Wilson, A. C., King, J., & Bishop, D. V. M. (2019).
Before discussing the specific results found here, it is important to consider potential biases in the sample of children with sex chromosome trisomies. We distinguished a High Bias group who were identified during investigation for neurodevelopmental
or behavioural problems; these cases may be useful in revealing genetic or other factors that may be associated with variable outcomes, but they will inevitably over-estimate prevalence of problems. Our Low Bias group is more appropriate for establishing risk of various outcomes, but it is important to note that there still will be some bias in those identified on prenatal screening: mothers will be older than average, and will have elected to continue with the pregnancy.
Children were also required to assent to take part in the study, after viewing a video that provided information about what was involved. It is plausible that this would deter children with significant social anxiety. In addition, only 89 of 142 (62%)of participating families completed the DAWBA, and our analysis showed that the child’s language level was a factor affecting this, with parents of more mildly affected children being more likely to complete the interview. Factors such as these will inevitably affect who takes part in any study of children with genetic conditions – even the original newborn screening studies – as one cannot compel people to participate.
The best we can do in this situation is to be transparent about sources of bias, and consider how far results are consistent across different studies where specific influences on who volunteers may assume different importance. Bearing this in mind, we turn to consider the specific results obtained here, which lead to six broad conclusions.
1. Sex chromosome trisomy as risk factor for autism. The current results add to those of previous research (Cordeiro et al., 2012; Joseph et al., 2018; Ross et al., 2012; van Rijn
et al., 2014; van Rijn et al., 2008) in indicating that the rate of autism is increased in children with an extra sex chromosome. On the basis of previous results by Bishop et al. (2011), we predicted that elevated rates of autism would be found only in boys with XXY and XYY. However, our results were consistent with the more recent studies of van Rijn et al. (2014) and Wigby et al. (2016), who found increased levels of autism in girls with trisomy X. In the current sample, two of 25 girls with XXX in the Low Bias group, as well as one of four girls in the High Bias group met DSM-IV criteria for autism.
Furthermore, if we adopt the broader definition of PDDNOS, which includes cases with significant autism symptoms but falling short of autism (usually because of insufficient evidence of repetitive behaviours) then we can compare rates with the gender-specific population norms for PDD of Meltzer et al. (2000). Although those with autism or PDDNOS were a minority of children, and the sample was too small to give a confident estimate of prevalence, these results show that the increased prevalence of autism applies to all three trisomies. Note that a minority of children in the current study had taken part in the earlier study by Bishop et al. (2011), and all of those with autism in the current sample were new cases.
2. Sex chromosome trisomy as risk factor for social anxiety disorder. Very few children in this study met diagnostic criteria for DSMIV Social Phobia: all but one of those who did were cases of XXY. Only two children (both from the Low Risk XXY group, N = 14) had Social Phobia in the absence of any autistic features. While this represents a substantial increase over expected values (relative to Meltzer et al.’s epidemiological prevalence of 0.4% in boys), numbers are too small for reliable estimates. When we considered milder symptoms of social anxiety, using DAWBA bands, a rather different picture emerged, with 11/25 (44%) of XXX girls, 6/14 (43%) of XXY boys and 3/13 (23%) of XYY boys being placed in band 2 or over, compared with a prevalence of 3.4% in the B-CAMHS epidemiological sample. Thus all three types of sex chromosome trisomy confer a risk of increased social anxiety, though this generally falls short of meeting diagnostic criteria for Social Phobia. As noted above, however, a child with severe Social Phobia is unlikely to have agreed to participate in the study, as this involved an individual assessment with an unfamiliar adult.
3. Many children show no signs of either autism or social anxiety. We used parental responses to the SRS to capture mild, subclinical aspects of psychiatric problems, and it did reveal elevated levels of impairment, even when children with diagnoses of autism/PDDNOS or Social Phobia were excluded. However, the SRS also confirmed the wide range of outcomes in all three trisomies, with around half the children in the Low Bias group scoring within the normal range.
4. The effects of ascertainment bias are substantial. As anticipated, children from the High Bias group had more severe problems than those from the Low Bias group. Although most studies that have looked at this question have found similar patterns, this is not always the case (cf. (Van Rijn et al., 2014)). The method for dividing cases into Low Bias and High Bias subgroups may affect results: if the latter group is defined purely in terms of late diagnosis, then it may include some whose trisomy was discovered in the course of medical investigations, rather than, as in our sample, being restricted to those where behavioural or cognitive impairments had prompted genetic investigation
5. More similarities than differences in the impact of an extra X or extra Y chromosome The notion that supplementary X and Y chromosomes might have different effects is attractive as well as plausible (Green et al., 2019; Skuse, 2018), particularly since it has potential to throw light on sex differences in psychiatric disorders in general. However, data from the current study indicate that insofar as such effects occur, they can be swamped by other influences, leading to a highly heterogeneous picture. We should not dismiss the idea that there could be complementary effects (more social anxiety with an extra X and more autism with an extra Y) on the basis of data from a single study. Our sample sizes were small, especially when attention was restricted to the Low Bias group, so we lacked statistical power to detect any but large associations. Insofar as there were trends in the data, they agreed with the prediction of higher levels of social anxiety in boys and girls with an extra X chromosome, and higher levels of autism and reduced prosocial behaviour in boys with an extra Y. But we can conclude that, insofar as such effects occur, risks for specific disorders are probabilistic rather than deterministic. A simple model that treats social anxiety and lack of social awareness as points at either end of a single dimension is also complicated by the finding of cases who met diagnostic criteria for both Social Phobia and autism/PDDNOS.
6. Variability in outcomes suggests that an extra sex chromosome creates general developmental instability. The more we discover about genetic causes of neuro -developmental disorders, the more we are confronted with the fact that the same aetiology can lead to very variable outcomes. This is true for a wide range of genetic conditions that involve copy number variants or point mutations, as well as chromosome aneuploidy (Mitchell, 2015). Conditions such as tuberous sclerosis, 22q11.2 deletion syndrome, 15q11.2 microdeletion syndrome are well-established as risk factors for neurodevelopmental disorders, but the clinical manifestations can vary widely from person to person, even within a family. The phenotypic outcomes of these conditions do not respect traditional diagnostic boundaries: the same genetic aetiology can lead to autism, schizophrenia, epilepsy or ADHD. It is often assumed that this variability is due either to interactions with the genetic background or to environmental factors that moderate gene effects.
This kind of explanation cannot, however, explain why in twin studies, one can often see marked phenotypic variability in concordant MZ twins: this has been shown by Le Couteur et al. (1996) for autism and by Corey et al. (2011) for epilepsy. Given that MZ twins have identical genes and closely similar environments, we need another explanation for phenotypic variability. One candidate is stochastic variation (Molenaar et al., 1993; Wolf, 1997). Purely random events early in neurodevelopment can have cascading effects so that quite minor differences in the prenatal brain lead to long-lasting differences in function. This alone, however, does not explain why there should be higher rates of impairment in those with chromosome trisomies. To account for that, we need to introduce the idea of developmental instability, i.e., a disturbance of the usual homeostatic processes that make the organisation robust to external influences (Nijhout & Davidowitz, 2003). This would mean that small stochastic effects would have larger impact than would otherwise be the case.
Evidence that aneuploidy may be linked to developmental instability was obtained by Beach et al. (2017), who found high variability in cell cycle progression among yeast cells with identical aneuploidies. Furthermore, these cells showed variable response to environmental stress. These authors went on to demonstrate wide phenotypic variation in inbred mice with genetically engineered trisomy 19, despite a uniform genetic background and environment. Mitchell (2018) proposed that a wide range of genetically-based neurodevelopmental disorders may involve a process of developmental instability. This can make sense of the wide range of outcomes seen with an extra sex chromosome – both in terms of the severity of impairments, and in terms of the wide range in symptoms that are observed. If the main effect of an additional chromosome is to increase neurodevelopmental instability, then the developing brain will be less robust to perturbations, so that small differences in early neurodevelopmental processes will have cascading effects that can lead to very different clinical pictures.
When a sex chromosome trisomy is identified prenatally, it is important that parents are given accurate information about the implications for the child’s development; the quality of information provided is one factor that influences whether the mother continues with the pregnancy (Jeon et al., 2012). It is crucial that parents are provided with information from unbiased samples: The recommendations of Linden et al. (2002) are generally still valid: as they noted, children with sex chromosome trisomies ‘are at an increased risk for developmental problems but… most are in the normal range of development, and marked abnormality is not usually seen’ (p. 4). However, at the time they were writing, there was no awareness of the risk of autism, possibly because the diagnostic criteria for autism were far more stringent at the time when the neonatal screening studies were conducted (Bishop et al., 2011). The companion paper on this sample confirms the recommendations that parents should be informed that presence of a sex chromosome trisomy raises the probability that the child will have educational difficulties (Bishop et al., 2018), but the current analysis indicates that in addition, although the majority of children will not merit a clinical diagnosis for autism and/or Social Phobia, the likelihood of these conditions is increased.
Advice on management of children with a sex chromosome trisomy has mostly focused on the provision of speech and language therapy and educational interventions to address problems with speech, language and educational progress that may arise. The current study emphasises that symptoms of social anxiety are common in all three trisomies, suggesting that early involvement of Child and Adolescent Mental Health Services may be warranted in some cases. The outcomes of children with sex chromosome trisomies are very varied, making it difficult to predict in advance the exact type of support that will be needed, but it is clear that problems often do not fit neatly into single diagnostic categories, and optimal support may require collaboration between different services.
© 2019 Meyer J. This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium provided the original work is properly cited.