TSC-associated neuropsychiatric disorders (TAND): findings from the TOSCA natural history study

Background Most evidence for TSC-associated neuropsychiatric disorders (TAND) to date have come from small studies and case reports, and very little is known about TAND in adults. We explored baseline TAND data from the large-scale international TOSCA natural history study to compare childhood and adult patterns, describe age-based patterns, and explore genotype-TAND correlations. Results The study enrolled 2216 eligible participants with TSC from 170 sites across 31 countries at the data cut-off for the third interim analysis (data cut-off date: September 30, 2015). The most common behavioural problems (reported in > 10% of participants) were overactivity, sleep difficulties, impulsivity, anxiety, mood swings, severe aggression, depressed mood, self-injury, and obsessions. Psychiatric disorders included autism spectrum disorder (ASD, 21.1%), attention deficit hyperactivity disorder (ADHD, 19.1%), anxiety disorder (9.7%), and depressive disorder (6.1%). Intelligence quotient (IQ) scores were available for 885 participants. Of these, 44.4% had normal IQ, while mild, moderate, severe, and profound degrees of intellectual disability (ID) were observed in 28.1, 15.1, 9.3, and 3.1%, respectively. Academic difficulties were identified in 58.6% of participants, and neuropsychological deficits (performance <5th percentile) in 55.7%. Significantly higher rates of overactivity and impulsivity were observed in children and higher rates of anxiety, depressed mood, mood swings, obsessions, psychosis and hallucinations were observed in adults. Genotype-TAND correlations showed a higher frequency of self-injury, ASD, academic difficulties and neuropsychological deficits in TSC2. Those with no mutations identified (NMI) showed a mixed pattern of TAND manifestations. Children and those with TSC2 had significantly higher rates of intellectual disability, suggesting that age and genotype comparisons should be interpreted with caution. Conclusions These results emphasize the magnitude of TAND in TSC and the importance of evaluating for neuropsychiatric comorbidity in all children and adults with TSC, across TSC1 and TSC2 genotypes, as well as in those with no mutations identified. However, the high rates of unreported or missing TAND data in this study underline the fact that, even in expert centres, TAND remains underdiagnosed and potentially undertreated.


Background
Tuberous sclerosis complex (TSC) is an autosomal dominant genetic disorder characterized by the formation of hamartomas in multiple organ systems with a wide diversity of symptoms and severity across individuals [1,2]. The majority of individuals with TSC have central nervous system involvement with a wide range of structural manifestations, such as cortical tubers, subependymal nodules, and aberrations of gray and white matter as well as high rates of functional manifestations, such as epilepsy, intellectual disability (ID), and behavioural problems [3][4][5]. The neurological, neuropsychiatric, and renal manifestations represent the greatest burden of disease of all TSC-related manifestations [3].
TSC-associated neuropsychiatric disorders (TAND) is an umbrella term coined by the Neuropsychiatry Panel of the 2012 International Consensus Conference for TSC, and encompasses a range of neuropsychiatric manifestations across various levels of investigation [2,3,5]. These include the behavioural level (observed behaviours such as sleep problems or aggressive behaviours), the psychiatric level (DSM/ICD defined psychiatric disorders such as autism spectrum disorders [ASD] or attention deficit hyperactivity disorder [ADHD]), the intellectual level (intellectual ability as defined by intelligence quotient [IQ]-type tests), the academic level (learning disorders, e.g., reading or mathematics difficulties), and the psychosocial level (e.g., self-esteem, family difficulties) [3,5,6]. TAND represent a significant burden of disease and have a major impact on quality of life of individuals with TSC and their families, given their impact on education, employment, and social life of patients and their family [3,5].
The rate and pattern of TAND vary greatly among patients with TSC [3,5,[7][8][9][10][11]. Overall, about 90% of individuals with TSC exhibit TAND features to some extent during their lifetime, with ASD and ID reported in up to 50% of individuals [3,5,6]. Results show differential rates in those with and without ID [12]. However, even individuals with normal intellectual abilities are at risk of developing TAND manifestations, particularly in the academic, neuropsychological, and psychosocial domains [8]. Despite the high rates of TAND, a 2010 survey of members of the Tuberous Sclerosis Association in the United Kingdom indicated that only 20% of individuals with TSC had ever received any assessment or treatment for TAND, suggesting a large assessment and treatment gap [5,6,12,13]. It is important to acknowledge that the majority of evidence for rates and patterns of TAND to date have come from relatively small studies and case reports. The largest published studies to date have included a few hundred participants [10,12,14]. Furthermore, very little TAND data are available about adults with TSC [15] or about the age-based pattern of TAND.
The underlying aetiology of TAND is likely to be combinatorial and multifactorial [16]. There is evidence that the genetic aberrations of TSC may be sufficient to cause TAND manifestations, [16,17] with combinatorial and interactive contribution from functional and structural factors [16][17][18][19][20][21][22]. Individuals with TSC2 mutations have been reported to have a greater likelihood of ID than those with TSC1, [23][24][25][26] but both TSC1 and TSC2 mutations have been associated with the full range of intellectual ability from high IQ to profound ID [25,26]. Individuals with no mutation identified (NMI) after genetic testing have typically been described to have ID profiles between those with TSC1 and TSC2 [23][24][25][26]. These studies have also been based on relatively small samples and no study to date has examined other aspects of TAND in relation to genotype.
The largest natural history study of TSC to datethe TOSCA (TuberOus SClerosis registry to increase disease Awareness) study is a multi-centre, international disease registry designed with the aim of providing deeper insights into the manifestations of TSC and its management [27,28]. In a previous publication outlining baseline findings from the TOSCA cohort of 2093 individuals, we presented topline findings of TAND features in the largest TSC cohort reported globally to date [28]. Results showed that ID was observed in 54% of the evaluated participants and suggested that psychiatric disorders were typically diagnosed late. We also identified significant non-reported or missing data, which suggested that even in expert TSC centres around the globe, TAND may be underdiagnosed and therefore under-treated [28].
Here, we performed a detailed exploration of the largest TAND dataset to date using TOSCA baseline data of all the patients enrolled at the cut-off of September 30, 2015, with the specific aim of comparing childhood and adult TAND profiles, describing the age-related pattern of TAND, and examining TAND in relation to genotype.

Methods
A detailed description of the methods of the TOSCA registry has been provided previously [27]. In short, individuals of any age with a documented visit for TSC before 12 months of enrollment or newly diagnosed with TSC were included in the study between August 2012 and August 2014.
Information on participant demography, family history, genotype, vital signs, prenatal history, clinical features of TSC across all organ systems, comorbidities, and rare manifestations were collected, both retrospectively and prospectively at baseline and annually thereafter for up to 5 years. Given that this was a natural history study, participants were followed up based on clinical need, and no clinical, laboratory or formal TAND evaluations were mandated by the protocol. The terms and operationalization of TAND manifestations and 'levels' were defined as in the primary TAND publication [5].
The TOSCA registry was designed and conducted in accordance with the Guidelines for Good Clinical Practice and ethical principles outlined in the Declaration of Helsinki [29,30]. Written informed consent, with prior endorsement by all local institutional review boards (human research ethics committees) was obtained from all participants, parents, or guardians prior to registry enrollment.
For the purposes of this manuscript, descriptive statistics were used to summarize TAND data. Frequency of TAND features were extracted and presented as percentage of individuals with available data (excluding non-reported or missing data). Intellectual ability was categorized as normal (IQ > 70), mild ID (IQ 51-70), moderate ID (IQ 36-50), severe ID (IQ 20-35), and profound ID (IQ < 20), according to DSM-5/ICD-10 [31,32]. Psychiatric disorders were defined according to the DSM-5/ICD-10. Age-based patterns of TAND (children vs adults) and the association between TAND and genotype (TSC1 vs TSC2) were analyzed using Chi-square test. The TOSCA study included data collection on all levels of TAND as outlined above and summarised in the primary TAND manuscript [5], apart from psychosocial characteristics, such as self-esteem, sibling, or family stressors, which were only included after the TOSCA database was set up.

Results
A total of 2216 participants (1154 females and 1062 males) with TSC were enrolled into the TOSCA registry from 170 sites across 31 countries at the data cut-off for the third interim analysis (September 30, 2015). The median age of the TOSCA cohort was 13 years (range, < 1-71) with 1410 individuals (63.6%) aged ≤18 years.
Overall TAND features Table 1 and Fig. 1 represent the frequencies of TAND features in the overall TOSCA cohort.

TAND features in children vs adults
Some differences were observed in the frequencies of TAND features between children (aged ≤18 years), and adults (aged > 18 years) ( Table 2). At the behavioural level, the rates of overactivity and impulsivity were significantly higher for children than adults (54.8% vs 21.4% and 46.7% vs 33.2% respectively, p < 0.001), while rates of anxiety, mood swings, depressed mood, obsessions, psychosis and hallucinations were significantly higher in adults than children (50.9% vs 25.8%; 40.8% vs 25.2%; 43.9% vs 8.2%; 19.2% vs 11.8%; 11.3% vs 3%; 10.3% vs 0.6%, respectively, P < 0.001 for all except obsessions, P < 0.01). Interestingly, some behavioural manifestations showed similar rates in children and adults. For instance, sleep difficulties and severe aggression were very similar between children and adults ( Table 2). At the psychiatric level, ASD and ADHD were reported at higher rates in children than in adults (23.1% vs 16.1%; 22.4% vs 10.5%; p = 0.0029 and P < 0.001 respectively), while rates of anxiety disorder and depressive disorder were higher in adults than in children (16.8% vs 7%; 16.3% vs 2.1% respectively; P < 0.001). We observed no major differences in the rates of academic difficulties and neuropsychological skills between children and adults (Table 2). However, highly significant difference was observed between the rates of intellectual disability between children and adults ( Table 2).

TAND features by age bands
Frequencies of all the TAND features by age bands at consent are depicted in Table 3. The majority of TAND behavioural characteristics showed varying frequencies across age bands, but anxiety and depressed mood showed a clear pattern of increased frequency across increasing age bands. Interestingly, the rates of diagnoses of anxiety and depressive disorder showed a pattern that was not entirely consistent with the behavioural observations for anxiety and depressed mood. Severe and profound ID was reported at low rates in the youngest age group (3%), increasing in the under 10-year olds (6.6 and 9.7%), and ranging between 12.7-19.3% in the older age bands.

TAND and genotype
Molecular testing for genetic mutations was performed in 1000 participants (45.1%). Of them, 197 had TSC1 mutations, 644 had TSC2 mutations, and 144 had no mutation identified (NMI). At the behavioural level, TSC2 mutations were associated only with significantly  Table 4). At the psychiatric level, ASD was observed at significantly higher frequency in participants with TSC2 than those with TSC1 mutations (28.6% vs 12.2%, P < 0.001). ADHD, anxiety disorder and depressive disorder were not significantly different between the two genotypes, but it was interesting to observe that all three showed higher absolute frequencies in association with TSC1 rather than TSC2 (ADHD TSC1 = 17.6%; TSC2 = 16%, P = 0.6881; Anxiety disorder TSC1 = 10.1%; TSC2 = 8.6%; P = 0.7809; Depressive disorders TSC1 = 10%; TSC2 = 5.2%; P = 0.0509) ( Table 4). The frequencies in behavioural and psychiatric manifestations of individuals with NMI ranged sometimes between and sometimes higher or lower than TSC1 and TSC2 (Table 4).
Academic/scholastic difficulties were more common in individuals with TSC2 mutation than those with TSC1 mutation (63.5% vs 49.2%; P = 0.0051, Table 4). More individuals with TSC2 mutation had neuropsychological performance scores falling below the 5th percentile compared to those with TSC1 mutation (63% vs 38.8%, P = 0.0024). Individuals with NMI showed IQ, academic and neuropsychological profiles in between the frequencies of the group with TSC1 and TSC2.

Discussion
The TOSCA registry allowed exploration of the frequency of a wide range of TAND features in the largest cohort of TSC reported to date. We set out to examine the overall TAND profile, to compare childhood and adult patterns, age-based patterns, and genotype-TAND correlations. Results showed lower rates of behavioural and psychiatric disorders than previously reported, but similar rates of ID [3,6,8,12]. The very high rates of non-reported or missing data in this study (in excess of 60% for behavioural and intellectual levels) may, at least in part, have contributed to the lower rates observed. The rates of academic difficulties and neuropsychological deficits, reported in this study for the first time, were very high and suggested that between half and two-thirds of individuals with TSC will have difficulties in these two TAND levels. In spite of relatively similar rates of ID between children and adults, we observed a pattern of higher overactivity and impulsivity in children, and higher rates of anxiety and depressed mood in adults. Interestingly, some TAND characteristics such as sleep problems and severe aggression remained high across all age bands, suggesting persistence of these difficulties across the lifespan. With regards to genotype-TAND correlations, we observed a genotype-intellectual phenotype correlation and a higher frequency of ASD in association with TSC2, similar to previous reports [8,26]. Interestingly, fewer other TAND-genotype correlations were observed. Self-injury, ASD, academic difficulties and neuropsychological deficits were the only other significant correlations with TSC2.
Given that all of these are known to be strongly correlated with intellectual level [12], and given the differences observed here in IQ groups, these results have to be treated with caution and should be explored in relation to matched or stratified IQ groups in future studies. We also observed a potential pattern of more depressed and anxious mood, and higher rates of anxiety and depressive disorders in association with TSC1 mutations. This was a novel observation not previously reported in the literature, but the same caveat in relation to IQ as potential confounder requires exploration in future studies. The TAND profile of individuals with NMI was also a novel finding. Intellectual, academic and neuropsychological profiles seemed to fall between the frequencies of those with TSC1 and TSC2 mutations, mapping onto intellectual findings previously reported [8,26].
However, the same pattern was not seen in behavioural and psychiatric manifestations. About 7% of individuals in the general population are expected to have clinically significant behavioural problems [33]. However, much higher rates of behavioural difficulties are reported in patients with TSC [7]. In a  [6]. The findings in this large-scale international study confirm the high rates of a wide range of behavioural TSC1 vs TSC2 f Chi-square test showing association between children and adults across all levels of IQ problems in TSC. The comparison between children and adults showed a pattern of lower overactive/impulsive behaviours in adults but higher anxiety or depressed mood. These findings map well onto typical age-based expectations in psychopathology [8,9,12]. However, our results also highlighted the fact that behavioural difficulties occurred across all ages in individuals with TSC. These findings underline the importance of expecting and evaluating for a changing profile of TAND difficulties from childhood into adulthood, as recommended in TAND assessment guidelines [2,34,35].
Among the psychiatric problems associated with TSC, ASD, and ADHD are the most common neurodevelopmental disorders in children, and anxiety/mood disorders the most common in adults [8]. The variable rates of ASD (17-68%) and ADHD (30-60%) reported in earlier studies can be understood based on different study methodologies, diagnostic criteria used, and level of intellectual ability of participants [7,8,36]. In this study, ASD and ADHD were reported in 21.1% and 19.1% of participants, respectively. As highlighted in our baseline paper [27] the diagnosis of ASD, in particular, was made very late (mean age, 7.8 years; median age, 5 years; range, 0-38 years) [28]. In spite of very high rates of anxiety and depressed mood symptoms in the cohort, the rates of diagnoses of anxiety disorders or depressive disorders were surprisingly low in our study (anxiety disorder, 9.7%; depressive disorder, 6.1%). Taking together the high rates of non-reported or missing data, late ages of diagnoses and frequencies observed in our cohort, we suggest that psychiatric disorders are underdiagnosed and potentially diagnosed late in individuals with TSC. The low rates of ASD and ADHD observed in adults here, may suggest a cohort effect where adults were not assessed for developmental disorders in the past decades.
In line with previous reports, [8,26,37] the TOSCA registry showed a genotype-intellectual phenotype pattern suggesting a greater likelihood of ID in participants with TSC2 than those with TSC1 mutations. However, it was also important to note that only 66.7% of those with TSC1 had normal intellectual ability, suggesting that a third of individuals with TSC1 may have ID. Similarly, even though TSC2 mutations were more likely to be associated with ID, 42% of all individuals with TSC2 mutations had normal intellectual ability. Our findings therefore reinforce the message that genotype (TSC1 vs TSC2) is not a clinically helpful predictor of intellectual ability at an individual level [26].
The differences between those with TSC1 and TSC2 mutations observed in other aspects of TAND were of interest, particularly as all previous genotype-phenotype studies have suggested a more "severe" phenotype associated with TSC2 [38]. The possibility that specific aspects of TAND may be more likely in association with TSC1 is therefore a potentially important observation. We acknowledge that a TSC1 vs TSC2 differentiation may be highly oversimplistic, given that specific TSC mutations may be associated with very different functional consequences at a biochemical level [16,26].
The challenge around unreported and missing TAND data in this study was significant. Fewer than 40% of behavioural data were available and only 39.9% of participants in the TOSCA registry received an evaluation of their intellectual ability. The overall rates of ID were consistent with previous studies, [8,12] but we did not observe a bimodal distribution of intellectual ability as previously reported [23,25,39,40]. Given that only about half of the TOSCA cohort received a formal evaluation for IQ, it is possible that more severely impaired individuals may not have been referred for an assessment of their intellectual or developmental quotient (as was done in the population-based study that observed the clear bimodal pattern) [39]. Academic data were not reported or were missing in 43.4% of the cohort and 38.9% were not reported or were missing regarding neuropsychological skills. Only 48.8% of those with academic difficulties ever received evaluations, and only 41.6% were evaluated for neuropsychological deficits. Whilst the very high rates of academic and neuropsychological deficits in TSC were a novel finding in the TOSCA registry and underlines the recommendation that all children with TSC should receive evaluation of their academic and neuropsychological needs, [2,35] the very low proportion assessed raises significant clinical concern.
Conclusions Together, findings from the TOSCA registry emphasize the magnitude of neuropsychiatric disorders in TSC, which has an enormous impact on quality of life of individuals with TSC and their families. However, there was a high proportion of non-reported or missing data, which may have impacted the overall findings of the study. Due to the observational nature of the registry, only data already available from clinical practice were collected. Moreover, considering the disease complexity and the fact that individuals with TSC were not always followed for all disease manifestations by the site involved in the registry, ensuring report of all disease manifestations of each participant was a major challenge. We therefore acknowledge the potential ascertainment biases that may come with large-scale clinic-based natural history studies such as this one.
There is clearly a highly dynamic interaction between development, genotype, intellectual ability, epilepsy, anti-epilepsy and other pharmacological treatments, and environment in TSC. Future studies should therefore aim to examine these interrelations in an integrated way using suitable multilevel computational modelling. We also acknowledge that further evaluations will be required to examine TAND in relation to intellectual level and sex. In addition, efforts are underway to determine if there may be "natural TAND clusters" consisting of natural groupings of TAND characteristics within and between individuals with TSC across levels of TAND [7,13]. The TOSCA cohort may provide a powerful dataset to examine this possibility further.
In spite of the limitations inherent to a large-scale natural history study, results provide strong data to encourage clinicians to evaluate for neuropsychiatric comorbidity in all children and adults with TSC. The 2012 TSC surveillance and management guidelines recommend annual screening for TAND [34]. A simple tool called the TAND Checklist has been developed and validated in a pilot study that could provide easy guidance to healthcare professionals in assessing neuropsychiatric difficulties in each individual with TSC [3,5,6].
Abbreviations ADHD: Attention deficit hyperactivity disorder; ASD: Autism spectrum disorders; ID: Intellectual disability; IQ: Intelligence quotient; TAND: TSCassociated neuropsychiatric disorders; TOSCA: TuberOus SClerosis registry to increase disease Awareness; TSC: Tuberous sclerosis complex drafting, revising, final review, and approval of the manuscript. SY Designing the study, data interpretation, drafting, revising, final review, and approval of the manuscript. BZ Designing the study, patient accrual, clinical care, data interpretation, drafting, revising, final review, and approval of the manuscript. ACJ Designing the study, patient accrual, clinical care, data interpretation, drafting, revising, final review, and approval of the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate All patients (or their legal representatives) provided written informed consent before enrolling in the TOSCA disease registry. The study was designed, implemented, and reported in accordance with the principles of Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Author details 1