A substantial amount of research shows a higher rate of autistic type problems in males compared to females. The 4:1 male to female ratio is one of the most consistent findings in autism spectrum disorder (ASD) research [1–3], and a gender difference has been a part of the description of ASD since the first characterisation of the disorders.

ASD occurs in conditions with X-linked recessive inheritance, but because of the rarity of these disorders, this inheritance cannot explain the male bias in prevalence of ASD [4]. The fact that most ASD risk loci are found in autosomal regions makes the male bias in ASD largely unexplained [5]. Most current data suggest that the male bias is more likely to be due to female protective factors rather than male-specific risk factors, but comprehensive molecular explanations are lacking for both [6].

Gender ratios in ASD differ substantially from study to study. Among individuals with ASD and normal cognitive functioning, gender differences as high as 9:1 have been reported [7]. A newer systematic review and meta-analysis from Loomes et al. [8] found a male to female ratio closer to 3:1 than 4:1. According to Loomes et al. [8], the main reasons for this change were both how ASD was diagnosed and what population were used to investigate the male to female ratio in ASD in different studies. Loomes [8] found that studies screening the general populations for ASD had a lower male to female ratio than studies investigating population with pre-existing diagnosis. In cohorts with ASD in combination with intellectual disability, the ratio varies between 2:1–7:1 [2, 4]. Loomes et al. [8] also found a lower male to female ratio in their meta-analysis in the subgroup of the studies including participants with lower IQ. Epidemiological studies describe the degree of intellectual disability and the ascertainment approach as major explanations behind the varying ratios that were reported [9].

The particular biological aetiologies of autistic problems are probably also relevant, even when the degree of intellectual disability (ID) is controlled for, but such a line of investigation has not yet been explored. How different biological pathways to ASD differ in the ASD-gender ratio may shed light on basic ASD biology.

ASD is in the Diagnostic and Statistical Manual V (DSM V) referred to as a dyad of impairments; difficulties in social interactions and social communications; and restricted and repetitive behaviour, interests, and activities [10]. Gender differences in profiles of autistic symptoms have a limited research base [11]. Several studies [11–13] have found that males have more restricted and repetitive behaviours than females. Some studies have found that females have more impairment in social reciprocity and communication than males, but these findings are not consistent [13]; others have found that females with ASD have better sociability skills than males with ASD [14].

Lately, the interest in studying ASD in genetic disorders has increased, and research shows a higher prevalence of ASD in some genetic disorders than in the general population [15]. The focus so far has been on the prevalence and phenomenology in different syndromes, and further studies are required to tell us more about the differences in ASD phenomenology between ASD in genetic syndromes and idiopathic autism. Using the Autism Screening Questionnaire (ASQ), Oliver et al. [16] found a high level of autism (> 45%) in individuals with Cornelia de Lange syndrome (CdLS) and fragile X syndrome (FXS) (only males with FXS participated in the study) but lower levels in individuals with cri du chat syndrome (CDCS), Angelman syndrome and Prader Willis syndrome (PWS) (< 20%). Individuals with Lowe syndrome and Smith-Magenis syndrome (SMS) were more in the middle with approximately 35% scoring above the cutoff for autism. No significant gender differences in any of the syndromes were found. Another study concerning tuberous sclerosis (TSC) found no significant differences between females and males regarding ASD [17]. Recently, Nærland et al. [18] published an article regarding gender differences in Down syndrome. The gender ratios in their sample were approximately 2M:1F, which is slightly less than expected in idiopathic ASD with the same degree of ID.

SMS is one of the rare disorders where ASD has been described as a prominent part of the disorder [19] but also a disorder where gender differences in ASD symptoms, favouring females, have been found. Laje et al. [19] found that females had higher T scores on the Social Responsiveness Scale (SRS) total and on the ‘Social cognition’ and ‘Autistic mannerisms’ subscales. In this study, they did not control for the gender differences already accounted for in the gender-specific norms. They did not find any significant gender differences in Social Communication Questionnaire (SCQ) scores [20].

SMS is a rare and complex genetic syndrome caused by an interstitial deletion of chromosome 17p11.2 [21] or a mutation on the retinoic acid induced 1 (RAI1) gene [22]. Most SMS patients have a deletion containing 76 genes [23], but the patients with mutations in the RAI1 gene also display most of the core features of SMS, which indicates that the RAI1 gene is a dosage-sensitive gene responsible for most of the symptoms in SMS patients [24]. The disorder is characterised by intellectual disability, multiple congenital anomalies, obesity, neurobehavioural abnormalities and a disrupted circadian sleep-wake pattern [25]. The incidence of SMS is estimated to range from 1:15,000–1:25,000 births [26]. Delayed diagnosis is common, although the use of array-CGH and SNP-array analyses in routine clinical practice, together with greater recognition of the syndrome in the last decade, has led to earlier diagnosis [27].

Children and adults with SMS appear to have unique neurobehavioural problems that are challenging for both parents and professionals. These problems include sleep disturbances, self-injurious and maladaptive behaviours, stereotypies, and sensory integration disorders [28]. A thorough investigation of aggressive behaviours of a cohort with SMS showed that self-injurious behaviour, physical aggression and destructive behaviour were all significantly more prevalent in persons with SMS compared with a cohort of persons with IDs of mixed aetiologies [29]. In this study, 96.9% of participants displayed self-injurious behaviour, 87.5% exhibited physical aggression, 81.3% showed destructive behaviour and 43.8% were verbally aggressive [29]. SRS scores consistent with ASD have also been identified in almost 90% of the investigated populations with SMS [19]. A progression of autistic-like behaviour has also been described in young children with SMS [20]. A study comparing several genetic disorders (PWS, FXS, CdLS, CDCS, etc.) found that persons with SMS scored higher (were more impaired) than PWS and CDCS in the social domain, but in the two other domains (communication and repetitive behaviour), they did not differ from the other groups [16].

In addition to the study by Laje [19] mentioned earlier, two other studies have looked at gender differences in SMS [19, 30]. In an animal model study, Huang et al. [24] found a sexually dimorphic phenotype regarding obesity in mice (females were significantly more obese than males) with loss of RAI1 functions, but due to high mortality, they did not investigate this any further and the cause of the sexually dimorphic phenotype is not clear. Edelman et al. [30] found some somatic differences between males and females such as myopia, cold hands and feet, eating/appetite problems and possible hypersensitivity (problems finding shoes to fit) in females; Edelman et al. also found that females had more frustration with communication than males.

Measuring ASD in genetic syndromes is fraught with some difficulties. Individuals with known genetic syndromes are usually excluded from the standardisation of ASD assessment tools, and it is known that degree of intellectual disability influences these tools [31]. Additionally, it is recently documented that the commonly used ASD assessment tools are highly influenced by parent-reported behavioural and emotional problems [32]. In the SMS population with its varying cognitive abilities and high rates of behavioural problems, it is therefore important to control for these factors when making claims about ASD symptomatology.

The main aim of this study was to investigate gender differences in rates and profile of ASD symptoms in SMS when controlling for rates of emotional and behaviour problems and adaptive behaviour as a proxy for developmental level. Based on previous research and our own clinical experiences, we hypothesised that the usual increased rate of ASD symptoms in males (the male bias) would be absent in a sample of individuals with SMS.

This study explored a number of ASD symptoms across gender in a Scandinavian SMS sample. The approximately three females per male above the SCQ cutoff is exactly the opposite of what we would expect to find in a sample of idiopathic ASD. It is particularly in the social domain of ASD that females with SMS differ substantially from females with other aetiological pathways to ASD.

The reversed gender ratio of ASD symptoms identified in this study cannot be explained by differences in neither developmental level nor in the amount of emotional and behavioural problems. The clinical diagnoses of intellectual disability differ between the genders, and we found a tendency for poorer development in females (VABS II total 53) than males (VABS II total 62), but this difference was not significant. In the regression model, the VABS II score did not have an independent contribution to the SCQ score. Emotional and behavioural problems, as measured with the DBC, did not differ between the sexes. In the regression model of the SRS, we found that DBC contributed in addition to gender. This probably indicates that the SRS is more sensitive to behavioural problems than the SCQ is [32, 45]. The SRS places a heavier emphasis on the reciprocal social interaction trait in ASD, whereas the SCQ places a similar emphasis on all three ASD domains [45].

Neither Oliver [16] nor Vignoli [17] found any significant gender differences in ASD symptomatology in other rare genetic syndromes such as cri du chat syndrome, Cornelia de Lange syndrome, Prader Willis syndrome or tuberous sclerosis complex.

We wanted to investigate whether a difference in ASD symptomatology could be the result of females having more severe phenotypes than males and if it could be linked to levels of ID or whether the emotional and behaviour problems in SMS affected gender differences. We found a strong tendency for lower degrees of ID in females than in males, but the difference was not significant. But as the difference is approaching significance (0.07), it would be interesting to investigate further if there could be a real gender difference in ID levels in SMS. As mentioned before, the accuracy of our ID levels is questionable and therefore not used to draw any conclusions. In general, administering formal psychometric assessments is often reported to be very difficult with people with SMS, due to the maladaptive behaviours, sleep disturbance and difficulties in expressive language skills [46].

The observed gender differences in ASD do not seem to be related to the main genetic mechanisms for SMS. The RAI1 mutations, associated with less severe SMS phenotype, were more frequent in females (3/20%) with more ASD symptoms than in males (0/0%) who had less ASD symptoms. The group of individuals with RAI1 mutations was too small to be tested as a separate subgroup in any of the analyses.

Current research suggest that female protective factors are more important than particular male-linked risk in explaining the male bias in ASD, but the mechanisms behind such female protection are not established [4, 6]. Whatever the female protective factor turns out to be, the current data suggest that it is not present in females with SMS.

We found three other papers presenting gender differences in SMS [19, 24, 30]. In the study from Edelman et al. [30], the authors found some gender differences, with the females showing more problems. Most of them were somatic (myopia, cold hands and feet, eating/appetite problems and possible hypersensitivity (problems finding shoes to fit)), but they also found that females had a significantly higher frustration with communication level. Neither of the questionnaires used in our study found a significant gender difference regarding communication, but a more thorough investigation of communication profiles in this syndrome would be beneficial both to investigate the gender difference more and to propose possible interventions. The study by Laje et al. [19] indicated an absence of the usual gender difference regarding ASD measured with SRS but not SCQ. In our study, we find gender differences both in the SRS and in the SCQ measure, both showing more problems among the females. It is particularly the social domain of ASD that has an unusual male/female ratio. Females with SMS have significantly more social problems than males. We did not find any difference in repetitive behaviour. Laje et al. [19] found a gender difference, favouring the females, in two subscales on the SRS but not in the total raw score or on the SCQ. In our study, we found a gender difference in ASD symptomatology, but neither in our study nor in the study by Laje et al. [19] could this difference be explained by differences in other traits in the syndrome. IQ level, adaptive behaviour and general emotional and behaviour problems have been investigated. A more thorough investigation of gender differences in adaptive behaviour profiles and the emotional and behaviour problems would be beneficial, alongside further molecular research regarding possible sexually dimorphic processes in SMS.

We found a clear reversed gender difference in the number of ASD symptoms in persons with SMS. This female bias in ASD symptoms is not explained by differences in the developmental level or the amount of emotional and behavioural problems. The deletion that is known to cause SMS is located on chromosome 17 (17p11.2), and there is no known reason to expect gender differences in any traits in this autosomal condition. The finding of a clear gender difference is therefore notable, and the mechanisms behind this require further study. A previous study found a sexually dimorphic phenotype in eating behaviour in mice with loss of RAI1 functions [24]. Whether this is related to our finding should be explored. Knowledge about the biological underpinnings of the reversed ASD gender ratio may be of relevance to understand gender differences in other biological pathways to ASD. The female protective factors assumed to explain the male bias in ASD seems to be lacking in SMS.