Behavioural and psychological features of PTEN mutations: a systematic review of the literature and meta-analysis of the prevalence of autism spectrum disorder characteristics

Background Phosphatase and tensin homologue (PTEN) is a cancer suppressor gene. Constitutional mutations affecting this gene are associated with several conditions, collectively termed PTEN hamartoma tumour syndromes (PHTS). In addition to hamartomas, PTEN aberrations have been associated with a range of non-tumoural phenotypes such as macrocephaly, and research indicates possibly increased rates of developmental delay and autism spectrum disorder (ASD) for people with germline mutations affecting PTEN. Method A systematic review of literature reporting behavioural and psychological variables for people with constitutional PTEN mutations/PHTS was conducted using four databases. Following in-depth screening, 25 articles met the inclusion criteria and were used in the review. Fourteen papers reported the proportion of people with PTEN mutations/PTHS meeting criteria for or having characteristics of ASD and were thus used in a pooled prevalence meta-analysis. Results Meta-analysis using a random effects model estimated pooled prevalence of ASD characteristics at 25% (95% CI 16–33%), although this should be interpreted cautiously due to possible biases in existing literature. Intellectual disability and developmental delay (global, motor and speech and language) were also reported frequently. Emotional difficulties and impaired cognitive functioning in specific domains were noted but assessed/reported less frequently. Methods of assessment of psychological/behavioural factors varied widely (with retrospective examination of medical records common). Conclusions Existing research suggests approximately 25% of people with constitutional PTEN mutations may meet criteria for or have characteristics of ASD. Studies have also begun to establish a range of possible cognitive impairments in affected individuals, especially when ASD is also reported. However, further large-scale studies are needed to elucidate psychological/behavioural corollaries of this mutation, and how they may relate to physiological/physical characteristics. Supplementary Information The online version contains supplementary material available at 10.1186/s11689-021-09406-w.


Background
Phosphatase and tensin homologue (PTEN), located on chromosome 10 (10q23. 3), was initially reported by Li et al. [1] and governs many processes in the cells which are disrupted in cancer [2]. For this reason, PTEN is recognised as a tumour suppressor gene. It has also been shown to play an important role in brain development [3]. PTEN mutations are related to an elevated risk of both malignant [4] and benign tumours. Conditions associated with constitutional PTEN mutations are collectively known as PTEN hamartoma tumour syndromes (PHTS). These include Cowden Syndrome (CS), Bannayan-Riley-Ruvalcaba Syndrome (BRRS) and Lhermitte-Duclos disease [5].
Constitutional PTEN mutations are found in 57 to 65% [6][7][8] and approximately 80% [5] of individuals diagnosed with BRRS and CS respectively. It has been suggested that a distinction between CS and BRRS is unnecessary, with age-related penetrance being the salient difference between features [9]. Indeed, 78% of individuals with a diagnosed constitutional PTEN mutation met criteria for both CS and BRRS [10], with common clinical features including (amongst other physical characteristics) hamartomas and macrocephaly [5,11], the latter of which is reported for 85% of those with a CS diagnosis.
Whilst autism spectrum disorder (ASD) is not a listed criterion for PHTS, it has frequently been reported in patients with constitutional PTEN mutations [12,13]. ASD and PTEN were initially linked in 2005 by Butler et al. [14] who reported that three of a group of eighteen individuals with ASD and macrocephaly had germline PTEN mutations. Mouse models suggest that deletion of PTEN in the cerebral cortex and hippocampus results in increased rates of macrocephaly and abnormal social interactions [15][16][17].
Whilst idiopathic ASD is considered multifactorial [18], elevated rates of ASD have also been observed in a number of genetic neurodevelopmental syndrome groups, such as Fragile X and Cornelia de Lange syndromes (see [19] for a meta-analysis). Previous research has also indicated that the precise profile of ASD-related behaviours may differ between different genetic syndrome groups and from that seen in idiopathic ASD (e.g. [20,21]). Furthermore, there is evidence that certain social and emotional characteristics, developmental sequalae and categories of psychological distress may be phenotypic of a number of the more extensively researched genetic neurodevelopmental syndromes. For example, social anxiety may characterise Fragile X syndrome [22], low mood is especially prevalent in Cornelia de Lange syndrome [23], and increased rates of psychosis have been recognised in those with 22q11.2 deletion syndrome. However, in the case of constitutional PTEN mutations, behavioural/ psychological research remains in its early stages. Whilst a number of papers have now been published in which ASD has been reported in a proportion of study participants with PTEN mutations, the overall prevalence of ASD in this population remains unknown. We are not aware of any previous pooled prevalence meta-analyses.
Research into neurodevelopmental, cognitive or behavioural features not related to ASD remains limited for those with PTEN mutations. Cognitive dysfunction has most frequently been reported in non-human animals [24][25][26]. Memory impairments, as well as repetitive and "depression-like" behaviours, have also been reported in PTEN-mutated mice [24]. To the authors' knowledge, no systematic reviews of psychological/behavioural corollaries of constitutional PTEN mutations in humans have been published to date. Previous reviews exploring a phenotype for PTEN mutations have focused specifically on ASD (without meta-analysing its frequency) [27,28] or individual disorders and their clinical features [29][30][31].

The present review
As is common with newly described conditions, research describing behavioural and psychological differences is often presented in disparate accounts and small studies. The current review aimed to systematically identify and synthesise literature reporting behavioural and psychological characteristics associated with PTEN mutations, including ASD, cognitive, emotional, social, sensory and motor aspects. A meta-analysis of prevalence rates of characteristics of ASD was also conducted. This may inform the theoretical understanding of implications of PTEN changes, and guide clinical practice and service development for those with PTEN mutations and their families.

Methods
The review was conducted in accordance with the Preferred Reporting Items for Systematic Review and metaanalysis protocols (PRISMA-P) 2015 statement [32].

Search strategy and selection criteria
Two comprehensive sets of search terms representing, respectively, PTEN mutations/PHTS and behavioural/psychological features, were developed (Table 1). These were Keywords: PTEN, PTEN hamartoma tumour syndrome, Autism spectrum disorder, Development, Cognition, Behaviour, Emotional difficulties Page 3 of 27 Cummings et al. Journal of Neurodevelopmental Disorders (2022) 14:1 informed by hand searches of terminology in relevant published research, reference to the OMIM website [33], and consultation with authors in the field and library staff at the Universities of Birmingham and Leicester (UK). These terms were used to search Web of Science, SCOPUS, Psy-cINFO and Cumulative Index of Nursing and Allied Health Literature between 3rd and 6th February 2020. Filters were applied to ensure the papers were written in English and were in peer-reviewed journal articles from 1997 onwards (when the PTEN gene was first reported on).
Following the removal of duplicates, 723 titles and abstracts were screened with reference to the following exclusion criteria: (a) no mention of PTEN mutations or diagnosis related to constitutional PTEN mutations; (b) non-human or molecular studies; (c) no behavioural, cognitive or developmental aspect, and (d) book chapters. This left 98 articles whose full text was screened using criteria in Table 2.
Twenty-five articles met full criteria for the review (Fig. 1).
Data regarding sample size, demographic information, recruitment procedure, assessments used, and relevant findings were extracted from included papers.
Two types of studies were identified: those in which participants were recruited/included on the basis of having an identified PTEN mutation or PTEN-related condition, with behavioural/psychological characteristics examined/reported (group A); those in which participants were selected on the basis of some other factor (e.g. macrocephaly), and these participants were tested for PTEN mutations (group B). For group B papers, details of behavioural/psychological characteristics of participants are only reported here for those with PTEN mutations or diagnoses of PTEN-related conditions.

Quality/bias appraisal tool
The 25 papers were assessed using the criteria developed by Richards et al. [19], adapted for the current review. Group B studies were rated on sample identification, confirmation of syndrome and quality of assessment of behavioural/psychological characteristic. For papers in group A, an additional criterion was assessed: presence and quality of a comparison group. It should be noted that these criteria are focused on establishing understanding of behavioural/psychological characteristics for

Meta-analysis of ASD prevalence
A meta-analysis was conducted of characteristics of ASD prevalence for all papers in which relevant data were reported. To determine the prevalence of diagnosis and characteristics of ASD in those with PTEN mutations or PHTS, the total number of these participants reported in the sample, and the number of those described as having ASD, or features of ASD such as "Autistic tendencies", Asperger's or "Autistic features" were extracted from each paper. The analysis was also repeated with only papers with more than 10 participants and papers specifically reporting on ASD/autism (excluding those reporting on "tendencies"/"features"). Meta-analytic weighted prevalence values were generated using the generic inverse variance method. A random effects model was selected to allow for between-study variation reflecting both sampling errors and other factors [34]. Initial Q-Q plots did not indicate marked deviations from normality for the prevalence estimates; therefore, the DerSimonian and Laird method was used to calculate between-studies variance. An additional quality effects model was also employed, with adjusted weightings according to studies' overall risk-of-bias ratings. In calculating the overall risk-of-bias rating for this analysis, the Assessment criterion focused solely on the assessment of ASD. The "quality of control group" criterion was removed for group A studies.
The existence of possible publication bias was assessed using the visual inspection of a funnel plot, in which the magnitude of the studies' proportion estimates are plotted against the square roots of the studies' sampling variances. Following Terrin et al. 's [35] demonstration of the unreliability of subjective judgements of funnel plot symmetry, Egger et al. 's [36] linear regression test of funnel plot asymmetry was also carried out. A trim and fill method was then used to model and correct for asymmetry due to potential publication bias [37,38], producing adjusted weighted average prevalence estimates.

Study characteristics
Information summarising the 25 studies analysed can be found in Tables 3 and 4. Those which were also used in the meta-analysis of prevalence of ASD/characteristic of ASD are marked with an asterisk. Overall, quality appraisal/risk of bias scores ranged from 0.5 to 0.89 (M = 0.64, SD = 0.12). Quality scores of the first author (KC) are presented in Tables 5 and 6. However, all studies were appraised independently by the second author (AWa), and inter-rater reliability was found to be excellent (two-way random effects, consistency, average-measures intraclass correlation coefficient for the overall score was .99, 95% CI .97-.99).

Participant characteristics Group a
The 13 studies in group A reported on a total of 1263 participants (51% male) with confirmed PTEN mutations or a PTEN-related condition (although, as with research into many rare syndrome groups, it is not possible to ascertain whether there is overlap between samples in some of these papers) and 93 control participants (including those with ASD with or without macrocephaly but no PTEN mutation and typically developing controls). Sample size ranged from six to 511 individuals. Age range varied from newborn to 89 years, with some studies reporting only a mean age. Only five studies included participants over the age of 30. The recruitment process was variable, and often only limited information about this was provided. Five papers had recruited people solely on the basis of PTEN mutations, three papers on the basis of PTEN mutations and some other feature (e.g. white matter lesions/disorders), and four papers reported on patients diagnosed with CS/BBRS and/or confirmed PTEN mutations.
Only two studies made comparisons of behavioural/ psychological characteristics of individuals with PTEN mutations with other groups. Frazier et al. [44] used comparison groups of individuals with macrocephaly and ASD (n = 16), ASD without macrocephaly (n = 38) and healthy controls (n = 14). Busch et al. [42] included a comparison group of individuals with macrocephaly and ASD (n = 25).

Group B
Group B papers (N = 12) reported on a total of 5353 participants, including data from two large prevalence studies: O'Roak et al. and Saskin et al. [55,57]. A total of 56 participants in these papers (1.0%) had a confirmed PTEN mutation or diagnosis of PHTS (confirmed number of cases ranged from three to 11 per paper), with age ranging from 1.6 to 35 years. Two studies did not provide demographic data specifically for those with PTEN mutations [55,57]. The nature of the overall samples varied, with participants recruited for studies on the basis of ASD (3 studies), macrocephaly and ASD (2 studies), macrocephaly and other developmental/cognitive/behavioural/neurological symptoms (3 studies), suspected PHTS (1 study) or having been tested for PTHS/PTEN mutation (2 studies).

Measures
In group A, four papers (31%) utilised or reported neuropsychological testing or measures, with seven (54%) gathering their data through medical records or developmental review and two (15%) not stating how the characteristics were assessed. In group B, nine papers (75%) reported at least one neuropsychological measure, and three studies did not provide this information [55,57,59].
The most commonly used measure to identify autistic features or record a diagnosis of ASD (where a measure was identified at all) was the Autism Diagnostic Observation Schedule (ADOS; [61]), used in five studies. The Autism Diagnostic Interview-Revised (ADI; [62]) was also frequently used (four studies). A range of measures were used to determine cognitive ability, including various editions of the Wechsler Adult Intelligence Scale [63] and the Wechsler Memory Scale [64].

Common themes
A range of characteristics were noted in those with a diagnosed PTEN mutation (Table 7). Papers in group A reported on a wider range of difficulties, including emotional difficulties [39,45] and specific types of cognitive impairments [41,42,44]. Group B largely focussed on the ability levels of their participants with only Orrico et al. [56] and McBride et al. [53] reporting emotional difficulties in their studies.

ASD and autism spectrum characteristics
ASD or autistic features were the most frequently reported characteristic of participants and were reported in 19 studies (76%).
The meta-analysis was also re-run including only studies with 10 or more total participants, due to the tendency for studies with smaller sample sizes to show greater variability in their measurement. This also omitted the two papers [51,59] which reported "autistic features" or "autistic tendencies" rather than ASD per se. This did not markedly affect prevalence estimates, with the random effects model again estimating a pooled prevalence of 25% (95% CI 14-36%). The meta-analysis was also re-run with group A papers only, since group B papers may be subject to extra/different sources of sampling bias (since samples comprised subsets of larger clinical groups, often defined by specific clinical characteristics). With only the eight papers from group A, the estimated prevalence was similar, at 23% (95% CI 13-33%).

Other aspects of ASD
A number of studies suggested relationships between ASD and other characteristics for participants with PTEN mutations. Three studies [41,42,44], compared participants with both PTEN mutations and ASD with other groups. These data indicated that PTEN mutations, lower general functioning and ASD may be interrelated: participants with PTEN mutations and ASD were more greatly impaired in a number of domains, including overall intellectual functioning, attention, inhibition, expressive and receptive language and motor coordination, than those with PTEN mutation but no ASD [42], and participants with PTEN mutations and ASD had lower average ability (including lower processing speed (d = 1.15), working memory (d = 1.07), auditory immediate memory and adaptive function most notably community living) than individuals with ASD associated with macrocephaly without PTEN mutations [44]. Effect size was reduced following adjustment for IQ scores (processing speed: Χ 2 = 3.71, p = 0.054 and working memory: Χ 2 = 2.63, p = 0.105). Scores on the ADI-R did not significantly differ between those with ASD and a PTEN mutation and those with ASD with or without macrocephaly [44] suggesting that levels of ASD symptomatology for those with PTEN mutations may not differ significantly from levels of ASD symptomatology reported for those with ASD of different aetiology. Unfortunately, however, scores on ASD measures representing ASD severity were reported only infrequently and not in a manner allowing robust comparison other groups (e.g. McBride et al. [53] reported ADOS II score for a single participant only). Sensory dysfunction, which features in DSM-5 criteria for ASD [65], was reported in one study [42]. Caregivers of participants with ASD and PTEN mutations observed greater symptoms of sensory dysfunction than the caregivers of those with PTEN mutations without ASD and the caregivers of children with macrocephaly and ASD. However, participants with PTEN mutations but without ASD were also reported to have impaired sensory processing. Butler et al. [14] also described sensory integration difficulties in one participant.

Cognitive ability, developmental delay and intellectual disability
Participants' ability levels were defined, assessed and reported in a variety of ways across papers. Some provided specific intelligence quotients (IQ), others report developmental quotients, and still others stated only whether intellectual disability or developmental delay was present for individuals. IQ was reported in three studies in group A, and four studies in group B. Hansen-Kiss et al. [45] found that 15 of 47 participants (32%) identified as having a PTEN mutation, PHTS, CS or BRRS in their electronic records had an IQ of less than 80 (mean = 65), with 18 more (38%) having a documented history of intellectual disability or developmental delay. The full IQ range in this study was 39 to 124. This variation may relate to their recruitment method of searching medical records and therefore not limiting participation to those who can complete certain measures. Busch et al. [41] and Lachlan et al. [9] reported global impairments (borderline or lower IQ) or learning difficulties in 12% of participants, respectively. Busch et al. [41] reported IQs ranging from 80 to 120. It must be noted that participants were required to sit through four hours of neuropsychological testing, which may have resulted in individuals with lower not volunteering the study. O'Roak et al. [55] reported the non-verbal IQs of three participants (with their overall sample selected on the basis of having ASD), ranging from 50 to 77.
Other descriptions of cognitive ability/disability included Yehia et al. [49] reporting a learning disability or "mental retardation" in 22 (4%). In group B, intellectual disability was reported in three papers (e.g. in one of four [53] and one of three [59] participants). Some papers did not differentiate between intellectual disability and developmental delay [43,47]. In these papers, intellectual disability and/or developmental delay were reported in 92% and 56% of participants respectively.
Developmental quotients (DQ) were also used to describe ability levels of participants in three group B studies [51,54,60], based on a variety of measures including the KIDS [66] and Kyoto Scale of Psychological Development [67]. DQs ranged from 30 [51] to 85 [51,54] across these studies.
Developmental delay (often with no further specification) was reported in 48% of the reviewed studies (n = 12; three group A studies and nine group B papers). Reported prevalence in group A papers varied: Ciaccio et al. [43] and Vanderver et al. [48] reported 9 of 16 and 23 of 23 participants respectively to have a developmental delay, and Yehia et al. [49] reported global or variable developmental delay in 91 out of 511 participants. In group B, prevalence rates of developmental delay in participants ranged from 16 to 66% (excluding Yeung et al. [60] who looked at PTEN in this population).
Amongst specific developmental delays, motor delay was the most commonly reported (n = 11 studies, seven in group A). Busch et al. [41] found that participants with PHTS scored significantly lower than normative data in motor functioning (t(22) = − 5.02, p = .001, d = − .94), specifically in fine manual dexterity. Kato et al. [51] described motor delay in four of their six participants with a PTEN mutation. In this group, participants began walking between 14 and 29 months, with three participants walking after 26 months and showing motor delay. As well as delays in walking, general and psychomotor delays [56] and fine motor delays [14] were also reported.
Along with motor delays, speech and language delays were reported in five studies in group A and 5 studies in group B. Prevalence rates varied between 27 and 57% across the studies [39,40,46] in group A. Smpokou et al. [47] did not delineate between motor and language delays. Reports of profiles of ability across domains for individuals are rare. Busa et al. [40] report on one participant with an uneven ability profile, with scores within most indices on the WAIS in the "normal" range but a working memory index of 67.

Attention, executive functioning and memory
Attentional difficulties were reported in four studies (three in group A). Reduced working memory abilities and processing speed were reported at group level in three group A studies [39,42,44]. Attention Deficit Hyperactivity Disorder (ADHD) was reported in two of 11 participants by Balci et al. [39], and in an unspecified number by Hansen-Kiss et al. [45], and a short attention span was noted in two of three individuals in one study [14]. Two studies in group A [39,41] identified poorer memory functioning in their participants with PTEN mutations. Busch et al. [41] reported a difference between people with PTEN mutations and data from a normative comparison group in the memory recall domain (small effect size [d = 0.38]), with 12 participants (47%) showing reduced performance on a memory recall measure, although no significant differences were found in recognition memory. Balci et al. [39] reported two patients (18%) with memory problems indicated in medical records (with no further details).
Busch et al. [41,42] reported impairments in the executive functioning domain, in which participants with PTEN mutations overall scored significantly lower than population controls (d = − 0.7, p = 0.001).
As previously mentioned, Frazier et al. [44] found that the large effect sizes for deficits in processing speed and working memory reduced following adjustment for full-scale IQ. However, when exploring the cognitive abilities of those with PHTS, most (88%) of whom had IQ scores in at least the low average range, Busch et al. [41] noted greater difficulty on measures of verbal fluency and fine motor skills than controls.

Emotional difficulties
In group A, four studies reported emotional difficulties, including mental health diagnoses, in their participants. Balci et al. [39] reported two of 11 paediatric participants were diagnosed with generalised anxiety disorder. One participant had also received a diagnosis of obsessive-compulsive disorder (OCD) following a suicide attempt and self-harm. A psychotic episode was reported in a 5-year-old [39]. Hansen-Kiss et al. [45] reported diagnoses of disruptive behaviour disorder, oppositional defiance disorder, aggression, anxiety, depression, bipolar disorder and OCD in 34% (n = 16), but with no further delineations. "Problem behaviour" and poor adaptive functioning was more commonly reported in those with PTEN mutations and ASD than just PTEN mutations [42,44].
Emotional difficulties were less frequently reported in group B studies. A nine-year-old female was reported to be diagnosed with an unspecified affective disorder and behavioural problems which were described as "oppositional and anger" with no further information [53].

Discussion
The current review examines literature reporting psychological and behavioural characteristics in individuals with constitutional PTEN mutations. The 25 studies meeting criteria for the review fell into two categories: those that investigated the characteristics of individuals recruited after confirmed PTEN mutations or PHTS (group A), and those which assessed for presence of PTEN mutations in a sample of participants with specific characteristics, such as ASD and macrocephaly (group B). There was a similar number of studies in each group, although the total number of participants with PTEN mutations was considerably greater for group A.
ASD was the most commonly reported characteristic. A meta-analysis of the prevalence of ASD or characteristics of ASD (including 14 papers) revealed a weighted average prevalence of 25% (95% CI 16-33%). This was not markedly changed by weighting papers by risk of bias/quality ratings, by including only group A papers, by omitting papers with ten or fewer participants, or by omitting those who referred to "features of ASD" rather than ASD. Asymmetry of papers' reported prevalence around the weighted average raised the possibility of publication bias; it is possible that ASD prevalence remains unreported/unpublished where this prevalence is lower. Following correction for possible publication bias, the estimate of prevalence decreased to 17% (95% CI 8-27%). However, even at the estimated lower confidence interval for this lower estimate, prevalence exceeds that in the general population (1-2% [68,69];). The calculated prevalence adjusted for publication bias is similar to that estimated in neurofibromatosis type 1 (18%) and Down's syndrome (16%), as noted by Richards et al. [19]. The impact of possible publication bias suggests that more large-scale studies looking at ASD and ASD characteristics in those with PTEN mutations and PHTS are needed to accurately estimate prevalence. Further to this, authors should endeavour to report prevalence of diagnoses in clinical samples where this information is available even if this is low.
Reviewed literature also suggests that individuals with ASD and PTEN mutations may differ on a number of psychological/behavioural dimensions from those without ASD, with evidence of lower ability in a number of areas (see below) for those also carrying diagnoses of ASD. This is in line with evidence that ASD is associated with lower ability more generally [70]. However, there was also evidence that those with PTEN mutations and ASD may also have more difficulties than those with ASD and macrocephaly of different aetiology [42], suggesting that the combination of PTEN mutations and ASD may be particularly associated with lower abilities, an association that should be explored more in further research. There was evidence of overall reduced IQ for individuals with PTEN mutations. It should also be noted, however, that individuals' reported cognitive abilities varied greatly, with some papers reporting on individuals with IQs over two standard deviations above the population mean (e.g. Busch et al. [41] reported an IQ range of 80-135). Data using standardised measures suggest that people with PTEN and ASD may on average have impairments in a number of domains [42,44], including attention, working memory, processing speed, language, visual-spatial abilities. Where no ASD is present, the picture may be less clear, with one study indicating that a group of 23 individuals with PTEN mutations but no ASD did not score statistically significantly differently from normative comparison groups on some measures of cognitive functioning, including attention and processing speed [42]. However, given the low sample size, the possibility of a type 2 error due to low power must be considered. Despite associations between ASD and lower ability, the frequency with which ASD occurs in the absence of significant impairments in general ability remains a question for future research.
Where standardised measures of motor abilities were used at a group level, functioning was found to be significantly lower for people with BBRS/CS than in normative data [41] and was impaired for people with PTEN mutations both with and without ASD (although more so for those with ASD) [42]. However, again, there is apparent variability at the individual level.
Attention, executive functioning and memory were also reported to be impaired at a group level [39,41,42,44,45]. A small number of individuals were reported to have diagnoses of ADHD [39,45], although this was not widely investigated/reported across papers so the degree to which impairments relate to specific developmental diagnoses is unclear. It has been postulated that deficits in processing speed and working memory associated with PTEN mutations may be related to poorly developed white matter [42,44], and details of this possible association should be explored in further research. Full-scale IQ has been shown to be significantly related to executive functioning [71] and scores on tasks tapping into working memory contribute to a full-scale IQ score. For this reason, when exploring impairments in these domains, it is important to question whether these are to be expected given the individual's IQ. The relative degree of impairment in these domains for people with PTEN differences, and the strength of relationships between impairments in different domains of cognitive functioning, are yet to be fully explored. Future research should build on existing work [41,44].
Emotional difficulties were reported/assessed only sporadically. Where reported, there were suggestions that the prevalence of these difficulties may be high: Hansen-Kiss et al. [45] identified these issues in 34% of their participants, citing anxiety, bipolar disorder and OCD (although with few further details). "Disruptive" or "problem" behaviour was also reported in three papers. However, the lack of systematic investigation, using established measures and appropriate comparison groups, precludes knowledge of whether emotional difficulties occur differently from or at a higher rate than in the general population and/or other genetic neurodevelopmental syndrome groups. How this may relate to other difficulties such as ASD (known to be associated with anxiety, for example), also remains to be ascertained.
Relationships between different behavioural/psychological variables and specific genetic, physical or physiological characteristics were not generally explored in the reviewed papers. The precise relationship between different PTEN variants and psychological corollaries remains to be delineated. Recent research has begun to explore this, e.g. Yehia et al. [50] found in their sample of 309 individuals with PTEN mutations that those with ASD/DD had an overall increased burden of copy number variants).

Strengths and limitations in the literature
For most studies, the presence of a PTEN mutation had been confirmed for all participants. Two studies [41,49], however, included participants with diagnoses of CS and BRRS but without a PTEN mutation (n = 204). The precise nature of the mutations, and the additional difficulties also featured in inclusion criteria/recruitment processes for many studies, leads to potential differences in the nature of participant groups. These factors mean that interpretation of synthesised results, including metaanalytic estimates, should remain cautious. It may also be that some of the heterogeneity (e.g., in reported ability levels) between studies reflects differences in recruitment of samples which cannot be entirely characterised in the present analysis (e.g., because of limited information given in the papers). This requires thorough consideration in future research.
Nine of the 25 studies recruited participants from multiple centres or databases either nationally [9,55,57] or internationally [12,48], which may enhance some aspects of generalisability. However, the data are largely from Western countries, and definitions and constructs of ASD and psychological distress may not relate to individuals in other cultures.
Small sample sizes may reflect the rarity (and relatively newly-identified) nature of the condition, which is likely to result in underpowered analyses. PTEN mutations are not routinely tested for and participants were frequently recruited through hospitals or clinics which may have led to a general bias in the literature, as this indicates that there was significant concern about the individual prior to genetic testing. The use of clinical review as a main method for identifying behavioural characteristics in these studies also introduces possible biases [72].

Clinical and research implications
Clinical evaluation and support of individuals with PTEN-related conditions should consider a wide range of possible corollaries, including ASD, cognitive and intellectual functioning, motor development and potentially, emotional difficulties.
Future research should employ a range of validated, standardised behavioural measures to allow a more comprehensive identification of the various domains associated with the behavioural phenotype of individuals with PTEN mutations (e.g. [73,74]). This will also aid more extensive comparisons with other groups (including those representing typical development, idiopathic ASD and other genetic neurodevelopmental conditions), which should also be included within studies, since this is important in defining the behavioural phenotype specifically associated with a syndrome group [75][76][77]. As more individuals with constitutional PTEN mutations are identified world-wide, and relevant support groups and databases grow, this may also allow researchers to assess psychological and behavioural factors for a greater number of individuals who may not otherwise have come to attention of services. This may then allow for samples which may be less biased towards specific difficulties, enhancing understanding of psychological/behavioural correlates of PTEN changes more broadly.

Conclusion
A systematic review of existing research with groups of people with constitutional PTEN-related conditions suggested a number of possible psychological/behavioural corollaries. Our meta-analysis estimated a prevalence of ASD or characteristics of ASD of approximately 25% (95% CI 16-33%), although it should be noted that this estimate may be inflated by publication bias, and should be interpreted with caution due to the varied nature of recruitment and basis on which ASD is determined. Further research is required on the qualitative nature of ASD phenomenology within this group. Research also indicates lower average cognitive abilities than in the general population, especially when ASD is also present, frequent reports of global developmental delay, motor and speech delay and cognitive impairment in those with PTEN mutations and PHTS. Wide variation in cognitive abilities is also noted. The relationship of psychological/behavioural variables with physiological or genetic factors remains relatively unexplored. Many studies are small scale, relying on retrospective reviews of medical records or unclear psychological assessment methods, and use of comparison groups was limited in available research. Future research, using detailed and well-established psychological assessment tools and appropriate comparison groups, may elucidate in greater depth the profile of possible characteristics associated with aberrations affecting PTEN.