Divergent structural brain abnormalities between different genetic subtypes of children with Prader–Willi syndrome

Prader–Willi syndrome (PWS) is a rare and poorly understood neurodevelopmental disorder that affects 1 in 15,000 to 20,000 live births [1]. PWS is characterized by pre- and postnatal hypotonia, endocrine problems, hyperphagia, temper tantrums and repetitive behavior, skin picking, dysmorphic facial features, high pain threshold, and developmental delays [2]. In addition, individuals with PWS carry a high risk of psychiatric illness, such as psychotic, obsessive-compulsive disorder (OCD), and autism spectrum disorders (ASD).

The underlying cause of PWS is the loss of function of paternally expressed genes on the long arm of chromosome 15q11-q13 due to a de novo deletion (DEL, approximately 70% of the cases [3]), maternal uniparental disomy (mUPD, 25% of the cases [4]), and unbalanced translocation or imprinting center defects (<5% of cases [5]).

Lifetime prevalence of psychotic illness in individuals with PWS is reported to be up to 60% in individuals with mUPD and up to 20% in individuals with DEL [1], which is at least 18 times higher than that in the general population [6]. In addition, the mUPD patients are more likely to have a more severe course of the psychiatric illness, a higher frequency of relapse, and a poorer response to medication [7]. The underlying neurobiology that places them at-risk is yet unknown. Recently, copy-number variations (CNVs) at the 15q11.2 locus were found to be associated with schizophrenia and related psychosis [8], which suggests common genetic pathways between schizophrenia, psychotic disorders, and PWS.

In addition, individuals with PWS have a higher prevalence of ASD traits, such as impaired social functioning, rigidity, resistance to change, and repetitive, stereotypic behavior [9]. These autistic traits are more common in individuals with mUPD than in those with DEL (40% versus 20%, respectively). The autistic symptoms found in PWS are between 20 to 40 times higher than the 1% reported in the general population [10]. Chromosomal rearrangements of 15q11-13 locus have been found in patients with autism [11], suggesting that defective 15q11-13 genes might underlie the ASD traits in individuals with PWS. Further, recent findings indicate a plausible common neurodevelopmental pathway of autism, schizophrenia, and bipolar disorder, as family history of schizophrenia and bipolar disorder increased risk of autism in other family members [12], and several CNVs in genome-wide association studies (GWASs) were associated with developing either of these disorders [13].

Key symptoms indicate that the central nervous system (CNS) is adversely affected in PWS. To date, only two quantitative structural magnetic resonance imaging (MRI) studies have been performed in adults with PWS, which reported smaller grey matter volumes in the frontal, temporal, and parietal [14, 15] lobes, and smaller white matter volumes in the frontal and temporal cortices, brainstem, and cerebellum [15]. An MRI qualitative study reported enlarged ventricles, decreased parieto-occipital lobar volume, and sylvian fissure abnormalities in adults with PWS [16]. However, no quantitative structural or functional MRI studies have been performed in children with PWS, and there are no studies evaluating differences between children with DEL and mUPD. Since the DEL and mUPD subtypes of PWS differ in their risk of developing severe psychiatric disorders, a better understanding of the neurobiology of these subtypes will shed light on at-risk states.

In the current study, we utilized high resolution MRI to investigate the brain morphology in children with two subtypes of PWS (DEL and mUPD) in order to gain insights into the brain structure of young children with PWS as compared with their healthy siblings. We hypothesized that children with PWS, when compared with healthy siblings, will show global deficits that will be reminiscent of those reported in the adults with PWS. Specifically, we expected smaller overall grey and white matter volume and larger lateral ventricular volume in children with PWS when compared with healthy controls. Since clinical data show that children with mUPD have greater social and cognitive deficits than children with DEL, we expected that the brain differences would be more pronounced in children with mUPD than in those with DEL. Furthermore, since children with mUPD show an increased risk of psychiatric disorders over children with DEL, we were also interested in brain differences which may indicate an increased risk of developing a severe psychiatric disorder.

The study population consisted of 25 children with PWS who participated in the Dutch PWS Cohort Study [17]. Patients fulfilled the following criteria: 1) genetically confirmed PWS subtype; 2) age 6 to 18 years old; and 3) no neurological or psychiatric history.

Eleven age- and gender-matched, typically developing siblings were included as a control group fulfilling the following inclusion criteria: 1) age 6 to 18 years old; and 2) no neurological or psychiatric history.

This study was approved by the Medical Ethical Committee of the Erasmus Medical Center Rotterdam, Rotterdam, The Netherlands. Written informed consent was obtained in all cases from the caregivers and children older than 12 years, and assent of children younger than 12 years.

Clinical data are presented in Table 1. No significant differences were found in age, handedness, and gender distribution among groups, and no differences in IQ scores and age at start of growth hormone (GH) treatment. Only two children in the mUPD group were diagnosed with ASD. None of the children had a history of treatment with psychotropic medication.

Our study is the first to investigate brain morphological differences in children with PWS using high resolution MRI techniques, and the first study to investigate differences between children with DEL and mUPD. We found that children with both DEL and mUPD had smaller brainstem volumes, and a trend towards smaller white matter volume and total cortical pial surface area compared with healthy controls, indicating early deviations in prenatal brain development in children with PWS. However, there were also differences between children with DEL and those with mUPD. Children with DEL had an overall smaller, but proportionately developed brain and normal cortical thickness, while children with mUPD had a significantly increased surface CSF and pronounced enlargement of the lateral ventricles compared to healthy controls. Further, exploratory cortical focal analyses revealed a significantly smaller white matter volume in children with mUPD in areas implicated in default mode network (DMN) and cognitive decision-making. Our results suggest that although both PWS groups show global brain deficits, children with mUPD are more severely affected.

Prenatal brain development is a well-orchestrated series of events consisting of neuronal proliferation and migration, establishment of synaptic connections, myelination, and elimination of ineffective synapses (pruning) [31]. Different regions of the brain develop at different time points and pace, which in turn provides different time windows for vulnerability to perturbations, but also for effective intervention. For instance, development of brainstem is finalized during the first trimester, while development of cerebellum is initialized during the first trimester, but extends into the first year after birth [31]. Myelination is most vulnerable to insults during the late gestational period, but it continues after birth as well and extends through adolescence and into early adulthood [32]. By investigating both global and local neuroanatomical measures in children with PWS, inferences can therefore be drawn about a plausible course of neurodevelopment and the phenotype of PWS.

Both children with DEL and mUPD had smaller brainstem volume. The brainstem is responsible for several basal bodily functions, such as pain perception, respiratory regulation [33], and sleep cycle [34], all of which have been reported to be impaired in PWS. Central sleep apneas are common in individuals with PWS and pose an increased mortality risk at a young age [35]. A higher pain threshold has been reported in children with PWS [2], as well as disturbed sleep cycle and rapid eye movement (REM) sleep phase abnormalities [36], supporting our findings of possible brainstem dysfunction in PWS. Mice lacking necdin (in PWS locus) had smaller medullar nuclei and a disturbed respiratory rhythmogenesis [37], further suggesting that PWS region is involved in the neurodevelopment of the brainstem and the respiratory problems in patients with PWS.

Regardless of the genetic subtype, children with PWS had a trend towards smaller white matter volume, indicating reduced structural connectivity or aberrant myelination in children with PWS. Children with developmental delay showed delayed myelination [38]. The smaller white matter volume in children with PWS may therefore reflect delay in brain maturation and might underlie cognitive deficits in these children. Important to note is that several endocrine factors, namely insulin-like growth factor I (IGF-I) and thyroid hormones (T3 and T4), which are important neurochemicals involving brain and axonal growth and myelination [39, 40], are impaired in children with PWS, unless treated. IGF-I levels are very low in most children with PWS prior to GH treatment [17]. Low free T4 levels were reported in children with PWS as well [41]. It is therefore plausible that lower levels of thyroid hormones and IGF-I had adverse effects on brain growth and myelination.

Children with mUPD had enlarged ventricles, together with increased surface CSF volume. In adult patients with PWS, enlarged ventricles have been reported [16]. Our results suggest that ventricular enlargement occurs early in life in individuals with PWS. We do not know when during neurodevelopment the enlargement of the ventricles took place. Lateral ventricles are large at the 17th gestational week (GW), and decrease in size between 18th and 24th GW, due to multiple factors, including the increasing thickness of the brain parenchyma, cortical gyrification, and the formation of the basal ganglia [42]. In the current study, the basal ganglia were enlarged relative to the intracranial volume, which may indicate that ventricular enlargement did not occur at the cost of the basal ganglia, but more likely at the cost of the cortical volumes. Enlarged ventricles, together with increased surface CSF, probably due to the widening of the sulci, indicate disturbances in early prenatal gyrification, dendritic arborization, or postnatal neuronal atrophy in these children. Enlarged ventricles are widely reported in patients with schizophrenia, both chronic [43] and first onset medication-naive patients with schizophrenia [44], and their first-degree unaffected relatives [45]. Furthermore, enlarged lateral ventricles were also found in young children with a 22q11.2 deletion who are at very high risk of schizophrenia [46] and in adolescents with a psychotic bipolar disorder [47]. Given that children with mUPD have an elevated risk of psychotic illness, our findings suggest that ventricular enlargement may be part of a predisposition for psychotic disease.

Children with DEL showed less enlargement of the lateral ventricles and no increase in surface CSF, suggesting that brain atrophy is less pronounced in these children compared to children with mUPD. Since ventricular enlargement has been reported in adults with DEL [16], our findings suggest that children with DEL are likely to develop ventricular enlargement later in life compared to children with mUPD.

Both children with DEL and mUPD showed a trend towards smaller cortical surface area, indicating impaired prenatal brain growth [31]. The development of cortical surface area is determined by the symmetric cell division in the neural tube during the first 6 GW, and by the pronounced growth and gyrification that occurs during the third trimester [42]. Environmental insults during the first 6 GW usually results in drastic reductions of cortical surface area [31]. It is therefore plausible that the observed mild reduction in cortical surface area may be the result of deviations in the gyrification processes during the third trimester.

Interestingly, children with DEL show a different pattern compared to children with mUPD. Smaller cortical and subcortical grey matter, brainstem, and cerebellum volumes were found, but cortical thickness was normal compared to healthy controls. The early development of cortical thickness is determined primarily through neuronal migration, which takes place between 6th and 24th GW [48]. Furthermore, from the perspective of phylogeny, cortical thickness is much more preserved compared to measures, such as brain volume and surface area [31]. Thus, normal cortical thickness may show that the genetic mechanisms involved in neuronal migration are not altered in children with DEL.

In contrast, children with mUPD showed a trend towards increased cortical thickness, which may indicate alterations in neuronal migration or impaired elimination of ineffective synapses (pruning). Increased cortical thickness has been reported in children with autism [49]. Knowing that ASD traits are very common in children with mUPD, impaired pruning might therefore affect integrative processing, complex executive, and social functions [49], and might underlie ASD symptoms in children with mUPD.

Holland et al. [50] proposed that developmental arrest might underlie the core PWS phenotype. In our study, both children with DEL and mUPD showed signs of impaired brain growth, however it was more pronounced in children with DEL. Furthermore, while children with DEL show signs of arrested, but not necessarily deviant, brain development, children with mUPD showed a plausible divergence in the neurodevelopmental trajectory. However, longitudinal studies are necessary to confirm the differences in neurodevelopmental trajectories between children with DEL and mUPD. An important genetic difference between individuals with DEL and mUPD is the overexpression of maternally imprinted UBE3A gene in brains of patients with mUPD [51]. UBE3A plays an important role in dendritic tree formation [52], and both knockout and overexpression of the UBE3A gene results in reduced growth and branching of the dendrites [52]. UBE3A regulates synapse development and is highly expressed during a novel learning situation [53]. Overexpression of UBE3A in mUPD and resulting failure to develop new synapses might be underlying the observed diversion, and possibly the cortical atrophy and ventricular enlargement in children with mUPD.

Exploratory cortical focal analyses revealed smaller white matter volume in the right caudal anterior and posterior cingulate cortex, left superior frontal gyrus, and bilateral precuneus in children with mUPD compared to those with DEL and to healthy controls. These areas are associated with cognitive control, moral decision-making, and emotion cognition [54], and are implicated in DMN [55, 56]. The DMN is a functional brain network that is activated in the absence of cognitive tasks [56], and is thought to reflect self-oriented and social cognitive processes [57]. DMN dysfunction is associated with multiple brain disorders, such as attention deficit and hyperactivity disorder, autism, and schizophrenia [56, 58‐60]. Interestingly, electrophysiological correlates of decision-making was found diminished in adults with mUPD, but not in those with DEL [61], and individuals with mUPD had more autistic-like symptoms on the social interaction scale [62]. Impairment of task-switching was found in individuals with PWS, although genetic subtypes were not reported [63]. While this should be confirmed by functional imaging studies, it is plausible that aberrant connections within these brain areas underlie social cognitive decision-making impairment in individuals with mUPD. Further, white matter volume in right pars triangularis and bilateral pars opercularis (inferior frontal gyrus (IFG)) was smaller in children with mUPD. Interestingly, lower functional connectivity in the IFG has been described in patients with schizophrenia and in individuals at ultra-high risk of developing psychosis [64], suggesting that IFG might be involved in the etiology of psychotic illness.

These results are limited by the small sample size, thus generalization to broader PWS population should be undertaken with great caution. Further, the control group consists of age- and gender-matched healthy siblings. A possible concern of recruiting siblings as the control group is that volumes of most brain structures are heritable [65]. However, as PWS occurs due to a de novo genetic event during conception, we assume that unaffected siblings are representative of a random sample of the general population. The great advantage of having siblings as the control group is that the effects of possible environmental and hereditary factors on brain development are greatly reduced, and that the observed significant differences are more likely PWS-specific.

All children with PWS were treated with GH (1 mg/m² per day) at the time of the study; therefore, our findings are confounded by GH treatment. However, as it is known that GH and IGF-I increase brain growth, myelination, and has neuroprotective properties [39] we could speculate that if the GH treatment had any effect of the brain, it would have a positive effect in terms of brain normalization.

Our findings provide preliminary insights into the brain anatomy of children with PWS. All children with PWS showed impaired brain growth. Children with mUPD showed signs of early brain atrophy and a trend towards increased cortical thickness. In contrast, children with DEL showed signs of fundamentally arrested, although not deviant brain development. The findings in children with mUPD are reminiscent of those in schizophrenia or autism, which confirms the clinical data of increased risk of ASD and psychotic illness in individuals with mUPD. The reported brain abnormalities are likely to precede psychiatric illness as none of the children were under psychiatrist treatment at the time of the study. Our results suggest that children with mUPD have a fundamentally different brain structure and divergent developmental trajectories compared with children with DEL and healthy controls.