Phelan–McDermid syndrome (PMS) is a micro-deletion syndrome with diagnosis based on cytogenetic, molecular cytogenetic and/or molecular evidence of loss or disruption at 22q13.3 [1]. The incidence of PMS is unknown, with under-diagnosis suspected due to the subtlety of the deletion [2]. Approximately 80% of people with PMS have de novo, simple terminal deletions, and the remaining 20% typically result from unbalanced translocations and ring chromosomes [1]. The 22q13 region contains the SHANK3 gene: haploinsufficiency of SHANK3 is proposed to cause the major features of PMS [3–5], and mutations in the SHANK1, SHANK2 and SHANK3 genes are associated with autism spectrum disorder [6]. Recent research has demonstrated that the SHANK3 mutation results in neuronal changes, including increase in input resistance to excitability, some impairment in dendritic branching and decreases in synaptic transmission [7]. These changes are mechanistically linked to impairments in hyperpolarization-activated cation (I_h) channels, with a suggestion that reduced I_h currents may account for some of the phenotypic characteristics observed in PMS [7]. Dysmorphic physical features associated with PMS are subtle and include hypotonia, normal to accelerated growth, long eye lashes, large ears, full brow, dolicocephaly, full cheeks, bulbous nose and pointed chin [1, 4, 8]. The most characteristic clinical features of PMS are moderate to profound intellectual disability and absent to severely delayed speech [1, 2, 8–10]. Preliminary research suggests that the physical features and severity of intellectual disability correlate with the size of the genetic deletion. However, expressive speech deficits are not associated with the size or type of deletion [8].

A number of behavioural characteristics have been reported in PMS. Hyperactivity, impulsivity and difficulties in sustaining attention have been identified; 34% of children with PMS were reported to have a diagnosis of attention deficit hyperactivity disorder (ADHD) [11], 36% of children scored above clinical cutoff for ADHD on a screening measure [12] and a high proportion of parents endorsed items indicative of impulsivity and inattention [11]. These findings suggest a potential association between PMS and ADHD phenomenology. However, few studies have employed measures designed for individuals with intellectual disabilities and none of the studies compared the results for the PMS group to contrast groups. Thus, it is unclear whether the presence of ADHD symptoms should be attributed to the behavioural phenotype of PMS, the severity of intellectual disability, age of the children assessed or the measures used. Similar threats to validity weaken results associating atypical affect with the behavioural phenotype of PMS. Cohort and case studies identified behaviours indicative of depression and psychosis/atypical bipolar disorder in PMS [11, 13]. However, given deficits in expressive language, it is unclear how internal experiences of positive symptoms of psychosis or depression have been reported and assessed. Nonetheless, given the clinical implications of mood disturbances, these findings warrant further investigation, utilising robust measures validated for individuals with intellectual disability and contrasting findings with appropriate comparison groups.

A final characteristic, frequently identified in PMS, is that of autism spectrum disorder (ASD) [2, 11, 12, 14] with a recent review suggesting that all individuals diagnosed with PMS should undergo specialist ASD assessments [15]. The putative association between ASD and PMS is of particular interest as the SHANK3 gene is one of the many implicated in the aetiology of idiopathic ASD [3, 16, 17] with SHANK3 deficiency associated with 0.5 to 2% of cases of ASD and intellectual disability [18]. Thus, delineation of the prevalence and phenomenology of ASD in PMS may have clinical implications for individuals with PMS and individuals with idiopathic ASD. Results from screening instruments have demonstrated convergent results: mean autism/pervasive developmental disorders scale scores for children on the Reiss Scales were above clinical cutoff [11]; 94% of children with PMS scored in the mild to moderate range for ASD and 67% in the severe range for ASD using the Childhood Autism Rating Scale [2]; 85% of children with PMS met the cutoff for ASD on the Social Communication Questionnaire (SCQ) and 67% met the more stringent cutoff for autism [12]. More robust evidence is found in studies employing clinical diagnostic measures of ASD. Soorya and colleagues [14] utilised both the Autism Diagnostic Observation Schedule (ADOS) [19] and Autism Diagnostic Interview-Revised (ADI-R) [20] and found that 84% of the sample with PMS met criteria for ASD and 75% met criteria for the more stringent classification of autistic disorder. However, whilst there appears to be a strong association between ASD phenomenology and PMS, no studies have employed contrast or comparison groups to evaluate whether ASD phenomenology can be identified as a specific component of the behavioural phenotype of PMS, rather than attributed to the level of intellectual disability associated with the syndrome. This is particularly important given the potential for over-estimating ASD when intellectual disability and expressive speech deficits are present [21].

Whilst there is a purportedly high prevalence of ASD phenomenology in PMS, the profile of ASD impairments in communication and social interaction and repetitive behaviour domains in the syndrome is less well described. The profile of ASD is known to vary across genetic syndromes [22]. For example, a recent meta-analysis demonstrated that ASD phenomenology is common in Cornelia de Lange and fragile X syndromes [23], and yet, detailed item-level analysis of screening [24] and diagnostic measures [25] reveal that both syndromes evidence an atypical profile of ASD. Philippe and colleagues [26] reported that whilst children with PMS attained high ADI-R scores, these only reached clinical thresholds in social interaction, play, and communication domains. They argue that the relative lack of repetitive behaviours distinguishes PMS from idiopathic ASD. However, the study was limited by not including an idiopathic ASD comparison group and relying upon visual inspection of data. Additionally, a number of sub-threshold items in the repetitive behaviour domains necessitated expressive language (e.g. delayed echolalia, verbal rituals). Interestingly, Soorya and colleagues [14] also found that interpretation of the ADI-R algorithm alone indicated that many children with PMS presented with sub-threshold levels of repetitive behaviour. However, when they included statistical analysis of all items, including a two factor algorithm of repetitive behaviour identified in research on the ADI-R, they found that repetitive and sensory-motor behaviours were present in the majority of the participants, and were similar in range to those reported in idiopathic ASD.

Finally, authors have suggested that behaviours indicative of psychopathology (psychosis and low mood) may be misinterpreted as ASD phenomenology in individuals with PMS [11]. Shaw and colleagues [11] report that some endorsed items could indicate both ASD and mental health problems, e.g. “Does not seem to listen when spoken to directly”, “Random and inappropriate speech” and “Appears confused”. Additionally, they suggest that other items such as “Maintains a rigid posture”, “Appears to be in a stupor, as if intoxicated” and “Laughs or appears angry for no apparent reason” may be more indicative of psychosis than ASD. However, it could be argued equally that these behaviours are indicative of repetitive behaviour, sensory difficulties or problems with emotional regulation, all of which are commonly reported in idiopathic ASD. Thus, there is a need to evaluate further the profile of ASD in PMS, utilising measures appropriate for individuals with intellectual disability, and with sufficient specificity and psychometric properties to allow for item-level statistical analysis. Additionally, these analyses need to be made in comparison to contrast groups, necessarily including individuals with idiopathic ASD, and ideally including groups with other genetic syndromes with known ASD profiles, in order to determine the relative position of the ASD profile in PMS.

A final point of interest is that the investigation of the profile of ASD impairments in PMS appears to have been largely driven by the hypothesised genetic links between PMS and idiopathic ASD. This has resulted in studies analysing the ASD profile of all participants in the PMS samples [11, 14, 26] in order to establish whether the profile in the syndrome is similar to individuals with idiopathic ASD. These data could support or weaken the hypothesised genetic SHANK3 link. A complementary analysis approach would be to restrict analyses to those who score above threshold for autism and ASD. These data would answer a second question about whether individuals with PMS meet criteria for autism for the same reasons as individuals with idiopathic ASD. Answers to this question would inform discussion of the specific clinical needs for individuals with PMS who evidence ASD behaviours, thus increasing the specificity of clinical provision and interventions for individuals with PMS.

The behavioural characteristics, prevalence and profile of ASD phenomenology in PMS were delineated in this study. The association between ASD phenomenology and broader behavioural and demographic characteristics was also evaluated. The novel recruitment of comparison groups with fragile X and Down syndrome, in which the profile of ASD phenomenology is well described, strengthens the validity of the study. The inclusion of a matched idiopathic ASD comparison group allows for robust delineation of the profile of ASD phenomenology in PMS. The use of validated measures, with appropriate psychometric properties established in populations with intellectual disabilities further contributes to the validity and reliability of the findings.

The results of the behavioural phenotype analyses revealed that individuals with PMS evidenced higher total mood scores than those with idiopathic ASD, but lower total mood scores than those with Down syndrome. Despite the between group differences identified at the total score level, there were no identified differences on the Mood or Interest and Pleasure subscales between the PMS and comparison groups. These findings support previous research identifying low mood in individuals with PMS [11], but also demonstrate the utility of including multiple comparison groups in order to position the behavioural phenotype in PMS relative to other syndromes. The PMS group achieved higher total mood scores than those with idiopathic ASD and comparable total mood scores to those with fragile X syndrome, suggesting that whilst lower mood is present in PMS, it may not be significantly atypical, given the degree of intellectual disability in the group.

The use of a carefully designed and detailed assessment of repetitive behaviour [36] revealed a mixed profile in individuals with PMS. The group evidenced similar levels of stereotyped behaviour, but lower levels of compulsive behaviour, insistence on sameness and total repetitive behaviour than both the fragile X syndrome and idiopathic ASD groups. This finding supports and synthesises divergent results demonstrating low levels of repetitive behaviour in PMS [26] and the presence of repetitive and sensory-motor behaviours in the group [14]. Individuals with PMS appear to evidence a dissociation between motor-driven repetitive behaviours, which are common in the sample, and more compulsive and routine-driven behaviours, which are less evident in the group. It is important to note that this finding is at the level of the total sample, including those who meet threshold for autism and those who do not. Finally, the results revealed no significant differences in levels of overactivity or impulsivity between the PMS and comparison groups. This finding differs from those previously reported, where high levels of ADHD-type behaviours were identified [11, 12]. However, previous research did not compare individuals with PMS to matched comparison groups, and thus, the high levels of activity and impulsivity may be more appropriately associated with the degree of intellectual disability in PMS rather than the behavioural phenotype of PMS per se.

The results demonstrated a high prevalence of ASD phenomenology in PMS, with 87% meeting threshold for ASD and 57% meeting the more stringent criteria for autism. These findings support the prevalence figures identified in previous studies using screening measures (94% mild–moderate ASD, 67% severe ASD [2]; 85% ASD, 67% autism [12]) and diagnostic tools (84% ASD, 75% autistic disorder [14]). The results of this study extend findings by demonstrating that a similar proportion of individuals with PMS meet threshold for ASD and autism as males with fragile X syndrome, in whom ASD phenomenology is common. Importantly, the proportion of individuals in the PMS group meeting clinical thresholds on the SCQ was significantly higher than the Down syndrome group, suggesting that a high prevalence of ASD phenomenology can be associated with the behavioural phenotype of PMS. It is important to note that whilst this study has demonstrated a high prevalence of ASD phenomenology in PMS, this does not directly equate to a high prevalence of ASD diagnoses in PMS, given the necessity of thorough, multimodal assessment in the clinical diagnoses of ASD.

Analyses to evaluate the profile of ASD phenomenology in the total PMS sample provided heterogeneous results across impairments in communication, social interaction and repetitive and restricted behaviour. Firstly, at subscale level, the group did not differ from the idiopathic ASD or fragile X syndrome groups in “ASD-like” communication impairments. Additionally, the PMS group evidenced more impairments than those with Down syndrome. This finding supports previous results highlighting “ASD-like” impairments in communication in PMS [14, 26]. Item-level analyses extended these findings to reveal that the PMS group evidenced specific impairments in “nodding to communicate yes”, with a higher proportion of the PMS sample scoring as impaired on this item than all three comparison groups, although this did not reach statistical significance when compared to the idiopathic ASD group who scored over the autism threshold. The PMS group did not significantly differ from the idiopathic ASD group on any item in the communication domain, suggesting that the profile of “ASD-like” communication impairments is similar in that of total PMS and idiopathic ASD groups.

Secondly, the PMS group did not differ from the idiopathic ASD, fragile X or Down syndrome groups in “ASD-like” repetitive behaviours. However, item-level analysis revealed that the PMS group was significantly less likely to engage in non-verbal ritualistic behaviours than those with fragile X syndrome or idiopathic ASD. This difference remained significant in the secondary analysis of individuals with PMS who scored above the clinical threshold for autism. Thus, the profile of repetitive behaviour is still somewhat unclear in PMS. Fine-grained observational analysis of repetitive behaviours would be beneficial, in order to detail topography, frequency and any potential management difficulties of repetitive behaviour in the syndrome.

Finally, at the subscale level, the PMS group evidenced significantly more impairments in social interaction than the Down syndrome group and showed comparable levels of impairment to the idiopathic ASD and fragile X syndrome groups. This finding supports data demonstrating “ASD-like” social interaction impairments in PMS [14, 26]. An interesting dissociation in social interaction was revealed at item level; the PMS group showed significantly more impairments in “Showing and directing attention” than both the Down syndrome and idiopathic ASD groups, but significantly less impairment in items assessing interest in, and responses to, other children. One interpretation of this finding is that there is a divergence in social skills and social motivation in PMS, with relatively preserved social motivation in contrast to deficits in social competence, potentially due to low levels of expressive speech. Alternatively, the result may represent a specific impairment in initiating interaction, with relatively preserved abilities to respond to interactions initiated by others. This finding warrants further investigation, including attempts to replicate the results in larger samples with PMS, using both indirect and direct assessments of social competence and motivation.

Whilst the profile of ASD impairments was varied within the total PMS sample, the results within the subgroup that scored above the autism threshold were very similar to the profile of behaviour in the idiopathic ASD group, suggesting that both groups reach clinical thresholds for autism due to a similar profile of behaviours. The results in this subgroup revealed that individuals with PMS were neither more nor less likely to score on items in the communication, social interaction or repetitive behaviour domain, than those with idiopathic ASD except for an increased impairment in “Showing and directing attention” and a decreased likelihood of ritualistic behaviour. This finding extends previous research, affording a more refined understanding of the nature ASD impairments in affected individuals with PMS. The result suggests that when individuals with PMS meet criteria for autism, they do so for similar reasons to those with idiopathic ASD. Clinically, this may indicate that interventions to support individuals with idiopathic ASD could be usefully applied to individuals with PMS who meet the diagnostic criteria. The result also replicates the specific deficit noted in the total sample in showing and directing attention. Interventions to extend the behavioural repertoires of individuals with PMS focused on behaviour to recruit and maintain others’ attention which may be warranted in this population.

The final results of this study demonstrated that across all demographic and behavioural scores, only “Interest and pleasure” was (negatively) correlated with SCQ score in the PMS group. The correlation between “Mood” and total SCQ score approached significance. These findings lend tangential support to previous research indicating an association between the presentation of mood disorders and ASD phenomenology in the syndrome [11]. However, given the strength of evidence of behaviours indicative of ASD in PMS, the correlation between interest and pleasure and SCQ score is not interpreted as substantiation of mood disorders being wholly explanatory for ASD phenomenology in PMS. Instead, it is possible that behaviours indicative of low mood are associated with ASD impairments in PMS. Alternatively, it may be that mood disorders and ASD impairments co-exist within PMS due to similar genetic underpinnings, perhaps with greater severity of mood disorder being associated with more significant genetic deletion, as ASD phenomenology is hypothesised to [8]. These hypotheses are tentative and further research is required to delineate the association between mood and ASD phenomenology in PMS, including any causal links between the two phenomena.

A number of caveats must be considered when interpreting the findings in this study. Firstly, the assessment of ASD phenomenology is somewhat limited, due to the utilisation of a screening measure rather than a diagnostic measure; the “gold standard” for assessment of ASD in individuals with intellectual disability is a combination of ADOS and ADI-R. However, utilising a brief parent screening measure reduced time and assessment demands and conferred the advantage of assessing multiple comparison groups in order to position the profile of ASD phenomenology in PMS relative to other syndromes [39]. Additionally, the SCQ is recognised as more appropriate for assessing ASD phenomenology in samples with intellectual disabilities than other ASD screening tools [38]. Similarly, the Wessex adaptive behaviour scores were utilised as a proxy measure for intellectual disability. Whilst it would have been beneficial to conduct full cognitive assessments of all of the participants, it would not have been possible within the scope of this study. Thus, a brief assessment of adaptive behaviour was chosen in order to balance the need to assess intellectual disability and the need to maximise participants in all the four groups. The limitations imposed by this method of assessing adaptive functioning should be considered as a caveat to the present study, particularly as the Wessex score was used as one of the indices for matching the groups. Future studies should seek to include robust cognitive assessments of intellectual functioning. Secondly, despite careful matching of the groups, it was not possible to reduce all differences in adaptive behaviour. Therefore, the Down syndrome group were significantly more able than the PMS, fragile X syndrome and idiopathic ASD samples. Previous researchers have argued that delineating the behavioural phenotype of a given genetic syndrome in relation to multiple other syndromes reduces the need for chronological or mental age matched comparison groups [39]. Additionally, the PMS, fragile X syndrome and idiopathic ASD groups were well matched for chronological age and adaptive ability. Nonetheless, the results should be interpreted with this caveat in mind. Finally, due to the relatively small PMS sample, there was insufficient statistical power to test causal associations between expressive speech, adaptive behaviour and ASD scores. Previous research has highlighted that it is important to explore these associations in samples with genetic syndromes [22]. Correlational evidence from this study indicates that adaptive behaviour was not associated with SCQ score; however, this still warrants further exploration in larger sample sizes, where causal statistical modelling is possible.

The results of this study have a number of important clinical implications. The results indicate that assessment of behaviours indicative of low mood should be routine in individuals with PMS. Research in individuals with severe intellectual disabilities has revealed that low mood scores may indicate pain and undiagnosed health conditions [40–42]. There are reports of gastro-oesophageal reflux and other painful conditions in PMS [1, 15]. Therefore, thorough health assessments should routinely be conducted for individuals with PMS. The results of this study have implications for research investigating the genetic underpinnings of idiopathic ASD. The results demonstrate that those with high levels of ASD impairment evidence a profile of ASD impairments that is similar to that of individuals with idiopathic ASD. However, the wider PMS sample presents a more atypical pattern with fewer impairments in repetitive behaviours. This may suggest that social and communicative impairments would be a useful autism endophenotype to be investigated in relation to 22q13.3 deletions and SHANK3 mutations more broadly [16]. Finally, the results of this study have implications for clinical trials in PMS. To translate recent pharmacological successes from pre-clinical studies to human trials [43, 44], nuanced behavioural phenotyping and identification of measures sensitive to change in behavioural characteristics are required. This study has identified unique aspects of the behavioural phenotype of PMS which should be considered as potential clinical targets (low mood, autism spectrum disorder characteristics) using measures appropriate for the level of intellectual disability present in the syndrome, highlighting both targets for intervention and measures with sensitivity to detect these difficulties.

In summary, this study has demonstrated that differences in mood and repetitive behaviour are common in PMS. Additionally, autism spectrum disorder phenomenology is prevalent within the syndrome. The profile of ASD impairments in the total sample with PMS is heterogeneous; the profile within those who meet clinical threshold for autism is analogous to those with idiopathic ASD. The presence of ASD phenomenology is associated with lower mood in those with PMS.

The results of this study demonstrate that ASD phenomenology is common in PMS and can be considered to be a component of the behavioural phenotype. These data support the assertion that SHANK3 mutations may be causally implicated in the development of ASD-type behaviours, particularly social and communication deficits, although further research is required to delineate the specificity of this endophenotype. The results extended previous studies by demonstrating that whilst the profile of ASD phenomenology in PMS is atypical, the profile in those who score above threshold on an autism screening measure is analogous to idiopathic ASD. This suggests that ASD-specific interventions could be usefully applied to groups with PMS who meet criteria for ASD. Fine-grained analysis of ASD phenomenology revealed an emerging dissociation between deficits in behaviours indicative of social skill, but relative preservation in behaviours indicative of social motivation. This suggests potential targets for psychological interventions in PMS.