sub-type assessed using the BASQ there was no significant difference in observed ADOS-2, and SRS-2 improvements in aloof, passive and active-but-odd subtype, indicating that OXT had broadly similar effects across the different sub-types. On the other hand, for the eye-tracking measures increased time spent looking at the eyes of angry faces was greater in the aloof group and decreased time spent looking at the eyes of fear faces was greater in the aloof and passive groups relative to the active-but-odd group. Thus, possibly OXT may enhance social attention most in the main aloof and passive autism subtypes. A previous study using a combined electro-acupuncture and behavioral intervention has reported greater improvements in autistic symptoms and increases in OXT in the aloof and passive subtypes [72].
In line with previous trials of intranasal OXT [19,22,25] there were no serious side effects and non-serious ones had a similar incidence during the OXT and PLC phases. The only significant effect of OXT was a small increase in daily urination frequency in line with a previous trial [22].
A 6-month follow-up after completion of the treatment phases revealed SRS-2 scores were significantly improved compared to those found at the end of the PLC phase of the trial and similar to those found after the OXT phase. This, together with some evidence for the wash-out period between phases being insufficient to reverse improvements following OXT treatment, suggests that the 6-week treatment may have produced some long-lasting benefits, although this remains to be demonstrated in a parallel design trial.
A limitation of the current study is that the cross-over design may have potentially reduced overall improvement due to some carry over influence in the group of participants receiving OXT first. A future parallel design study could address this and better assess the duration of OXT effects. The eye-tracking results obtained only achieved significance after the end of the 6-week treatmentperiod and possibly longer treatment durations may have achieved greater improvements in these and other measures. Finally, the required period of positive social interaction/play between caretakers and children following intranasal treatments was not strictly controlled or structured in order to allow individual flexibility. Future trials using more formal positive behavioral interventions might result in greater symptom improvements.
In summary, this cross-over trial on 41 young children with ASD has demonstrated that a 6-week intranasal OXT treatment using a novel infrequent dose regime followed by a period of positive social interaction is safe and effective in improving core symptoms of ASD. Additionally, objective eye-tracking assessments of attention to social cues were applied for the first time to demonstrate improved social attention following OXT treatment. Important implications of the current findings are that to achieve optimal beneficial improvements in autistic symptoms, exogenous OXT administration may need to be given infrequently to avoid receptor desensitization and in the context of subsequent positive social experience. Our findings suggest that we need to move away from the previous treatment philosophy of trying simply to increase basal concentrations of the peptide using more frequent and progressively higher doses, and without deliberately associating administration with positive social experience.
Data Availability
All data for the Tables and Figures in the main paper and supplementary will be provided in the final published version after peer review
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
Funding Sources
This work was supported by: Key Technological Projects of Guangdong Province “Development of New Tools for Diagnosis and Treatment of Autism” grant 2018B030335001 (KMK) ;UESTC High-end Expert Project Development grant Y0301902610100201 (KMK). These funding sources played no role in the trial itself or in the writing or submission of the paper.
Author Contributions
Authors contributions are as follows: Conceptualization: KMK; Methodology: JL, LZ, WZ, JK, YZ, CM, RZ, BB, KMK; Investigation: JL, SZ, JK, QZ, YZ, QL, MF; Visualization: JL, WZ, QL, KMK; Funding acquisition: KMK; Project administration: JL, LZ, SZ, CL, QL, ZC, WY; Supervision: JL, LZ, WY, KMK; Writing – original draft: JL, WZ, KMK; Writing – review & editing: JL, LZ, WZ, SZ, CL, JK, QZ, YZ, QL, ZC, MF, CM, RZ, WY, BB, KMK.
Acknowledgement
We thank the children and their families for their participation and support in this study.
References
