Cortical dysplasia and autistic trait severity in children with Tuberous Sclerosis Complex: a clinical epidemiological study

Cortical dysplasia and autistic trait severity in children with Tuberous Sclerosis Complex: a... Eur Child Adolesc Psychiatry (2018) 27:753–765 https://doi.org/10.1007/s00787-017-1066-z ORIGINAL CONTRIBUTION Cortical dysplasia and autistic trait severity in children with Tuberous Sclerosis Complex: a clinical epidemiological study 1,2 2,3 1,2 1 Sabine E. Mous  · Iris E. Overwater  · Rita Vidal Gato  · Jorieke Duvekot  · 1,2 4 2,3 Leontine W. ten Hoopen  · Maarten H. Lequin  · Marie‑Claire Y. de Wit  · 1,2 Gwendolyn C. Dieleman   Received: 3 March 2017 / Accepted: 9 October 2017 / Published online: 23 October 2017 © The Author(s) 2017. This article is an open access publication Abstract Tuberous Sclerosis Complex (TSC) is charac- and the relevance of separately studying the two ASD terized by a high prevalence of autism spectrum disorders subdomains. (ASD). Little is known about the relation between corti- cal dysplasia and ASD severity in TSC. We assessed ASD Keywords Tubers · Radial migration lines · Autism · severity (using the Autism Diagnostic Observation Scale), Quantitative autistic traits · Intelligence · Cognitive tuber and radial migration line (RML) count and location, functioning and cognitive functioning in 52 children with TSC and performed regression and mediation analyses. Tuber and RML count were strongly positively related to ASD sever- Introduction ity. However, when correcting for cognitive functioning, the majority of associations became insignificant and only total Tuberous Sclerosis Complex (TSC) is an autosomal domi- tuber count remained associated to the severity of restricted/ nant disorder affecting 1 in 6000 people, caused by inacti - repetitive behaviors. Occipital RML count remained associ- vating TSC1 (chromosome 9) or TSC2 (chromosome 16) ated with overall ASD severity, and social communication/ variants. The TSC1 and TSC2 protein products form the interaction deficit severity specifically. This study shows the intracellular TSC1-TSC2 protein complex, which serves as important explanatory role of cognitive functioning in the a regulator of the mammalian target of rapamycin (mTOR) association between cortical dysplasia and ASD severity, pathway. Mutations in the TSC1 or TSC2 gene lead to a upregulation of the mTOR pathway, causing uncontrolled cell growth and abnormal differentiation and the prolifera - Electronic supplementary material The online version of this tion of benign overgrowths of cells and tissue in several article (doi:10.1007/s00787-017-1066-z) contains supplementary material, which is available to authorized users. organ systems including the brain, skin, kidneys, heart, eyes, lungs and bones [1]. In the brain this may lead to cortical * Sabine E. Mous dysplasia. The most common form of cortical dysplasia in s.mous@erasmusmc.nl TSC is the presence of cortical tubers, affecting over 80% Department of Child and Adolescent Psychiatry/Psychology, of all TSC patients. Cortical tubers are focal developmen- Erasmus Medical Center-Sophia Children’s Hospital, P.O. tal abnormalities of the cortex, characterized by disorgan- Box 2060, 3000 CB Rotterdam, The Netherlands ized lamination and atypical cellular growth, differentiation ENCORE Expertise Center for Neurodevelopmental and maturation [2, 3], which develop during prenatal brain Disorders, Erasmus Medical Center-Sophia Children’s development and can be detected by MRI from 20 weeks Hospital, P.O. Box 2060, 3000 CB Rotterdam, of gestation onwards. Postnatally, no new tubers arise, but The Netherlands in older children tubers may calcify or become cystic [1]. Department of Pediatric Neurology, Erasmus Medical Another form of cortical dysplasia is the presence of radial Center-Sophia Children’s Hospital, P.O. Box 2060, 3000 CB Rotterdam, The Netherlands migration lines (RMLs). RMLs are linear abnormalities that 4 extend from the ventricles to the cortex, representing areas Department of Radiology, University Medical Center Utrecht, P.O. Box 85500, 3508 GA Utrecht, The Netherlands of hypomyelination and white matter heterotopia [4]. RMLs Vol.:(0123456789) 1 3 754 Eur Child Adolesc Psychiatry (2018) 27:753–765 are a marker of abnormal neural migration and cortical that child psychopathology, such as ASD, might be better organization and are often associated with a tuber, but can described within a quantitative—or dimensional—frame- also be isolated. Both forms of cortical dysplasia may act as work has gained support in the last years [16]. Within this epileptogenic lesions. In Fig. 1, an example of a Magnetic framework of continuous symptom levels, the entire spec- Resonance Imaging (MRI) scan showing cortical tubers and trum of symptom severity is covered. Most likely, this is RMLs is provided. a more naturalistic representation of psychopathology, as Other features associated with brain pathology in TSC compared to the use of all-or-none dichotomous diagnostic include the presence of epilepsy (72–85% of all patients) and categories. Previous studies have suggested that the symp- cognitive impairment, with about 50% of patients having an toms and etiology of ASD indeed form such a spectrum [17], intellectual disability (IQ < 70), and a range of behavioral seemingly even extending into the general population [18]. and psychiatric symptoms. Autism spectrum disorder (ASD) Studying ASD as a quantitative trait (and thus also includ- is highly prevalent in children with TSC, with prevalence ing subclinical traits) rather than as a categorically defined rates estimated around 40–50% [5–8]. disorder can contribute to a better understanding of the dis- Previous studies have suggested the total number of corti- order and the potentially contributing biological pathways. cal tubers to be an important predictor for an ASD diagnosis An additional advantage of the use of quantitative severity in TSC, and the temporal lobes were suggested to be specifi- scores in research is that these continuous scores provide cally implicated [9, 10]. Other studies found the presence more statistical power and allow the application of advanced (yes/no) of temporal tubers to be associated with a higher statistical methods [17]. We found only a single study that likelihood of an ASD diagnosis [11], while others specifi- previously investigated the relation between a quantitative cally found the number of cyst-like tubers to be related to measure of ASD severity and tuber count, which did not find ASD diagnostic status [12] or found a diagnosis of autism to an association between cortical tuber count or location and be related to frontal and posterior tubers [13]. Still others did overall ASD severity [19]. Furthermore, ASD is character- not find an association between the occurrence of cortical ized by various difficulties that, according to the latest edi- tubers and an autism diagnosis [14], or found the number of tion of the Diagnostic and Statistical Manual of Mental Dis- tubers to be equally prevalent in intellectually disabled non- orders (DSM-5) [20], can be divided in two main domains; autistic and intellectually disabled autistic children and thus (1) deficits in social communication and interaction and non-specific for ASD [ 15]. These inconsistencies in n fi dings (2) restricted or repetitive patterns of behavior, interests or point out that the association between tuber burden and ASD activities. The nature of the symptoms in these domains is is still poorly understood. substantially different and symptom severity is not necessar - Although these previous studies have experimented with ily equal in both domains, making it plausible that distinct different ways of defining cortical tuber involvement (i.e. mechanisms in different brain regions may underlie these tuber presence (yes/no), tuber count, or tuber size), in these two different ASD symptom domains. Therefore, it would studies ASD has always only been categorically defined as be useful to not only study overall ASD severity, but also the presence or absence of an ASD diagnosis. The notion study the association between symptom severity in these Fig. 1 Example of T2-weighted images with arrows indicating a cortical tubers, and b radial migration lines 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 755 two distinct domains and tuber burden and location in TSC CSS for the two separate subdomains of the ADOS (social patients. affect (SA) and restricted and repetitive behaviors (RRB) In addition, it is known that TSC is characterized by a [29–31]), providing an indication of ASD severity rela- wide range of cognitive abilities and it has been shown that tive to the child’s age and expressive language level [31]. the severity of cognitive impairment is related to tuber bur- The ADOS CSS is a truly continuous measure, covering den [11, 21–23]. Similarly, an association between intellec- the entire spectrum of autistic traits. CSS range between 1 tual (dis)ability and autism severity has been demonstrated and 10, with lower scores indicating no to very little prob- [24]. It is of interest, but still unclear, whether tuber burden lems, and higher scores indicating severe ASD symptoms. is an independent factor in determining autism severity or if Raw ADOS total scores corresponding to an ADOS clas- cognitive impairment is an important (potentially mediating) sification ‘Non-spectrum’ are distributed across CSS 1–3, determinant in this association. ‘ASD’ across CSS 4–5, and ‘Autism’ across CSS 6–10. For Most neuroimaging studies in TSC have focused spe- descriptive purposes, children were classified as having ASD cifically on tuber characteristics and the association with (autism or the broader ASD phenotype) or not according to cognitive or behavioral problems. RMLs can be more dif- the revised ADOS-2 algorithms [27]. ficult to detect, and less is known about the contribution of RMLs to the neurocognitive phenotype in TSC. There are Cognitive functioning data, however, demonstrating that RMLs are associated with intelligence [25, 26], as well as with the severity of autistic Cognitive functioning was assessed using different intel- traits [26]. ligence measures according to best practice standards; in In the present study, we aim to investigate the associa- the majority of children (n = 32, 62%) this was either the tion between cortical dysplasia (i.e. the number and location Wechsler Intelligence Scale for Children-III (WISC-III) of cortical tubers and RMLs) and a clinical observational or Wechsler Preschool and Primary Scale of Intelligence- quantitative measure of ASD severity, and to study the role III (WPPSI-III) [32, 33]. In some children (n = 7, 13%) of cognitive functioning in this association. Furthermore, a non-verbal intelligence test was used, which was either we aim to investigate the specific association of tuber and the Wechsler Non Verbal scale of ability (WNV) [34] or RML count and location and ASD severity within the two Snijders-Oomen Nonverbal Intelligence Test (SON-R) [35]. main domains of ASD symptomatology (deficits in social From all intelligence tests full-scale intelligence quotients communication and interaction, and restricted or repetitive (TIQs) were used. For children who were at the floor of their behaviors). age-appropriate standardized scores, a developmental quo- tient (DQ) was calculated (developmental age/chronological age × 100) [36]. Like IQ scores, a DQ of 100 is considered Methods the mean. In a number of children (n = 13, 25.0%) no formal intelligence test could be performed due to a young (devel- Participants opmental) age. In these children the cognitive developmental age was evaluated using one of the Bayley Scales of Infant The medical records of all pediatric TSC patients in care at and Toddler Development (BSID-II or Bayley-III [37, 38]) the expertise center ENCORE (Erasmus MC-Sophia Chil- or the Vineland Screener [39]. Again, we used the estimated dren’s Hospital, Rotterdam, the Netherlands) were retrospec- cognitive developmental age to calculate a DQ, according to tively reviewed. In 75 patients, MRI scans of the brain were the formula provided above. available for review. Of this sample, 52 patients (24 boys, 28 girls, 2–17 years of age) also had data on ASD severity and cognitive functioning. Neuroimaging Measures Brain MRI scans were made on a 1.5 Tesla General Electric scanner. For children with more than one available MRI, the Autism spectrum symptoms MRI closest in time to the ADOS assessment was selected. All MRI scans were visually inspected by two trained medi- The severity of ASD was assessed using the Autism Diag- cal students, and re-assessed by a pediatric neuroradiologist nostic Observation Scale (ADOS) [27, 28]. All ADOS and a pediatric neurologist. A protocol for data collection assessments were performed and scored by a trained, expe- was developed in which fluid-attenuated inversion recovery rienced and certified psychologist or psychiatrist. (FLAIR) images were used to assess the number and loca- For our main analyses, a continuous total standardized tion of tubers and RMLs. If FLAIR images were not present, calibrated severity score (CSS) was calculated, as well as a or for clarification purposes, T2-weighted images were used. 1 3 756 Eur Child Adolesc Psychiatry (2018) 27:753–765 Statistical analysis In supplementary Table S1 (online resource) the Pearson correlations between all variables of interest are shown. In Data were analyzed in IBM SPSS Statistics version 21 supplementary Figure S1 (online resource) the distribution [40]. Associations between the various variables of interest of the ADOS total CSS by IQ/DQ is shown, split by ADOS were studied calculating Pearson correlation coefficients. classification. To investigate the association between ASD severity and total tuber and RML count, linear regression analyses were Association ASD severity and tuber count performed. When these yielded significant results, post hoc analyses were performed, assessing the associations in the The association between the ADOS total severity score separate lobes of the brain. A Bonferroni correction was and total tuber count was highly significant (β  =  0.46, applied to correct for multiple testing. Because of the con- p < 0.001), and about 20% of the variance in the ADOS total siderable intercorrelations between the number of tubers severity score could be explained by the total number of cor- or RMLs in the separate lobes (ranging between 0.63–0.78 tical tubers. Post-hoc analyses assessing the separate lobes and 0.32–0.51, respectively), we first calculated the effec- of the brain indicated similar results for all lobes (Table 2). tive number of tests and adjusted the Bonferroni correction Because IQ/DQ was significantly related to both the accordingly to account for this lack of independence [41]. ADOS total severity score and the total number of tubers The calculation yielded an effective number of 2.98 tests (as well as to the two separate ADOS subdomain scores and for the tuber analyses and 3.57 tests for the RML analyses. the number of tubers in all separate lobes) (supplementary In all tables, both uncorrected as well as Bonferroni cor- Table S1, online resource), the analyses were repeated with rected (p ) p-values are provided, as well as β and adjusted IQ/DQ added as a covariate. The results of these analyses corr R effect size measures. Supplementary linear regression show that this correction rendered all associations insignifi- analyses were performed studying the association of ASD cant, indicating that IQ/DQ was an important explanatory severity with the number of cystic and calcified tubers. We variable in the associations (Table 2). also performed supplementary t tests and logistic regression Supplementary analyses studying the association between analyses, using the categorical ADOS classification instead the categorical ADOS classification and tuber count show of the continuous severity scores. similar, but reduced in magnitude, results (supplementary To assess whether IQ/DQ was a mediator in the associa- Tables S2 and S3 (online resource). tion between ASD severity and tuber or RML count, formal Next, we studied the association between total tuber count mediation analyses were performed using the ‘PROCESS’ and ASD severity in the two subdomains of the ADOS; macro for SPSS, version 2.15 (http://www.afhayes.com/) social affect (SA) and restricted and repetitive behaviors with bias-corrected bootstrapping using 1000 replications (RRB) (Table  3). Again, strong associations were found [42]. For the mediation analyses, effect sizes are reported as between the total number of tubers and ADOS SA and κ , with values of 0.01, 0.09 and 0.25 considered as small, RRB severity scores (β  =  0.37, p = 0.008 and β  =  0.49, medium and large respectively [43]. Finally, supplementary p < 0.001), and 12 and 22% of the variance in respectively mediation analyses were performed to study the role of epi- SA and RRB severity score was explained by total tuber lepsy in the association between ASD severity and tuber or count. RML burden. After adding IQ/DQ as a covariate, the total number of tubers only remained significantly associated with the RRB severity score (β  =  0.29, p  =  0.046). Post-hoc analyses Results studying the separate lobes of the brain indicated that this association was mainly driven by tuber count in the frontal Patient characteristics lobes (β = 0.30, p = 0.042, p  = 0.124). An (uncorrected) corr trend was visible for the association between temporal tuber In total, the data of 52 TSC patients (24 boys, 28 girls) were count and RRB severity (β = 0.25, p = 0.071, p  = 0.211). corr included. The mean age at time of the MRI was 7.0 years To formally assess whether IQ/DQ was a mediator in (range 0–17) and the mean age at time of the ADOS assess- the association between the total number of tubers and the ment was 8.8 years (range 2–17). The total number of tubers ADOS total severity score, a mediation analysis was per- ranged between 0 and 81 and the total number of RMLs formed (Fig.  2, panel a). The mediation analysis showed between 0 and 37. The largest number of tubers and RMLs that the direct effect (c’ path) of total tuber count on the were located in the frontal lobes. According to the ADOS, total severity score was insignificant. The indirect (a*b path) a total number of 25 children (48.1%) met the criteria for effect through IQ/DQ was large and statistically significant an ASD classification. Additional patient characteristics are (B = 0.04, 95% CI = 0.02; 0.06, κ  = 0.263). This implies shown in Table 1. that the total effect (c path) between total tuber count and 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 757 Table 1 Patient characteristics n (%) Mean (SD) Min–Max Gender, male 24 (46.2) Age in years  During MRI 7.0 (3.9) 0–17  During ADOS 8.8 (4.2) 2–17 Age difference in years between ADOS-MRI 1.8 (2.9) 0–14 Mutation  TSC1 14 (26.9)  TSC2 36 (69.2)  No mutation identified 1 (1.9)  Not tested 1 (1.9) Epilepsy  Present, yes 46 (88.5)  West syndrome, yes (from n = 46) 16 (34.8)  Age of onset, months (from n = 46) 15.4 (18.6) 1–91 Tubers  Present, yes 51 (98.1)  Number   Total 27.5 (20.2) 0–81   Frontal lobe 16.0 (12.1) 0–54   Parietal lobe 5.5 (4.6) 0–19   Temporal lobe 3.7 (3.3) 0–12   Occipital lobe 2.4 (2.6) 0–9 Cystic tubers  Present, yes 19 (36.5)  Number   Total 1.85 (3.88) 0–18   Frontal lobe 0.96 (2.21) 0–10   Parietal lobe 0.44 (1.18) 0–7   Temporal lobe 0.33 (0.98) 0–4   Occipital lobe 0.12 (0.32) 0–1 Calcified tubers  Present, yes 9 (17.3)  Number   Total 0.71 (2.52) 0–17   Frontal lobe 0.33 (1.32) 0–9   Parietal lobe 0.15 (0.75) 0–5   Temporal lobe 0.06 (0.31) 0–2   Occipital lobe 0.17 (0.62) 0–3 Radial migration lines  Present, yes 51 (98.1)  Number   Total 16.0 (10.2) 0–37   Frontal lobe 8.2 (5.6) 0–23   Parietal lobe 3.9 (3.3) 0–15   Temporal lobe 3.0 (3.0) 0–12   Occipital lobe 0.9 (1.2) 0–4 IQ/DQ 59.7 (24.5) 8–114 Intellectual disability (IQ/DQ < 70), yes 34 (65.4) ADOS module  Module 1 15 (28.8)  Module 2 11 (21.2) 1 3 758 Eur Child Adolesc Psychiatry (2018) 27:753–765 Table 1 (continued) n (%) Mean (SD) Min–Max  Module 3 18 (34.6)  Module 4 8 (15.4) ADOS classification  Non-spectrum 27 (51.9)  Autism spectrum disorder 25 (48.1) ADOS calibrated severity score  Total 4.0 (2.7) 1–10  Social affect domain 4.3 (2.6) 1–10  Restricted and repetitive behaviors domain 4.8 (2.8) 1–10 n = 52 ADOS Autism Diagnostic Observation Scale, DQ developmental quotient, IQ intelligence quotient, MRI magnetic resonance imaging Raw ADOS total scores corresponding to an ADOS classification of ‘Autism’ were distributed across cali- brated severity scores (CSS) 6–10, ‘ASD’ across 4–5 and ‘Non-spectrum’ across 1–3 Table 2 Association ADOS total calibrated severity score and tuber count Model I  Model I + IQ/DQ a a 2 2 B 95% CI β p p B 95% CI β p p R R corr corr adj adj Total number of tubers 0.06 0.03; 0.09 0.46 < 0.001 – 0.195 0.02 − 0.01; 0.06 0.18 0.188 – 0.356 Frontal lobes 0.09 0.03; 0.15 0.41 0.002 0.007 0.155 0.03 − 0.04; 0.09 0.11 0.414 1 0.341 Parietal lobes 0.24 0.10; 0.39 0.43 0.002 0.005 0.166 0.11 − 0.04; 0.25 0.19 0.150 0.447 0.360 Temporal lobes 0.31 0.10; 0.52 0.38 0.005 0.016 0.129 0.11 − 0.09; 0.31 0.14 0.278 0.829 0.348 Occipital lobes 0.40 0.13; 0.67 0.39 0.004 0.012 0.136 0.24 − 0.00; 0.47 0.23 0.054 0.160 0.382 n = 52 ADOS Autism Diagnostic Observation Scale, IQ intelligence quotient, DQ developmental quotient, R adjusted R squared model adj Multiple testing correction (2.98 effective tests) applied Significant associations are highlighted in bold font the total severity score was fully mediated by IQ/DQ. Post- count on the RRB severity score also remained significant hoc mediation analyses for the separate lobes were also per- (B  =  0.04, p = 0.046). This means that a direct effect of formed (figures not shown), also showing full mediation by total tuber count on the RRB severity score was present and IQ/DQ for all lobes. that the total effect (c path) was only partly mediated by IQ/ Next, mediation analyses were performed studying the DQ. Again, in line with the regression analyses, post hoc role of IQ/DQ in the association between total tuber count analyses of the separate lobes (figures not shown) indicated and the severity score of the two separate ADOS domains. full mediation by IQ/DQ, with the exception of the frontal For the SA domain (Fig. 2, panel b), the direct effect (c’ lobes. For the frontal lobe association with the RRB score, path) of total tuber count on the SA severity score was insig- the indirect effect (a*b path) through IQ/DQ was medium nificant. The indirect effect (a*b path) through IQ/DQ was to large and significant (B  = 0.05, 95% CI = 0.01; 0.09, medium to large and statistically significant (B  = 0.03, 95% κ  = 0.188), but the direct effect (c’ path) between frontal CI = 0.02; 0.05, κ  = 0.242). Again, this means that the lobe tuber count and the RRB severity score remained sig- total effect (c path) between total tuber count and the SA nificant as well (B  = 0.07, p = 0.042), again implying only severity score was fully mediated by IQ/DQ. However, in partial mediation by IQ/DQ. line with the regression analyses, the mediation analysis per- Because cystic and calcified tubers have been suggested formed with the RRB domain (Fig. 2, panel c) showed that to be more epileptogenic and related to a more severe phe- although the indirect effect (a*b path) through IQ/DQ was notype [44–47], we additionally studied the associations medium to large and significant (B  = 0.03, 95% CI = 0.01; between the ADOS severity scores and the number of cystic 0.05, κ  = 0.189), the direct effect (c’ path) of total tuber and calcified cortical tubers (supplementary Tables S4 and 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 759 Table 3 Association ADOS subdomain calibrated severity scores and tuber coufnt Model I Model I + IQ/DQ a a 2 2 B 95% CI β p p B 95% CI β p p R R corr corr adj adj Total number of tubers  SA domain CSS 0.05 0.01; 0.08 0.37 0.008 – 0.117 0.01 − 0.02; 0.05 0.10 0.497 – 0.262  RRB domain CSS 0.07 0.03; 0.10 0.49 <0.001 – 0.224 0.04 0.00; 0.08 0.29 0.046 – 0.299 Frontal lobes  SA domain CSS 0.07 0.01; 0.12 0.32 0.023 0.069 0.081 0.00 − 0.06; 0.07 0.02 0.882 1 0.255  RRB domain CSS 0.12 0.06; 0.17 0.49 < 0.001 0.001 0.229 0.07 0.00; 0.14 0.30 0.042 0.124 0.301 Parietal lobes  SA domain CSS 0.22 0.08; 0.36 0.40 0.003 0.010 0.141 0.10 − 0.05; 0.25 0.19 0.169 0.503 0.283  RRB domain CSS 0.22 0.06; 0.38 0.36 0.008 0.025 0.114 0.09 − 0.08; 0.26 0.15 0.275 0.821 0.257 Temporal lobes  SA domain CSS 0.20 − 0.01; 0.41 0.26 0.064 0.192 0.048 0.02 − 0.19; 0.23 0.02 0.876 1 0.255  RRB domain CSS 0.37 0.15; 0.58 0.43 0.001 0.004 0.168 0.21 − 0.02; 0.43 0.25 0.071 0.211 0.288 Occipital lobes  SA domain CSS 0.34 0.08; 0.61 0.35 0.012 0.036 0.102 0.20 − 0.05; 0.45 0.20 0.111 0.330 0.293  RRB domain CSS 0.34 0.04; 0.63 0.31 0.026 0.079 0.077 0.18 − 0.10; 0.46 0.16 0.201 0.598 0.264 n = 52 ADOS Autism Diagnostic Observation Scale, CSS calibrated severity score, SA social affect, RRB  restricted and repetitive behaviors, IQ intel- ligence quotient, DQ developmental quotient, R adjusted R squared model adj Multiple testing correction (2.98 effective tests) applied Significant associations are highlighted in bold font Fig. 2 Mediation analyses tuber count, ASD severity score and IQ/DQ. a ADOS total severity score, b ADOS Social Affect (SA) domain sever - ity score, c ADOS Restricted and Repetitive Behaviors (RRB) domain severity score S5, online resource). These analyses showed that, after cor- (β = 0.27, p = 0.020). This association was strongest in, and rection for IQ/DQ, only the number of calcified tubers was mainly driven by, the occipital lobe (β = 0.33, p = 0.003, significantly associated with the ADOS total severity score p  = 0.009) (Table S4). When studying the two ADOS corr 1 3 760 Eur Child Adolesc Psychiatry (2018) 27:753–765 subdomain scores, we found that after correction for IQ/DQ to the two separate ADOS subdomain scores and the number the total number of calcified tubers was significantly related of RMLs in the frontal and parietal lobes) (supplementary to the RRB subdomain severity score (β = 0.32, p = 0.010). Table S1, online resource), the analyses were repeated with Post-hoc analyses studying the separate lobes of the brain IQ/DQ added as a covariate. After the correction for IQ/DQ revealed that this association was strongest in the occipital only the association for the occipital lobe remained significant lobes (β  =  0.29, p  =  0.016, p  = 0.046), but also pre- (β = 0.28, p = 0.013, p  = 0.047) (Table 4). corr corr sent in the parietal and temporal lobes (β = 0.26, p = 0.033, Supplementary analyses studying the association between p  = 0.094, and β = 0.29, p = 0.021, p  = 0.060 respec- the categorical ADOS classification and the number of radial corr corr tively) (Table S5). Correction for age during MRI did not migration lines were performed, showing similar effects, but change these results. considerably reduced in magnitude and mostly insignificant Because studies have suggested that an early onset of (supplementary Tables S6 and S7 (online resource). epilepsy may have a deleterious influence on early brain Next, we studied the association between RML count and development and may act as a risk factor for autism [48], ASD severity in the two subdomains of the ADOS; social supplementary mediation analyses were performed to study affect (SA) and restricted and repetitive behaviors (RRB) the role of epilepsy in the association between tuber bur- (Table 5). Again, highly significant associations were found den and ASD severity (supplementary Figure S2, online between total RML count and ADOS SA and RRB severity resource). The results of these multiple mediation models (β = 0.41, p = 0.002 and β = 0.33, p = 0.016), and 16% and (that simultaneously included age of epilepsy onset and IQ/ 9% of the variance in respectively SA and RRB severity DQ as mediators) show that the association between tuber score was explained by total RML count. burden and ASD severity was not mediated by age of epi- After adding IQ/DQ as a covariate, only the number of lepsy onset (path a2*b2: B = 0.01, 95% CI = − 0.01;0.03), RMLs in the occipital lobes remained significantly asso- and fully mediated by IQ/DQ (path a1*b1: B = 0.04, 95% ciated with the SA severity score (β  =  0.30, p  =  0.013, CI = 0.02; 0.06). Similar results were found when separately p  = 0.047). corr analyzing the two ADOS subdomains. Again, a formal mediation analysis was performed, assessing whether IQ/DQ was a mediator in the associa- Association ASD severity and radial migration line tion between the total number of RMLs and the ADOS count total severity score (Fig. 3, panel a). The mediation analy- sis showed that the direct effect (c’ path) of total RML A significant association was found between the ADOS total count on the total severity score was insignificant. The severity score and the total number of radial migration lines indirect effect (a*b path) through IQ/DQ was medium (RMLs) (β = 0.40, p = 0.003), and 15% of the variance in to large and statistically significant (B  =  0.06, 95% ADOS total severity score was accounted for by the total num- CI = 0.03; 0.10, κ  = 0.223). This implies that the total ber of radial migration lines. Post-hoc analyses assessing the effect (c path) between total RML count and the total separate lobes of the brain indicated significant associations severity score was fully mediated by IQ/DQ. Post-hoc for the parietal lobes and occipital lobes specifically (Table  4). mediation analyses for the frontal and parietal lobes were Because IQ/DQ was significantly related to both the ADOS performed as well (figures not shown), also showing full total severity score and the total number of RMLs (as well as mediation by IQ/DQ. Since IQ/DQ was not significantly Table 4 Association ADOS total calibrated severity score (CSS) and the number of radial migration lines (RMLs) Model I Model I + IQ/DQ a a 2 2 B 95% CI β p p R B 95% CI β p p R corr corr adj adj Total number of RMLs 0.11 0.04;0.17 0.40 0.003 – 0.146 0.05 − 0.02; 0.11 0.19 0.141 – 0.362 Frontal lobes 0.14 0.02;0.27 0.31 0.025 0.090 0.078 0.04 − 0.08; 0.15 0.08 0.505 1 0.338 Parietal lobes 0.28 0.07;0.49 0.35 0.011 0.038 0.106 0.14 − 0.05; 0.33 0.18 0.140 0.501 0.362 Temporal lobes 0.21 − 0.04; 0.46 0.24 0.094 0.336 0.036 0.10 − 0.11; 0.31 0.11 0.364 1 0.344 Occipital lobes 0.90 0.30;1.50 0.39 0.004 0.015 0.135 0.65 0.14; 1.16 0.28 0.013 0.047 0.412 n = 52 ADOS Autism Diagnostic Observation Scale, IQ intelligence quotient, DQ developmental quotient, R adjusted R squared model adj Multiple testing correction (3.57 effective tests) applied Significant associations are highlighted in bold font 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 761 Table 5 Association ADOS subdomain calibrated severity scores (CSS) and the number of radial migration lines (RMLs) Model I Model I + IQ/DQ a a 2 2 B 95% CI β p p B 95% CI β p p R R corr corr adj adj Total number of RMLs  SA domain CSS 0.10 0.04;0.17 0.41 0.002 – 0.155 0.06 − 0.01; 0.12 0.23 0.080 – 0.300  RRB domain CSS 0.09 0.02; 0.17 0.33 0.016 – 0.093 0.04 − 0.04; 0.11 0.14 0.300 – 0.255 Frontal lobes  SA domain CSS 0.15 0.03; 0.27 0.33 0.017 0.061 0.091 0.06 − 0.06; 0.18 0.14 0.296 1 0.271  RRB domain CSS 0.13 − 0.00; 0.27 0.27 0.057 0.203 0.052 0.03 − 0.10; 0.17 0.07 0.608 1 0.243 Parietal lobes  SA domain CSS 0.27 0.07; 0.47 0.35 0.010 0.037 0.107 0.15 − 0.04; 0.35 0.20 0.110 0.394 0.293  RRB domain CSS 0.21 − 0.03; 0.44 0.25 0.081 0.288 0.041 0.07 − 0.14; 0.29 0.09 0.504 1 0.246 Temporal lobes  SA domain CSS 0.20 − 0.04; 0.44 0.23 0.101 0.359 0.034 0.10 − 0.11; 0.32 0.12 0.343 1 0.268  RRB domain CSS 0.24 − 0.03; 0.50 0.25 0.075 0.269 0.043 0.13 − 0.10; 0.37 0.14 0.262 0.935 0.258 Occipital lobes SA domain CSS 0.87 0.29; 1.45 0.39 0.004 0.015 0.137 0.66 0.14; 1.18 0.30 0.013 0.047 0.343 RRB domain CSS 0.64 − 0.03; 1.31 0.26 0.062 0.222 0.049 0.40 − 0.20; 1.00 0.16 0.186 0.665 0.266 n = 52 ADOS Autism Diagnostic Observation Scale, SA social affect, RRB restricted and repetitive behaviors, IQ intelligence quotient, DQ developmen- tal quotient, R adjusted R squared model adj Multiple testing correction (3.57 effective tests) applied Significant associations are highlighted in bold font Fig. 3 Mediation analyses RML count, ASD severity score and IQ/DQ. a ADOS total severity score, b ADOS Social Affect (SA) domain sever - ity score, c ADOS Restricted and Repetitive Behaviors (RRB) domain severity score related to RML count in the temporal and occipital lobes, score could not be mediated by IQ/DQ, implying a direct the earlier identified association between total number of effect of occipital lobe RML count on the total ADOS RMLs in the occipital lobes and the ADOS total severity severity score. 1 3 762 Eur Child Adolesc Psychiatry (2018) 27:753–765 As with tuber count, the mediation analyses were also from the fact that the frontal lobes represent the largest brain performed studying the role of IQ/DQ in the association area of all lobes. When studying the association between between total RML count and the severity score of the two RML count and ASD severity we found similar results, again separate ADOS subdomains (Fig. 3, panel b and c). For both initially showing strong associations between the number of subdomains, the direct effect (c’ path) of total RML count RMLs and ASD severity, but which were for the most part on the subdomain severity score was insignificant. The indi- rendered insignificant when corrected for IQ/DQ, except for rect effect (a*b path) through IQ/DQ was medium to large the relation between the number of RMLs in the occipital and statistically significant (B = 0.05, 95% CI = 0.02; 0.09, lobes and total and social affect ASD severity. The formal κ  = 0.184 for SA domain, and B = 0.05, 95% CI = 0.02; mediation analyses confirmed all results and showed that, 0.10, κ  = 0.190 for RRB domain). This implies that the indeed, all other initial findings were fully mediated by IQ/ total effect (c path) between total RML count and the SA DQ. and RRB severity scores was fully mediated by IQ/DQ. Post- The results emphasize the importance of taking cognitive hoc analyses of the separate lobes (figures not shown) again functioning into account when studying the relation between showed full mediation by IQ/DQ for the frontal and parietal brain pathology and ASD in patients with TSC. TSC is lobes for both the SA and RRB subdomains. Because IQ/DQ often characterized by cognitive impairment, and previous was not significantly related to RML count in the temporal studies have shown that cognitive impairment is strongly and occipital lobes, the earlier found association between related to both brain pathology [11, 21–23] and ASD sever- RML count in the occipital lobes and the SA severity score ity [24], thereby acting as an important confounding (or could not be mediated by IQ/DQ, implying a direct effect of rather explanatory) factor in this association. However, one occipital lobe RML count on the SA severity score. should also realize that, regardless of the explanatory role of Finally, supplementary mediation analyses were per- cognitive functioning, ASD symptoms remain a significant formed to study the effect of epilepsy on the association problem in patients with TSC. We found a direct association, between the number of RMLs and ASD severity (sup- regardless of cognitive functioning, between RML count in plementary Figure S3, online resource). These multiple the occipital lobes and the severity of problems in social mediation models (simultaneously including age of epi- communication and interaction. The main function of the lepsy onset and IQ/DQ as mediators) show that the asso- occipital lobes is processing visual stimuli, and structural ciation between RML count and ASD severity was not and functional abnormalities in the occipital and occipito- mediated by age of epilepsy onset (path a2*b2: B = 0.01, temporal regions have been frequently reported in ASD [49]. 95% CI = − 0.00;0.03), and was fully mediated by IQ/DQ One might argue that, next to IQ, epilepsy severity could (path a1*b1: B = 0.05, 95% CI = 0.02;0.10). Similar results be an important confounding/explanatory factor in the asso- were obtained when separately analyzing the two ADOS ciation between cortical dysplasia and autistic trait severity. subdomains. A recent study by our group has shown that, in a multivari- able model including various epilepsy severity indicators, age of epilepsy onset was the only significant predictor for Discussion cognitive functioning later in life [50]. Therefore, we ran supplementary mediation analyses additionally including In the current clinical epidemiological study, the association this variable (as proxy for epilepsy severity) as mediator. between cortical dysplasia and a quantitative observational The results of these analyses showed that, although there measure of ASD severity was studied in a clinical sample of was a significant association between the number of tubers children with TSC. The specificity of the association with and the age of epilepsy onset, the age of epilepsy onset was the two main subdomains of ASD symptomatology (deficits not related to any of the ASD severity scores or the number in social communication and interaction, and restricted or of RMLs, and was no mediator in the associations studied. repetitive behaviors) was studied as well. Finally, we focused Because the large majority of our sample (46/52, 88.5%) on the role of cognitive functioning in these associations. was using anti-epileptic drugs (AEDs), this sample does not The initial analyses, not corrected for IQ/DQ, showed that allow us to study the effect of AED use on the described total cortical tuber count, as well as tuber count in the sepa- associations. Although the exact effect of AED use on ASD rate lobes of the brain, was strongly related to the severity severity remains unclear due to a small number of studies of ASD, visible in both ASD subdomains. However, when and limited sample sizes, it has been suggested that the use IQ/DQ was added as a covariate to the analyses, only total of AEDs may have a beneficial, but most likely very small, and frontal tuber count remained related to the severity of effect on ASD severity [ 51, 52]. If this is indeed the case, restricted and repetitive behaviors, although it must be noted the use of AEDs might have attenuated our results. To assist that the frontal association did not survive correction for in further elucidating the association between epilepsy and multiple testing and that the relationship may partly arise ASD severity in TSC, future studies might not only consider 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 763 to study the effect of AEDs on ASD severity, but also the findings is enhanced by referring all TSC patients within our effect of other epilepsy indicators such as infantile spasms, expertise center for a developmental and psychiatric evalua- epilepsy refractoriness, and current epilepsy status. tion (regardless of whether or not the child experiences cog- The difference in findings between our and other stud- nitive or behavioral difficulties), the risk of residual selec - ies (but also between previous studies) can most likely be tion bias remains; a first selection takes place when parents explained by large differences in methodology, such as (1) decide whether or not to visit the expertise center with their different ASD measures (i.e. clinical observational meas- child, and a second selection occurs when parents decide ure vs. screening questionnaire or clinical diagnosis, and whether or not they want to visit the department of Child continuous severity scores vs. dichotomous diagnostic cat- and Adolescent Psychiatry/Psychology for a developmental egories), (2) different ways of defining brain involvement and psychiatric evaluation. (i.e. tuber/RML count vs. volume or absence/presence of To conclude, our study initially showed strong associa- tubers/RMLs), (3) different statistical techniques, (4) the tions between cortical dysplasia and ASD severity, with in- or exclusion of confounding variables in the statistical children with more cortical tubers and RMLs having more models and (5) participant selection. severe ASD symptoms. However, for the majority of these A strength and novel aspect of our study is the use of a associations, cognitive functioning was identified as an quantitative measure of ASD severity. Not only does this important confounding—or rather explanatory—factor, approach provide a more naturalistic representation of ASD highlighting the importance of taking cognitive function- symptoms and more statistical power [17], it also allowed us ing into account when studying the relation between brain to study the two different main domains of ASD symptoma- pathology and ASD symptomatology. Regardless of cogni- tology; difficulties in social communication and interaction, tive functioning, children with more tubers overall showed and restricted and repetitive behaviors. Another strength of more severe restrictive and repetitive behaviors, and children this study is the use of a standardized observational measure with more RMLs in the occipital lobes specifically showed of ASD, thereby reducing reporter bias. It must be noted more difficulties in social communication and interaction. that, although the ADOS is an instrument aiming at meas- These findings underline the importance of separately study - uring autistic traits, it remains unclear whether these traits ing problems in social communication and interaction on (especially in non-spectrum patients) truly originate from an the one hand, and restricted and repetitive behaviors on the ASD predisposition or are caused by other factors that might other hand. affect social behavior and restricted or repetitive behaviors. Acknowledgements This research was financially supported by the Also, the direction of effect between cortical dysplasia and Sophia Children’s Hospital Fund (Rotterdam, the Netherlands) under autistic trait severity remains unclear. Although it seems grant number SSWO B14-02. Further financial support was provided plausible that the cortical abnormalities have an adverse by the Dutch Brain Foundation (Hersenstichting) and the Dutch Epi- lepsy Foundation (Epilepsiefonds). Funders were not involved in the impact on brain development, consequently leading to more design of the study, nor in data collection, analysis, interpretation or severe ASD symptoms and developmental delay, it might writing the manuscript. The authors thank Karen Bindels-de Heus well be that in fact all result from another shared factor. for patient care and data collection, and Simone Eijk, Emma van der The stepwise approach and correction for multiple test- Ende and Kimberley Hanemaayer for their help in data collection and cleaning. ing in the current study makes it less plausible that findings are false positive, although this cannot be ruled out entirely. Compliance with ethical standards Mitigating this concern however, is the modest sample size in which these results were obtained. This relatively small Ethical approval All procedures performed in studies involving sample size, which is a limitation of our study, might have human participants were in accordance with the ethical standards of reduced the power to reveal relatively subtle effects, poten- the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments. This retrospective study tially resulting in an underestimation of effects. It is of great was approved by the Medical Ethics Committee of the Erasmus Medi- importance that future larger studies attempt to replicate our cal Center, the Netherlands. findings, before any strong conclusions can be made regard- ing the association between cortical dysplasia and ASD Informed consent Written informed consent was formally waived symptom severity. Furthermore, our study employs clinical as there is no patient burden and no privacy concern. MRI scans that were made on an 1.5 Tesla scanner. This might have led to limited power to detect RMLs, and made it Conflict of interest The authors declare that they have no conflict impossible to accurately retrospectively measure the volume of interest. of cortical tubers, thereby preventing us from studying the relation between ASD severity and tuber volume or tuber- Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://crea- brain proportion. Finally, even though the risk of selection tivecommons.org/licenses/by/4.0/), which permits unrestricted use, bias in our sample is reduced and the generalizability of 1 3 764 Eur Child Adolesc Psychiatry (2018) 27:753–765 distribution, and reproduction in any medium, provided you give appro- 18. Blanken LME, Mous SE, Ghassabian A et al (2015) Cortical mor- priate credit to the original author(s) and the source, provide a link to phology in 6- to 10-year old children with autistic traits: a popu- the Creative Commons license, and indicate if changes were made. lation-based neuroimaging study. Am J Psychiatry 172:479–486. doi:10.1176/appi.ajp.2014.14040482 19. Weber A, Egelhoff J, McKellop J, Franz D (2000) Autism and the cerebellum: evidence from tuberous sclerosis. J Autism Dev Disord 30:511–517 References 20. American Psychiatric Association (2013) Diagnostic and statisti- cal manual of mental disorders (5th ed.). Am J Psychiatry. doi: 1. Crino PB (2013) Evolving neurobiology of tuberous sclero- 10.1176/appi.books.9780890425596.744053 sis complex. Acta Neuropathol 125:317–332. doi:10.1007/ 21. Kaczorowska M, Jurkiewicz E, Domańska-Pakieła D et  al s00401-013-1085-x (2011) Cerebral tuber count and its impact on mental outcome 2. Talos DM, Kwiatkowski DJ, Cordero K et al (2008) Cell-specific of patients with tuberous sclerosis complex. Epilepsia 52:22–27. alterations of glutamate receptor expression in tuberous sclerosis doi:10.1111/j.1528-1167.2010.02892.x complex cortical tubers. Ann Neurol 63:454–465. doi:10.1002/ 22. Kassiri J, Snyder TJ, Bhargava R et al (2011) Cortical tubers, ana.21342 cognition, and epilepsy in tuberous sclerosis. Pediatr Neurol 3. Crino PB, Trojanowski JQ, Dichter MA, Eberwine J (1996) 44:328–332. doi:10.1016/j.pediatrneurol.2011.01.001 Embryonic neuronal markers in tuberous sclerosis: single- 23. O’Callaghan FJK, Harris T, Joinson C et al (2004) The rela- cell molecular pathology. Neurobiology 93:14152–14157. tion of infantile spasms, tubers, and intelligence in tuberous doi:10.1073/pnas.93.24.14152 sclerosis complex. Arch Dis Child 89:530–533. doi:10.1136/ 4. Griffiths PD, Bolton P, Verity C (1998) White matter abnor - adc.2003.026815 malities in tuberous sclerosis complex. Acta Radiol 39:482–486. 24. Hoekstra RA, Happé F, Baron-Cohen S, Ronald A (2009) Asso- doi:10.1080/02841859809172211 ciation between extreme autistic traits and intellectual disabil- 5. Curatolo P, Moavero R, de Vries PJ (2015) Neurological and ity: insights from a general population twin study. Br J Psychia- neuropsychiatric aspects of tuberous sclerosis complex. Lancet try 195:531–536. doi:10.1192/bjp.bp.108.060889 Neurol 14:733–745. doi:10.1016/S1474-4422(15)00069-1 25. Shepherd CW, Houser OW, Gomez MR (1995) MR findings in 6. de Vries PJ, Whittemore VH, Leclezio L et al (2015) Tuberous tuberous sclerosis complex and correlation with seizure devel- Sclerosis Associated Neuropsychiatric Disorders (TAND) and opment and mental impairment. Am J Neuroradiol 16:149–155 the TAND checklist. Pediatr Neurol 52:25–35. doi:10.1016/j. 26. van Eeghen AM, Terán LO, Johnson J et al (2013) The neu- pediatrneurol.2014.10.004 roanatomical phenotype of tuberous sclerosis complex: focus 7. de Vries PJ, Hunt A, Bolton PF et al (2007) The psychopatholo- on radial migration lines. Neuroradiology 55:1007–1014. gies of children and adolescents with tuberous sclerosis complex doi:10.1007/s00234-013-1184-3 (TSC). Eur Child Adolesc Psychiatry 16:16–24. doi:10.1007/ 27. Lord C, Rutter M, DiLavore PC et al (2012) Autism diagnostic s00787-006-0570-3 observation schedule, second edition (ADOS-2). Manual (Part 8. Jeste SS, Varcin KJ, Hellemann GS et al (2016) Symptom profiles I). Western Psychological Services, Los Angeles of autism spectrum disorder in tuberous sclerosis complex. Neu- 28. Lord C, Rutter M, DiLavore PC, Risi S (1999) Autism diagnos- rology 87:766–772. doi:10.1212/WNL.0000000000003002 tic observation schedule (ADOS). Manual. Western Psychologi- 9. Bolton PF, Griffiths PD (1997) Association of tuberous sclero- cal Services, Los Angeles sis of temporal lobes with autism and atypical autism. Lancet 29. Hus V, Gotham K, Lord C (2014) Standardizing ADOS Domain 349:392–395. doi:10.1016/S0140-6736(97)80012-8 Scores: separating severity of social affect and restricted and 10. Huang CH, Peng SSF, Weng WC et al (2015) The relationship repetitive behaviors. J Autism Dev Disord 44:2400–2412. of neuroimaging findings and neuropsychiatric comorbidities in doi:10.1007/s10803-012-1719-1 children with tuberous sclerosis complex. J Formos Med Assoc 30. Gotham K, Pickles A, Lord C (2009) Standardizing ADOS 114:849–854. doi:10.1016/j.jfma.2014.02.008 scores for a measure of severity in autism spectrum disorders. 11. Bolton PF, Park RJ, Higgins JNP et al (2002) Neuro-epileptic J Autism Dev Disord 39:693–705. doi:10.1007/s10803-008- determinants of autism spectrum disorders in tuberous sclerosis 0674-3.Standardizing complex. Brain 125:1247–1255. doi:10.1093/brain/awf124 31. Hus V, Lord C (2014) The Autism diagnostic observation 12. Numis AJ, Major P, Montenegro MA et  al (2011) Identifica- schedule, module 4: revised algorithm and standardized sever- tion of risk factors for autism spectrum disorders in tuber- ity scores. J Autism Dev Disord 44:1996–2012. doi:10.1007/ ous sclerosis complex. Neurology 76:981–987. doi:10.1212/ s10803-014-2080-3 WNL.0b013e3182104347 32. Wechsler D (1991) Wechsler intelligence scale for children, 3rd 13. Curatolo P, Cusmai R, Cortesi F et  al (1991) Neuropsychiat- edn. Psychological Corporation, San Antonio ric aspects of tuberous sclerosis. Ann N Y Acad Sci 615:8–16. 33. Wechsler D (2002) Wechsler preschool and primary scale of intel- doi:10.1111/j.1749-6632.1991.tb37743.x ligence, 3rd edn. Psychological Corporation, San Antonio 14. Walz NC, Byars AW, Egelhoff JC, Franz DN (2002) Supratentorial 34. Wechsler D, Naglieri JA (2006) Wechsler nonverbal scale of abil- tuber location and autism in tuberous sclerosis complex. J Child ity (WNV). Harcourt Assessment, San Antonio Neurol 17:830–832. doi:10.1177/08830738020170111401 35. Tellegen P, Winkel M, Wijnberg-Williams B, Laros J (2005) 15. Asano E, Chugani DC, Muzik O et al (2001) Autism in tuberous Snijders-Oomen Niet-Verbale Intelligentietest (SON-R) 2 1/2–7: sclerosis complex is related to both cortical and subcortical dys- Handleiding en Verantwoording. Boom Testuitgevers, Amsterdam function. Neurology 57:1269–1277. doi:10.1212/WNL.57.7.1269 36. van Eeghen AM, Black ME, Pulsifer MB et al (2012) Genotype 16. Hudziak J, Achenbach T, Althoff R, Pine D (2007) A dimensional and cognitive phenotype of patients with tuberous sclerosis com- approach to developmental psychopathology. Int J Methods Psy- plex. Eur J Hum Genet 20:510–515. doi:10.1038/ejhg.2011.241 chiatr Res 16:S16–S23. doi:10.1002/mpr 37. Bayley N (1993) Bayley scales of infant development, vol 2. Psy- 17. Constantino JN (2011) The quantitative nature of autis- chological Corporation, San Antonio tic social impairment. Pediatr Res 69:55–62. doi:10.1203/ 38. Bayley N (2006) Bayley scales of infant and toddler development, PDR.0b013e318212ec6e vol 3. Harcourt Assessment, San Antonio 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 765 39. Scholte E, van Duijn G, Dijkxhoorn Y et al (2008) Handleiding tuberous sclerosis complex. J Neurol 257:1373–1381. Vineland Screener 0–6 [Manual Vineland Screener 0–6]. PITS, doi:10.1007/s00415-010-5535-2 Leiden 47. Chu-Shore CJ, Major P, Montenegro M, Thiele E (2009) Cyst- 40. IBM Corp (2012) IBM SPSS Statistics for Windows, version 21.0 like tubers are associated with TSC2 and epilepsy in tuberous 41. Galwey NW (2009) A new measure of the effective number of sclerosis complex. Neurology 72:1165–1169. doi:10.1212/01. tests, a practical tool for comparing families of non-independent wnl.0000345365.92821.86 significance tests. Genet Epidemiol 33:559–568. doi:10.1002/ 48. Curatolo P, Napolioni V, Moavero R (2010) Autism spec- gepi.20408 trum disorders in tuberous sclerosis: pathogenetic pathways 42. Preacher KJ, Hayes AF (2008) Asymptotic and resampling strat- and implications for treatment. J Child Neurol 25:873–880. egies for assessing and comparing indirect effects in multiple doi:10.1177/0883073810361789 mediator models. Behav Res Methods 40:879–891. doi:10.3758/ 49. O’Connor K, Kirk I (2008) Brief report: atypical social cognition BRM.40.3.879 and social behaviours in autism spectrum disorder: A different 43. Preacher KJ, Kelley K (2011) Effect size measures for mediation way of processing rather than an impairment. J Autism Dev Dis- models: quantitative strategies for communicating indirect ee ff cts. ord 38:1989–1997. doi:10.1007/s10803-008-0559-5 Psychol Methods 16:93–115. doi:10.1037/a0022658 50. Overwater IE, Verhaar BJH, Lingsma HF et al (2017) Interde- 44. Gallagher A, Madan N, Stemmer-Rachamimov A, Thiele EA pendence of clinical factors predicting cognition in children with (2010) Progressive calcified tuber in a young male with tuber - tuberous sclerosis complex. J Neurol 264:161–167 ous sclerosis complex. Dev Med Child Neurol 52:1062–1065. 51. Hirota T, Veenstra-Vanderweele J, Hollander E, Kishi T (2014) doi:10.1111/j.1469-8749.2010.03792.x Antiepileptic medications in autism spectrum disorder: a system- 45. Koh S, Jayakar P, Dunoyer C et al (2000) Epilepsy surgery in chil- atic review and meta-analysis. J Autism Dev Disord 44:948–957. dren with tuberous sclerosis complex: presurgical evaluation and doi:10.1007/s10803-013-1952-2 outcome. Epilepsia 41:1206–1213. doi:10.1111/j.1528-1157.2000. 52. Tuchman R (2004) AEDs and psychotropic drugs in children with tb00327.x autism and epilepsy. Ment Retard Dev Disabil Res Rev 10:135– 46. Gallagher A, Grant EP, Madan N et  al (2010) MRI find- 138. doi:10.1002/mrdd.20026 ings reveal three different types of tubers in patients with 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Child & Adolescent Psychiatry Springer Journals

Cortical dysplasia and autistic trait severity in children with Tuberous Sclerosis Complex: a clinical epidemiological study

Free
13 pages

Loading next page...
 
/lp/springer_journal/cortical-dysplasia-and-autistic-trait-severity-in-children-with-WJQkjy3plP
Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s)
Subject
Medicine & Public Health; Psychiatry
ISSN
1018-8827
eISSN
1435-165X
D.O.I.
10.1007/s00787-017-1066-z
Publisher site
See Article on Publisher Site

Abstract

Eur Child Adolesc Psychiatry (2018) 27:753–765 https://doi.org/10.1007/s00787-017-1066-z ORIGINAL CONTRIBUTION Cortical dysplasia and autistic trait severity in children with Tuberous Sclerosis Complex: a clinical epidemiological study 1,2 2,3 1,2 1 Sabine E. Mous  · Iris E. Overwater  · Rita Vidal Gato  · Jorieke Duvekot  · 1,2 4 2,3 Leontine W. ten Hoopen  · Maarten H. Lequin  · Marie‑Claire Y. de Wit  · 1,2 Gwendolyn C. Dieleman   Received: 3 March 2017 / Accepted: 9 October 2017 / Published online: 23 October 2017 © The Author(s) 2017. This article is an open access publication Abstract Tuberous Sclerosis Complex (TSC) is charac- and the relevance of separately studying the two ASD terized by a high prevalence of autism spectrum disorders subdomains. (ASD). Little is known about the relation between corti- cal dysplasia and ASD severity in TSC. We assessed ASD Keywords Tubers · Radial migration lines · Autism · severity (using the Autism Diagnostic Observation Scale), Quantitative autistic traits · Intelligence · Cognitive tuber and radial migration line (RML) count and location, functioning and cognitive functioning in 52 children with TSC and performed regression and mediation analyses. Tuber and RML count were strongly positively related to ASD sever- Introduction ity. However, when correcting for cognitive functioning, the majority of associations became insignificant and only total Tuberous Sclerosis Complex (TSC) is an autosomal domi- tuber count remained associated to the severity of restricted/ nant disorder affecting 1 in 6000 people, caused by inacti - repetitive behaviors. Occipital RML count remained associ- vating TSC1 (chromosome 9) or TSC2 (chromosome 16) ated with overall ASD severity, and social communication/ variants. The TSC1 and TSC2 protein products form the interaction deficit severity specifically. This study shows the intracellular TSC1-TSC2 protein complex, which serves as important explanatory role of cognitive functioning in the a regulator of the mammalian target of rapamycin (mTOR) association between cortical dysplasia and ASD severity, pathway. Mutations in the TSC1 or TSC2 gene lead to a upregulation of the mTOR pathway, causing uncontrolled cell growth and abnormal differentiation and the prolifera - Electronic supplementary material The online version of this tion of benign overgrowths of cells and tissue in several article (doi:10.1007/s00787-017-1066-z) contains supplementary material, which is available to authorized users. organ systems including the brain, skin, kidneys, heart, eyes, lungs and bones [1]. In the brain this may lead to cortical * Sabine E. Mous dysplasia. The most common form of cortical dysplasia in s.mous@erasmusmc.nl TSC is the presence of cortical tubers, affecting over 80% Department of Child and Adolescent Psychiatry/Psychology, of all TSC patients. Cortical tubers are focal developmen- Erasmus Medical Center-Sophia Children’s Hospital, P.O. tal abnormalities of the cortex, characterized by disorgan- Box 2060, 3000 CB Rotterdam, The Netherlands ized lamination and atypical cellular growth, differentiation ENCORE Expertise Center for Neurodevelopmental and maturation [2, 3], which develop during prenatal brain Disorders, Erasmus Medical Center-Sophia Children’s development and can be detected by MRI from 20 weeks Hospital, P.O. Box 2060, 3000 CB Rotterdam, of gestation onwards. Postnatally, no new tubers arise, but The Netherlands in older children tubers may calcify or become cystic [1]. Department of Pediatric Neurology, Erasmus Medical Another form of cortical dysplasia is the presence of radial Center-Sophia Children’s Hospital, P.O. Box 2060, 3000 CB Rotterdam, The Netherlands migration lines (RMLs). RMLs are linear abnormalities that 4 extend from the ventricles to the cortex, representing areas Department of Radiology, University Medical Center Utrecht, P.O. Box 85500, 3508 GA Utrecht, The Netherlands of hypomyelination and white matter heterotopia [4]. RMLs Vol.:(0123456789) 1 3 754 Eur Child Adolesc Psychiatry (2018) 27:753–765 are a marker of abnormal neural migration and cortical that child psychopathology, such as ASD, might be better organization and are often associated with a tuber, but can described within a quantitative—or dimensional—frame- also be isolated. Both forms of cortical dysplasia may act as work has gained support in the last years [16]. Within this epileptogenic lesions. In Fig. 1, an example of a Magnetic framework of continuous symptom levels, the entire spec- Resonance Imaging (MRI) scan showing cortical tubers and trum of symptom severity is covered. Most likely, this is RMLs is provided. a more naturalistic representation of psychopathology, as Other features associated with brain pathology in TSC compared to the use of all-or-none dichotomous diagnostic include the presence of epilepsy (72–85% of all patients) and categories. Previous studies have suggested that the symp- cognitive impairment, with about 50% of patients having an toms and etiology of ASD indeed form such a spectrum [17], intellectual disability (IQ < 70), and a range of behavioral seemingly even extending into the general population [18]. and psychiatric symptoms. Autism spectrum disorder (ASD) Studying ASD as a quantitative trait (and thus also includ- is highly prevalent in children with TSC, with prevalence ing subclinical traits) rather than as a categorically defined rates estimated around 40–50% [5–8]. disorder can contribute to a better understanding of the dis- Previous studies have suggested the total number of corti- order and the potentially contributing biological pathways. cal tubers to be an important predictor for an ASD diagnosis An additional advantage of the use of quantitative severity in TSC, and the temporal lobes were suggested to be specifi- scores in research is that these continuous scores provide cally implicated [9, 10]. Other studies found the presence more statistical power and allow the application of advanced (yes/no) of temporal tubers to be associated with a higher statistical methods [17]. We found only a single study that likelihood of an ASD diagnosis [11], while others specifi- previously investigated the relation between a quantitative cally found the number of cyst-like tubers to be related to measure of ASD severity and tuber count, which did not find ASD diagnostic status [12] or found a diagnosis of autism to an association between cortical tuber count or location and be related to frontal and posterior tubers [13]. Still others did overall ASD severity [19]. Furthermore, ASD is character- not find an association between the occurrence of cortical ized by various difficulties that, according to the latest edi- tubers and an autism diagnosis [14], or found the number of tion of the Diagnostic and Statistical Manual of Mental Dis- tubers to be equally prevalent in intellectually disabled non- orders (DSM-5) [20], can be divided in two main domains; autistic and intellectually disabled autistic children and thus (1) deficits in social communication and interaction and non-specific for ASD [ 15]. These inconsistencies in n fi dings (2) restricted or repetitive patterns of behavior, interests or point out that the association between tuber burden and ASD activities. The nature of the symptoms in these domains is is still poorly understood. substantially different and symptom severity is not necessar - Although these previous studies have experimented with ily equal in both domains, making it plausible that distinct different ways of defining cortical tuber involvement (i.e. mechanisms in different brain regions may underlie these tuber presence (yes/no), tuber count, or tuber size), in these two different ASD symptom domains. Therefore, it would studies ASD has always only been categorically defined as be useful to not only study overall ASD severity, but also the presence or absence of an ASD diagnosis. The notion study the association between symptom severity in these Fig. 1 Example of T2-weighted images with arrows indicating a cortical tubers, and b radial migration lines 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 755 two distinct domains and tuber burden and location in TSC CSS for the two separate subdomains of the ADOS (social patients. affect (SA) and restricted and repetitive behaviors (RRB) In addition, it is known that TSC is characterized by a [29–31]), providing an indication of ASD severity rela- wide range of cognitive abilities and it has been shown that tive to the child’s age and expressive language level [31]. the severity of cognitive impairment is related to tuber bur- The ADOS CSS is a truly continuous measure, covering den [11, 21–23]. Similarly, an association between intellec- the entire spectrum of autistic traits. CSS range between 1 tual (dis)ability and autism severity has been demonstrated and 10, with lower scores indicating no to very little prob- [24]. It is of interest, but still unclear, whether tuber burden lems, and higher scores indicating severe ASD symptoms. is an independent factor in determining autism severity or if Raw ADOS total scores corresponding to an ADOS clas- cognitive impairment is an important (potentially mediating) sification ‘Non-spectrum’ are distributed across CSS 1–3, determinant in this association. ‘ASD’ across CSS 4–5, and ‘Autism’ across CSS 6–10. For Most neuroimaging studies in TSC have focused spe- descriptive purposes, children were classified as having ASD cifically on tuber characteristics and the association with (autism or the broader ASD phenotype) or not according to cognitive or behavioral problems. RMLs can be more dif- the revised ADOS-2 algorithms [27]. ficult to detect, and less is known about the contribution of RMLs to the neurocognitive phenotype in TSC. There are Cognitive functioning data, however, demonstrating that RMLs are associated with intelligence [25, 26], as well as with the severity of autistic Cognitive functioning was assessed using different intel- traits [26]. ligence measures according to best practice standards; in In the present study, we aim to investigate the associa- the majority of children (n = 32, 62%) this was either the tion between cortical dysplasia (i.e. the number and location Wechsler Intelligence Scale for Children-III (WISC-III) of cortical tubers and RMLs) and a clinical observational or Wechsler Preschool and Primary Scale of Intelligence- quantitative measure of ASD severity, and to study the role III (WPPSI-III) [32, 33]. In some children (n = 7, 13%) of cognitive functioning in this association. Furthermore, a non-verbal intelligence test was used, which was either we aim to investigate the specific association of tuber and the Wechsler Non Verbal scale of ability (WNV) [34] or RML count and location and ASD severity within the two Snijders-Oomen Nonverbal Intelligence Test (SON-R) [35]. main domains of ASD symptomatology (deficits in social From all intelligence tests full-scale intelligence quotients communication and interaction, and restricted or repetitive (TIQs) were used. For children who were at the floor of their behaviors). age-appropriate standardized scores, a developmental quo- tient (DQ) was calculated (developmental age/chronological age × 100) [36]. Like IQ scores, a DQ of 100 is considered Methods the mean. In a number of children (n = 13, 25.0%) no formal intelligence test could be performed due to a young (devel- Participants opmental) age. In these children the cognitive developmental age was evaluated using one of the Bayley Scales of Infant The medical records of all pediatric TSC patients in care at and Toddler Development (BSID-II or Bayley-III [37, 38]) the expertise center ENCORE (Erasmus MC-Sophia Chil- or the Vineland Screener [39]. Again, we used the estimated dren’s Hospital, Rotterdam, the Netherlands) were retrospec- cognitive developmental age to calculate a DQ, according to tively reviewed. In 75 patients, MRI scans of the brain were the formula provided above. available for review. Of this sample, 52 patients (24 boys, 28 girls, 2–17 years of age) also had data on ASD severity and cognitive functioning. Neuroimaging Measures Brain MRI scans were made on a 1.5 Tesla General Electric scanner. For children with more than one available MRI, the Autism spectrum symptoms MRI closest in time to the ADOS assessment was selected. All MRI scans were visually inspected by two trained medi- The severity of ASD was assessed using the Autism Diag- cal students, and re-assessed by a pediatric neuroradiologist nostic Observation Scale (ADOS) [27, 28]. All ADOS and a pediatric neurologist. A protocol for data collection assessments were performed and scored by a trained, expe- was developed in which fluid-attenuated inversion recovery rienced and certified psychologist or psychiatrist. (FLAIR) images were used to assess the number and loca- For our main analyses, a continuous total standardized tion of tubers and RMLs. If FLAIR images were not present, calibrated severity score (CSS) was calculated, as well as a or for clarification purposes, T2-weighted images were used. 1 3 756 Eur Child Adolesc Psychiatry (2018) 27:753–765 Statistical analysis In supplementary Table S1 (online resource) the Pearson correlations between all variables of interest are shown. In Data were analyzed in IBM SPSS Statistics version 21 supplementary Figure S1 (online resource) the distribution [40]. Associations between the various variables of interest of the ADOS total CSS by IQ/DQ is shown, split by ADOS were studied calculating Pearson correlation coefficients. classification. To investigate the association between ASD severity and total tuber and RML count, linear regression analyses were Association ASD severity and tuber count performed. When these yielded significant results, post hoc analyses were performed, assessing the associations in the The association between the ADOS total severity score separate lobes of the brain. A Bonferroni correction was and total tuber count was highly significant (β  =  0.46, applied to correct for multiple testing. Because of the con- p < 0.001), and about 20% of the variance in the ADOS total siderable intercorrelations between the number of tubers severity score could be explained by the total number of cor- or RMLs in the separate lobes (ranging between 0.63–0.78 tical tubers. Post-hoc analyses assessing the separate lobes and 0.32–0.51, respectively), we first calculated the effec- of the brain indicated similar results for all lobes (Table 2). tive number of tests and adjusted the Bonferroni correction Because IQ/DQ was significantly related to both the accordingly to account for this lack of independence [41]. ADOS total severity score and the total number of tubers The calculation yielded an effective number of 2.98 tests (as well as to the two separate ADOS subdomain scores and for the tuber analyses and 3.57 tests for the RML analyses. the number of tubers in all separate lobes) (supplementary In all tables, both uncorrected as well as Bonferroni cor- Table S1, online resource), the analyses were repeated with rected (p ) p-values are provided, as well as β and adjusted IQ/DQ added as a covariate. The results of these analyses corr R effect size measures. Supplementary linear regression show that this correction rendered all associations insignifi- analyses were performed studying the association of ASD cant, indicating that IQ/DQ was an important explanatory severity with the number of cystic and calcified tubers. We variable in the associations (Table 2). also performed supplementary t tests and logistic regression Supplementary analyses studying the association between analyses, using the categorical ADOS classification instead the categorical ADOS classification and tuber count show of the continuous severity scores. similar, but reduced in magnitude, results (supplementary To assess whether IQ/DQ was a mediator in the associa- Tables S2 and S3 (online resource). tion between ASD severity and tuber or RML count, formal Next, we studied the association between total tuber count mediation analyses were performed using the ‘PROCESS’ and ASD severity in the two subdomains of the ADOS; macro for SPSS, version 2.15 (http://www.afhayes.com/) social affect (SA) and restricted and repetitive behaviors with bias-corrected bootstrapping using 1000 replications (RRB) (Table  3). Again, strong associations were found [42]. For the mediation analyses, effect sizes are reported as between the total number of tubers and ADOS SA and κ , with values of 0.01, 0.09 and 0.25 considered as small, RRB severity scores (β  =  0.37, p = 0.008 and β  =  0.49, medium and large respectively [43]. Finally, supplementary p < 0.001), and 12 and 22% of the variance in respectively mediation analyses were performed to study the role of epi- SA and RRB severity score was explained by total tuber lepsy in the association between ASD severity and tuber or count. RML burden. After adding IQ/DQ as a covariate, the total number of tubers only remained significantly associated with the RRB severity score (β  =  0.29, p  =  0.046). Post-hoc analyses Results studying the separate lobes of the brain indicated that this association was mainly driven by tuber count in the frontal Patient characteristics lobes (β = 0.30, p = 0.042, p  = 0.124). An (uncorrected) corr trend was visible for the association between temporal tuber In total, the data of 52 TSC patients (24 boys, 28 girls) were count and RRB severity (β = 0.25, p = 0.071, p  = 0.211). corr included. The mean age at time of the MRI was 7.0 years To formally assess whether IQ/DQ was a mediator in (range 0–17) and the mean age at time of the ADOS assess- the association between the total number of tubers and the ment was 8.8 years (range 2–17). The total number of tubers ADOS total severity score, a mediation analysis was per- ranged between 0 and 81 and the total number of RMLs formed (Fig.  2, panel a). The mediation analysis showed between 0 and 37. The largest number of tubers and RMLs that the direct effect (c’ path) of total tuber count on the were located in the frontal lobes. According to the ADOS, total severity score was insignificant. The indirect (a*b path) a total number of 25 children (48.1%) met the criteria for effect through IQ/DQ was large and statistically significant an ASD classification. Additional patient characteristics are (B = 0.04, 95% CI = 0.02; 0.06, κ  = 0.263). This implies shown in Table 1. that the total effect (c path) between total tuber count and 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 757 Table 1 Patient characteristics n (%) Mean (SD) Min–Max Gender, male 24 (46.2) Age in years  During MRI 7.0 (3.9) 0–17  During ADOS 8.8 (4.2) 2–17 Age difference in years between ADOS-MRI 1.8 (2.9) 0–14 Mutation  TSC1 14 (26.9)  TSC2 36 (69.2)  No mutation identified 1 (1.9)  Not tested 1 (1.9) Epilepsy  Present, yes 46 (88.5)  West syndrome, yes (from n = 46) 16 (34.8)  Age of onset, months (from n = 46) 15.4 (18.6) 1–91 Tubers  Present, yes 51 (98.1)  Number   Total 27.5 (20.2) 0–81   Frontal lobe 16.0 (12.1) 0–54   Parietal lobe 5.5 (4.6) 0–19   Temporal lobe 3.7 (3.3) 0–12   Occipital lobe 2.4 (2.6) 0–9 Cystic tubers  Present, yes 19 (36.5)  Number   Total 1.85 (3.88) 0–18   Frontal lobe 0.96 (2.21) 0–10   Parietal lobe 0.44 (1.18) 0–7   Temporal lobe 0.33 (0.98) 0–4   Occipital lobe 0.12 (0.32) 0–1 Calcified tubers  Present, yes 9 (17.3)  Number   Total 0.71 (2.52) 0–17   Frontal lobe 0.33 (1.32) 0–9   Parietal lobe 0.15 (0.75) 0–5   Temporal lobe 0.06 (0.31) 0–2   Occipital lobe 0.17 (0.62) 0–3 Radial migration lines  Present, yes 51 (98.1)  Number   Total 16.0 (10.2) 0–37   Frontal lobe 8.2 (5.6) 0–23   Parietal lobe 3.9 (3.3) 0–15   Temporal lobe 3.0 (3.0) 0–12   Occipital lobe 0.9 (1.2) 0–4 IQ/DQ 59.7 (24.5) 8–114 Intellectual disability (IQ/DQ < 70), yes 34 (65.4) ADOS module  Module 1 15 (28.8)  Module 2 11 (21.2) 1 3 758 Eur Child Adolesc Psychiatry (2018) 27:753–765 Table 1 (continued) n (%) Mean (SD) Min–Max  Module 3 18 (34.6)  Module 4 8 (15.4) ADOS classification  Non-spectrum 27 (51.9)  Autism spectrum disorder 25 (48.1) ADOS calibrated severity score  Total 4.0 (2.7) 1–10  Social affect domain 4.3 (2.6) 1–10  Restricted and repetitive behaviors domain 4.8 (2.8) 1–10 n = 52 ADOS Autism Diagnostic Observation Scale, DQ developmental quotient, IQ intelligence quotient, MRI magnetic resonance imaging Raw ADOS total scores corresponding to an ADOS classification of ‘Autism’ were distributed across cali- brated severity scores (CSS) 6–10, ‘ASD’ across 4–5 and ‘Non-spectrum’ across 1–3 Table 2 Association ADOS total calibrated severity score and tuber count Model I  Model I + IQ/DQ a a 2 2 B 95% CI β p p B 95% CI β p p R R corr corr adj adj Total number of tubers 0.06 0.03; 0.09 0.46 < 0.001 – 0.195 0.02 − 0.01; 0.06 0.18 0.188 – 0.356 Frontal lobes 0.09 0.03; 0.15 0.41 0.002 0.007 0.155 0.03 − 0.04; 0.09 0.11 0.414 1 0.341 Parietal lobes 0.24 0.10; 0.39 0.43 0.002 0.005 0.166 0.11 − 0.04; 0.25 0.19 0.150 0.447 0.360 Temporal lobes 0.31 0.10; 0.52 0.38 0.005 0.016 0.129 0.11 − 0.09; 0.31 0.14 0.278 0.829 0.348 Occipital lobes 0.40 0.13; 0.67 0.39 0.004 0.012 0.136 0.24 − 0.00; 0.47 0.23 0.054 0.160 0.382 n = 52 ADOS Autism Diagnostic Observation Scale, IQ intelligence quotient, DQ developmental quotient, R adjusted R squared model adj Multiple testing correction (2.98 effective tests) applied Significant associations are highlighted in bold font the total severity score was fully mediated by IQ/DQ. Post- count on the RRB severity score also remained significant hoc mediation analyses for the separate lobes were also per- (B  =  0.04, p = 0.046). This means that a direct effect of formed (figures not shown), also showing full mediation by total tuber count on the RRB severity score was present and IQ/DQ for all lobes. that the total effect (c path) was only partly mediated by IQ/ Next, mediation analyses were performed studying the DQ. Again, in line with the regression analyses, post hoc role of IQ/DQ in the association between total tuber count analyses of the separate lobes (figures not shown) indicated and the severity score of the two separate ADOS domains. full mediation by IQ/DQ, with the exception of the frontal For the SA domain (Fig. 2, panel b), the direct effect (c’ lobes. For the frontal lobe association with the RRB score, path) of total tuber count on the SA severity score was insig- the indirect effect (a*b path) through IQ/DQ was medium nificant. The indirect effect (a*b path) through IQ/DQ was to large and significant (B  = 0.05, 95% CI = 0.01; 0.09, medium to large and statistically significant (B  = 0.03, 95% κ  = 0.188), but the direct effect (c’ path) between frontal CI = 0.02; 0.05, κ  = 0.242). Again, this means that the lobe tuber count and the RRB severity score remained sig- total effect (c path) between total tuber count and the SA nificant as well (B  = 0.07, p = 0.042), again implying only severity score was fully mediated by IQ/DQ. However, in partial mediation by IQ/DQ. line with the regression analyses, the mediation analysis per- Because cystic and calcified tubers have been suggested formed with the RRB domain (Fig. 2, panel c) showed that to be more epileptogenic and related to a more severe phe- although the indirect effect (a*b path) through IQ/DQ was notype [44–47], we additionally studied the associations medium to large and significant (B  = 0.03, 95% CI = 0.01; between the ADOS severity scores and the number of cystic 0.05, κ  = 0.189), the direct effect (c’ path) of total tuber and calcified cortical tubers (supplementary Tables S4 and 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 759 Table 3 Association ADOS subdomain calibrated severity scores and tuber coufnt Model I Model I + IQ/DQ a a 2 2 B 95% CI β p p B 95% CI β p p R R corr corr adj adj Total number of tubers  SA domain CSS 0.05 0.01; 0.08 0.37 0.008 – 0.117 0.01 − 0.02; 0.05 0.10 0.497 – 0.262  RRB domain CSS 0.07 0.03; 0.10 0.49 <0.001 – 0.224 0.04 0.00; 0.08 0.29 0.046 – 0.299 Frontal lobes  SA domain CSS 0.07 0.01; 0.12 0.32 0.023 0.069 0.081 0.00 − 0.06; 0.07 0.02 0.882 1 0.255  RRB domain CSS 0.12 0.06; 0.17 0.49 < 0.001 0.001 0.229 0.07 0.00; 0.14 0.30 0.042 0.124 0.301 Parietal lobes  SA domain CSS 0.22 0.08; 0.36 0.40 0.003 0.010 0.141 0.10 − 0.05; 0.25 0.19 0.169 0.503 0.283  RRB domain CSS 0.22 0.06; 0.38 0.36 0.008 0.025 0.114 0.09 − 0.08; 0.26 0.15 0.275 0.821 0.257 Temporal lobes  SA domain CSS 0.20 − 0.01; 0.41 0.26 0.064 0.192 0.048 0.02 − 0.19; 0.23 0.02 0.876 1 0.255  RRB domain CSS 0.37 0.15; 0.58 0.43 0.001 0.004 0.168 0.21 − 0.02; 0.43 0.25 0.071 0.211 0.288 Occipital lobes  SA domain CSS 0.34 0.08; 0.61 0.35 0.012 0.036 0.102 0.20 − 0.05; 0.45 0.20 0.111 0.330 0.293  RRB domain CSS 0.34 0.04; 0.63 0.31 0.026 0.079 0.077 0.18 − 0.10; 0.46 0.16 0.201 0.598 0.264 n = 52 ADOS Autism Diagnostic Observation Scale, CSS calibrated severity score, SA social affect, RRB  restricted and repetitive behaviors, IQ intel- ligence quotient, DQ developmental quotient, R adjusted R squared model adj Multiple testing correction (2.98 effective tests) applied Significant associations are highlighted in bold font Fig. 2 Mediation analyses tuber count, ASD severity score and IQ/DQ. a ADOS total severity score, b ADOS Social Affect (SA) domain sever - ity score, c ADOS Restricted and Repetitive Behaviors (RRB) domain severity score S5, online resource). These analyses showed that, after cor- (β = 0.27, p = 0.020). This association was strongest in, and rection for IQ/DQ, only the number of calcified tubers was mainly driven by, the occipital lobe (β = 0.33, p = 0.003, significantly associated with the ADOS total severity score p  = 0.009) (Table S4). When studying the two ADOS corr 1 3 760 Eur Child Adolesc Psychiatry (2018) 27:753–765 subdomain scores, we found that after correction for IQ/DQ to the two separate ADOS subdomain scores and the number the total number of calcified tubers was significantly related of RMLs in the frontal and parietal lobes) (supplementary to the RRB subdomain severity score (β = 0.32, p = 0.010). Table S1, online resource), the analyses were repeated with Post-hoc analyses studying the separate lobes of the brain IQ/DQ added as a covariate. After the correction for IQ/DQ revealed that this association was strongest in the occipital only the association for the occipital lobe remained significant lobes (β  =  0.29, p  =  0.016, p  = 0.046), but also pre- (β = 0.28, p = 0.013, p  = 0.047) (Table 4). corr corr sent in the parietal and temporal lobes (β = 0.26, p = 0.033, Supplementary analyses studying the association between p  = 0.094, and β = 0.29, p = 0.021, p  = 0.060 respec- the categorical ADOS classification and the number of radial corr corr tively) (Table S5). Correction for age during MRI did not migration lines were performed, showing similar effects, but change these results. considerably reduced in magnitude and mostly insignificant Because studies have suggested that an early onset of (supplementary Tables S6 and S7 (online resource). epilepsy may have a deleterious influence on early brain Next, we studied the association between RML count and development and may act as a risk factor for autism [48], ASD severity in the two subdomains of the ADOS; social supplementary mediation analyses were performed to study affect (SA) and restricted and repetitive behaviors (RRB) the role of epilepsy in the association between tuber bur- (Table 5). Again, highly significant associations were found den and ASD severity (supplementary Figure S2, online between total RML count and ADOS SA and RRB severity resource). The results of these multiple mediation models (β = 0.41, p = 0.002 and β = 0.33, p = 0.016), and 16% and (that simultaneously included age of epilepsy onset and IQ/ 9% of the variance in respectively SA and RRB severity DQ as mediators) show that the association between tuber score was explained by total RML count. burden and ASD severity was not mediated by age of epi- After adding IQ/DQ as a covariate, only the number of lepsy onset (path a2*b2: B = 0.01, 95% CI = − 0.01;0.03), RMLs in the occipital lobes remained significantly asso- and fully mediated by IQ/DQ (path a1*b1: B = 0.04, 95% ciated with the SA severity score (β  =  0.30, p  =  0.013, CI = 0.02; 0.06). Similar results were found when separately p  = 0.047). corr analyzing the two ADOS subdomains. Again, a formal mediation analysis was performed, assessing whether IQ/DQ was a mediator in the associa- Association ASD severity and radial migration line tion between the total number of RMLs and the ADOS count total severity score (Fig. 3, panel a). The mediation analy- sis showed that the direct effect (c’ path) of total RML A significant association was found between the ADOS total count on the total severity score was insignificant. The severity score and the total number of radial migration lines indirect effect (a*b path) through IQ/DQ was medium (RMLs) (β = 0.40, p = 0.003), and 15% of the variance in to large and statistically significant (B  =  0.06, 95% ADOS total severity score was accounted for by the total num- CI = 0.03; 0.10, κ  = 0.223). This implies that the total ber of radial migration lines. Post-hoc analyses assessing the effect (c path) between total RML count and the total separate lobes of the brain indicated significant associations severity score was fully mediated by IQ/DQ. Post-hoc for the parietal lobes and occipital lobes specifically (Table  4). mediation analyses for the frontal and parietal lobes were Because IQ/DQ was significantly related to both the ADOS performed as well (figures not shown), also showing full total severity score and the total number of RMLs (as well as mediation by IQ/DQ. Since IQ/DQ was not significantly Table 4 Association ADOS total calibrated severity score (CSS) and the number of radial migration lines (RMLs) Model I Model I + IQ/DQ a a 2 2 B 95% CI β p p R B 95% CI β p p R corr corr adj adj Total number of RMLs 0.11 0.04;0.17 0.40 0.003 – 0.146 0.05 − 0.02; 0.11 0.19 0.141 – 0.362 Frontal lobes 0.14 0.02;0.27 0.31 0.025 0.090 0.078 0.04 − 0.08; 0.15 0.08 0.505 1 0.338 Parietal lobes 0.28 0.07;0.49 0.35 0.011 0.038 0.106 0.14 − 0.05; 0.33 0.18 0.140 0.501 0.362 Temporal lobes 0.21 − 0.04; 0.46 0.24 0.094 0.336 0.036 0.10 − 0.11; 0.31 0.11 0.364 1 0.344 Occipital lobes 0.90 0.30;1.50 0.39 0.004 0.015 0.135 0.65 0.14; 1.16 0.28 0.013 0.047 0.412 n = 52 ADOS Autism Diagnostic Observation Scale, IQ intelligence quotient, DQ developmental quotient, R adjusted R squared model adj Multiple testing correction (3.57 effective tests) applied Significant associations are highlighted in bold font 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 761 Table 5 Association ADOS subdomain calibrated severity scores (CSS) and the number of radial migration lines (RMLs) Model I Model I + IQ/DQ a a 2 2 B 95% CI β p p B 95% CI β p p R R corr corr adj adj Total number of RMLs  SA domain CSS 0.10 0.04;0.17 0.41 0.002 – 0.155 0.06 − 0.01; 0.12 0.23 0.080 – 0.300  RRB domain CSS 0.09 0.02; 0.17 0.33 0.016 – 0.093 0.04 − 0.04; 0.11 0.14 0.300 – 0.255 Frontal lobes  SA domain CSS 0.15 0.03; 0.27 0.33 0.017 0.061 0.091 0.06 − 0.06; 0.18 0.14 0.296 1 0.271  RRB domain CSS 0.13 − 0.00; 0.27 0.27 0.057 0.203 0.052 0.03 − 0.10; 0.17 0.07 0.608 1 0.243 Parietal lobes  SA domain CSS 0.27 0.07; 0.47 0.35 0.010 0.037 0.107 0.15 − 0.04; 0.35 0.20 0.110 0.394 0.293  RRB domain CSS 0.21 − 0.03; 0.44 0.25 0.081 0.288 0.041 0.07 − 0.14; 0.29 0.09 0.504 1 0.246 Temporal lobes  SA domain CSS 0.20 − 0.04; 0.44 0.23 0.101 0.359 0.034 0.10 − 0.11; 0.32 0.12 0.343 1 0.268  RRB domain CSS 0.24 − 0.03; 0.50 0.25 0.075 0.269 0.043 0.13 − 0.10; 0.37 0.14 0.262 0.935 0.258 Occipital lobes SA domain CSS 0.87 0.29; 1.45 0.39 0.004 0.015 0.137 0.66 0.14; 1.18 0.30 0.013 0.047 0.343 RRB domain CSS 0.64 − 0.03; 1.31 0.26 0.062 0.222 0.049 0.40 − 0.20; 1.00 0.16 0.186 0.665 0.266 n = 52 ADOS Autism Diagnostic Observation Scale, SA social affect, RRB restricted and repetitive behaviors, IQ intelligence quotient, DQ developmen- tal quotient, R adjusted R squared model adj Multiple testing correction (3.57 effective tests) applied Significant associations are highlighted in bold font Fig. 3 Mediation analyses RML count, ASD severity score and IQ/DQ. a ADOS total severity score, b ADOS Social Affect (SA) domain sever - ity score, c ADOS Restricted and Repetitive Behaviors (RRB) domain severity score related to RML count in the temporal and occipital lobes, score could not be mediated by IQ/DQ, implying a direct the earlier identified association between total number of effect of occipital lobe RML count on the total ADOS RMLs in the occipital lobes and the ADOS total severity severity score. 1 3 762 Eur Child Adolesc Psychiatry (2018) 27:753–765 As with tuber count, the mediation analyses were also from the fact that the frontal lobes represent the largest brain performed studying the role of IQ/DQ in the association area of all lobes. When studying the association between between total RML count and the severity score of the two RML count and ASD severity we found similar results, again separate ADOS subdomains (Fig. 3, panel b and c). For both initially showing strong associations between the number of subdomains, the direct effect (c’ path) of total RML count RMLs and ASD severity, but which were for the most part on the subdomain severity score was insignificant. The indi- rendered insignificant when corrected for IQ/DQ, except for rect effect (a*b path) through IQ/DQ was medium to large the relation between the number of RMLs in the occipital and statistically significant (B = 0.05, 95% CI = 0.02; 0.09, lobes and total and social affect ASD severity. The formal κ  = 0.184 for SA domain, and B = 0.05, 95% CI = 0.02; mediation analyses confirmed all results and showed that, 0.10, κ  = 0.190 for RRB domain). This implies that the indeed, all other initial findings were fully mediated by IQ/ total effect (c path) between total RML count and the SA DQ. and RRB severity scores was fully mediated by IQ/DQ. Post- The results emphasize the importance of taking cognitive hoc analyses of the separate lobes (figures not shown) again functioning into account when studying the relation between showed full mediation by IQ/DQ for the frontal and parietal brain pathology and ASD in patients with TSC. TSC is lobes for both the SA and RRB subdomains. Because IQ/DQ often characterized by cognitive impairment, and previous was not significantly related to RML count in the temporal studies have shown that cognitive impairment is strongly and occipital lobes, the earlier found association between related to both brain pathology [11, 21–23] and ASD sever- RML count in the occipital lobes and the SA severity score ity [24], thereby acting as an important confounding (or could not be mediated by IQ/DQ, implying a direct effect of rather explanatory) factor in this association. However, one occipital lobe RML count on the SA severity score. should also realize that, regardless of the explanatory role of Finally, supplementary mediation analyses were per- cognitive functioning, ASD symptoms remain a significant formed to study the effect of epilepsy on the association problem in patients with TSC. We found a direct association, between the number of RMLs and ASD severity (sup- regardless of cognitive functioning, between RML count in plementary Figure S3, online resource). These multiple the occipital lobes and the severity of problems in social mediation models (simultaneously including age of epi- communication and interaction. The main function of the lepsy onset and IQ/DQ as mediators) show that the asso- occipital lobes is processing visual stimuli, and structural ciation between RML count and ASD severity was not and functional abnormalities in the occipital and occipito- mediated by age of epilepsy onset (path a2*b2: B = 0.01, temporal regions have been frequently reported in ASD [49]. 95% CI = − 0.00;0.03), and was fully mediated by IQ/DQ One might argue that, next to IQ, epilepsy severity could (path a1*b1: B = 0.05, 95% CI = 0.02;0.10). Similar results be an important confounding/explanatory factor in the asso- were obtained when separately analyzing the two ADOS ciation between cortical dysplasia and autistic trait severity. subdomains. A recent study by our group has shown that, in a multivari- able model including various epilepsy severity indicators, age of epilepsy onset was the only significant predictor for Discussion cognitive functioning later in life [50]. Therefore, we ran supplementary mediation analyses additionally including In the current clinical epidemiological study, the association this variable (as proxy for epilepsy severity) as mediator. between cortical dysplasia and a quantitative observational The results of these analyses showed that, although there measure of ASD severity was studied in a clinical sample of was a significant association between the number of tubers children with TSC. The specificity of the association with and the age of epilepsy onset, the age of epilepsy onset was the two main subdomains of ASD symptomatology (deficits not related to any of the ASD severity scores or the number in social communication and interaction, and restricted or of RMLs, and was no mediator in the associations studied. repetitive behaviors) was studied as well. Finally, we focused Because the large majority of our sample (46/52, 88.5%) on the role of cognitive functioning in these associations. was using anti-epileptic drugs (AEDs), this sample does not The initial analyses, not corrected for IQ/DQ, showed that allow us to study the effect of AED use on the described total cortical tuber count, as well as tuber count in the sepa- associations. Although the exact effect of AED use on ASD rate lobes of the brain, was strongly related to the severity severity remains unclear due to a small number of studies of ASD, visible in both ASD subdomains. However, when and limited sample sizes, it has been suggested that the use IQ/DQ was added as a covariate to the analyses, only total of AEDs may have a beneficial, but most likely very small, and frontal tuber count remained related to the severity of effect on ASD severity [ 51, 52]. If this is indeed the case, restricted and repetitive behaviors, although it must be noted the use of AEDs might have attenuated our results. To assist that the frontal association did not survive correction for in further elucidating the association between epilepsy and multiple testing and that the relationship may partly arise ASD severity in TSC, future studies might not only consider 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 763 to study the effect of AEDs on ASD severity, but also the findings is enhanced by referring all TSC patients within our effect of other epilepsy indicators such as infantile spasms, expertise center for a developmental and psychiatric evalua- epilepsy refractoriness, and current epilepsy status. tion (regardless of whether or not the child experiences cog- The difference in findings between our and other stud- nitive or behavioral difficulties), the risk of residual selec - ies (but also between previous studies) can most likely be tion bias remains; a first selection takes place when parents explained by large differences in methodology, such as (1) decide whether or not to visit the expertise center with their different ASD measures (i.e. clinical observational meas- child, and a second selection occurs when parents decide ure vs. screening questionnaire or clinical diagnosis, and whether or not they want to visit the department of Child continuous severity scores vs. dichotomous diagnostic cat- and Adolescent Psychiatry/Psychology for a developmental egories), (2) different ways of defining brain involvement and psychiatric evaluation. (i.e. tuber/RML count vs. volume or absence/presence of To conclude, our study initially showed strong associa- tubers/RMLs), (3) different statistical techniques, (4) the tions between cortical dysplasia and ASD severity, with in- or exclusion of confounding variables in the statistical children with more cortical tubers and RMLs having more models and (5) participant selection. severe ASD symptoms. However, for the majority of these A strength and novel aspect of our study is the use of a associations, cognitive functioning was identified as an quantitative measure of ASD severity. Not only does this important confounding—or rather explanatory—factor, approach provide a more naturalistic representation of ASD highlighting the importance of taking cognitive function- symptoms and more statistical power [17], it also allowed us ing into account when studying the relation between brain to study the two different main domains of ASD symptoma- pathology and ASD symptomatology. Regardless of cogni- tology; difficulties in social communication and interaction, tive functioning, children with more tubers overall showed and restricted and repetitive behaviors. Another strength of more severe restrictive and repetitive behaviors, and children this study is the use of a standardized observational measure with more RMLs in the occipital lobes specifically showed of ASD, thereby reducing reporter bias. It must be noted more difficulties in social communication and interaction. that, although the ADOS is an instrument aiming at meas- These findings underline the importance of separately study - uring autistic traits, it remains unclear whether these traits ing problems in social communication and interaction on (especially in non-spectrum patients) truly originate from an the one hand, and restricted and repetitive behaviors on the ASD predisposition or are caused by other factors that might other hand. affect social behavior and restricted or repetitive behaviors. Acknowledgements This research was financially supported by the Also, the direction of effect between cortical dysplasia and Sophia Children’s Hospital Fund (Rotterdam, the Netherlands) under autistic trait severity remains unclear. Although it seems grant number SSWO B14-02. Further financial support was provided plausible that the cortical abnormalities have an adverse by the Dutch Brain Foundation (Hersenstichting) and the Dutch Epi- lepsy Foundation (Epilepsiefonds). Funders were not involved in the impact on brain development, consequently leading to more design of the study, nor in data collection, analysis, interpretation or severe ASD symptoms and developmental delay, it might writing the manuscript. The authors thank Karen Bindels-de Heus well be that in fact all result from another shared factor. for patient care and data collection, and Simone Eijk, Emma van der The stepwise approach and correction for multiple test- Ende and Kimberley Hanemaayer for their help in data collection and cleaning. ing in the current study makes it less plausible that findings are false positive, although this cannot be ruled out entirely. Compliance with ethical standards Mitigating this concern however, is the modest sample size in which these results were obtained. This relatively small Ethical approval All procedures performed in studies involving sample size, which is a limitation of our study, might have human participants were in accordance with the ethical standards of reduced the power to reveal relatively subtle effects, poten- the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments. This retrospective study tially resulting in an underestimation of effects. It is of great was approved by the Medical Ethics Committee of the Erasmus Medi- importance that future larger studies attempt to replicate our cal Center, the Netherlands. findings, before any strong conclusions can be made regard- ing the association between cortical dysplasia and ASD Informed consent Written informed consent was formally waived symptom severity. Furthermore, our study employs clinical as there is no patient burden and no privacy concern. MRI scans that were made on an 1.5 Tesla scanner. This might have led to limited power to detect RMLs, and made it Conflict of interest The authors declare that they have no conflict impossible to accurately retrospectively measure the volume of interest. of cortical tubers, thereby preventing us from studying the relation between ASD severity and tuber volume or tuber- Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://crea- brain proportion. Finally, even though the risk of selection tivecommons.org/licenses/by/4.0/), which permits unrestricted use, bias in our sample is reduced and the generalizability of 1 3 764 Eur Child Adolesc Psychiatry (2018) 27:753–765 distribution, and reproduction in any medium, provided you give appro- 18. Blanken LME, Mous SE, Ghassabian A et al (2015) Cortical mor- priate credit to the original author(s) and the source, provide a link to phology in 6- to 10-year old children with autistic traits: a popu- the Creative Commons license, and indicate if changes were made. lation-based neuroimaging study. Am J Psychiatry 172:479–486. doi:10.1176/appi.ajp.2014.14040482 19. Weber A, Egelhoff J, McKellop J, Franz D (2000) Autism and the cerebellum: evidence from tuberous sclerosis. J Autism Dev Disord 30:511–517 References 20. American Psychiatric Association (2013) Diagnostic and statisti- cal manual of mental disorders (5th ed.). Am J Psychiatry. doi: 1. Crino PB (2013) Evolving neurobiology of tuberous sclero- 10.1176/appi.books.9780890425596.744053 sis complex. Acta Neuropathol 125:317–332. doi:10.1007/ 21. Kaczorowska M, Jurkiewicz E, Domańska-Pakieła D et  al s00401-013-1085-x (2011) Cerebral tuber count and its impact on mental outcome 2. Talos DM, Kwiatkowski DJ, Cordero K et al (2008) Cell-specific of patients with tuberous sclerosis complex. Epilepsia 52:22–27. alterations of glutamate receptor expression in tuberous sclerosis doi:10.1111/j.1528-1167.2010.02892.x complex cortical tubers. Ann Neurol 63:454–465. doi:10.1002/ 22. Kassiri J, Snyder TJ, Bhargava R et al (2011) Cortical tubers, ana.21342 cognition, and epilepsy in tuberous sclerosis. Pediatr Neurol 3. Crino PB, Trojanowski JQ, Dichter MA, Eberwine J (1996) 44:328–332. doi:10.1016/j.pediatrneurol.2011.01.001 Embryonic neuronal markers in tuberous sclerosis: single- 23. O’Callaghan FJK, Harris T, Joinson C et al (2004) The rela- cell molecular pathology. Neurobiology 93:14152–14157. tion of infantile spasms, tubers, and intelligence in tuberous doi:10.1073/pnas.93.24.14152 sclerosis complex. Arch Dis Child 89:530–533. doi:10.1136/ 4. Griffiths PD, Bolton P, Verity C (1998) White matter abnor - adc.2003.026815 malities in tuberous sclerosis complex. Acta Radiol 39:482–486. 24. Hoekstra RA, Happé F, Baron-Cohen S, Ronald A (2009) Asso- doi:10.1080/02841859809172211 ciation between extreme autistic traits and intellectual disabil- 5. Curatolo P, Moavero R, de Vries PJ (2015) Neurological and ity: insights from a general population twin study. Br J Psychia- neuropsychiatric aspects of tuberous sclerosis complex. Lancet try 195:531–536. doi:10.1192/bjp.bp.108.060889 Neurol 14:733–745. doi:10.1016/S1474-4422(15)00069-1 25. Shepherd CW, Houser OW, Gomez MR (1995) MR findings in 6. de Vries PJ, Whittemore VH, Leclezio L et al (2015) Tuberous tuberous sclerosis complex and correlation with seizure devel- Sclerosis Associated Neuropsychiatric Disorders (TAND) and opment and mental impairment. Am J Neuroradiol 16:149–155 the TAND checklist. Pediatr Neurol 52:25–35. doi:10.1016/j. 26. van Eeghen AM, Terán LO, Johnson J et al (2013) The neu- pediatrneurol.2014.10.004 roanatomical phenotype of tuberous sclerosis complex: focus 7. de Vries PJ, Hunt A, Bolton PF et al (2007) The psychopatholo- on radial migration lines. Neuroradiology 55:1007–1014. gies of children and adolescents with tuberous sclerosis complex doi:10.1007/s00234-013-1184-3 (TSC). Eur Child Adolesc Psychiatry 16:16–24. doi:10.1007/ 27. Lord C, Rutter M, DiLavore PC et al (2012) Autism diagnostic s00787-006-0570-3 observation schedule, second edition (ADOS-2). Manual (Part 8. Jeste SS, Varcin KJ, Hellemann GS et al (2016) Symptom profiles I). Western Psychological Services, Los Angeles of autism spectrum disorder in tuberous sclerosis complex. Neu- 28. Lord C, Rutter M, DiLavore PC, Risi S (1999) Autism diagnos- rology 87:766–772. doi:10.1212/WNL.0000000000003002 tic observation schedule (ADOS). Manual. Western Psychologi- 9. Bolton PF, Griffiths PD (1997) Association of tuberous sclero- cal Services, Los Angeles sis of temporal lobes with autism and atypical autism. Lancet 29. Hus V, Gotham K, Lord C (2014) Standardizing ADOS Domain 349:392–395. doi:10.1016/S0140-6736(97)80012-8 Scores: separating severity of social affect and restricted and 10. Huang CH, Peng SSF, Weng WC et al (2015) The relationship repetitive behaviors. J Autism Dev Disord 44:2400–2412. of neuroimaging findings and neuropsychiatric comorbidities in doi:10.1007/s10803-012-1719-1 children with tuberous sclerosis complex. J Formos Med Assoc 30. Gotham K, Pickles A, Lord C (2009) Standardizing ADOS 114:849–854. doi:10.1016/j.jfma.2014.02.008 scores for a measure of severity in autism spectrum disorders. 11. Bolton PF, Park RJ, Higgins JNP et al (2002) Neuro-epileptic J Autism Dev Disord 39:693–705. doi:10.1007/s10803-008- determinants of autism spectrum disorders in tuberous sclerosis 0674-3.Standardizing complex. Brain 125:1247–1255. doi:10.1093/brain/awf124 31. Hus V, Lord C (2014) The Autism diagnostic observation 12. Numis AJ, Major P, Montenegro MA et  al (2011) Identifica- schedule, module 4: revised algorithm and standardized sever- tion of risk factors for autism spectrum disorders in tuber- ity scores. J Autism Dev Disord 44:1996–2012. doi:10.1007/ ous sclerosis complex. Neurology 76:981–987. doi:10.1212/ s10803-014-2080-3 WNL.0b013e3182104347 32. Wechsler D (1991) Wechsler intelligence scale for children, 3rd 13. Curatolo P, Cusmai R, Cortesi F et  al (1991) Neuropsychiat- edn. Psychological Corporation, San Antonio ric aspects of tuberous sclerosis. Ann N Y Acad Sci 615:8–16. 33. Wechsler D (2002) Wechsler preschool and primary scale of intel- doi:10.1111/j.1749-6632.1991.tb37743.x ligence, 3rd edn. Psychological Corporation, San Antonio 14. Walz NC, Byars AW, Egelhoff JC, Franz DN (2002) Supratentorial 34. Wechsler D, Naglieri JA (2006) Wechsler nonverbal scale of abil- tuber location and autism in tuberous sclerosis complex. J Child ity (WNV). Harcourt Assessment, San Antonio Neurol 17:830–832. doi:10.1177/08830738020170111401 35. Tellegen P, Winkel M, Wijnberg-Williams B, Laros J (2005) 15. Asano E, Chugani DC, Muzik O et al (2001) Autism in tuberous Snijders-Oomen Niet-Verbale Intelligentietest (SON-R) 2 1/2–7: sclerosis complex is related to both cortical and subcortical dys- Handleiding en Verantwoording. Boom Testuitgevers, Amsterdam function. Neurology 57:1269–1277. doi:10.1212/WNL.57.7.1269 36. van Eeghen AM, Black ME, Pulsifer MB et al (2012) Genotype 16. Hudziak J, Achenbach T, Althoff R, Pine D (2007) A dimensional and cognitive phenotype of patients with tuberous sclerosis com- approach to developmental psychopathology. Int J Methods Psy- plex. Eur J Hum Genet 20:510–515. doi:10.1038/ejhg.2011.241 chiatr Res 16:S16–S23. doi:10.1002/mpr 37. Bayley N (1993) Bayley scales of infant development, vol 2. Psy- 17. Constantino JN (2011) The quantitative nature of autis- chological Corporation, San Antonio tic social impairment. Pediatr Res 69:55–62. doi:10.1203/ 38. Bayley N (2006) Bayley scales of infant and toddler development, PDR.0b013e318212ec6e vol 3. Harcourt Assessment, San Antonio 1 3 Eur Child Adolesc Psychiatry (2018) 27:753–765 765 39. Scholte E, van Duijn G, Dijkxhoorn Y et al (2008) Handleiding tuberous sclerosis complex. J Neurol 257:1373–1381. Vineland Screener 0–6 [Manual Vineland Screener 0–6]. PITS, doi:10.1007/s00415-010-5535-2 Leiden 47. Chu-Shore CJ, Major P, Montenegro M, Thiele E (2009) Cyst- 40. IBM Corp (2012) IBM SPSS Statistics for Windows, version 21.0 like tubers are associated with TSC2 and epilepsy in tuberous 41. Galwey NW (2009) A new measure of the effective number of sclerosis complex. Neurology 72:1165–1169. doi:10.1212/01. tests, a practical tool for comparing families of non-independent wnl.0000345365.92821.86 significance tests. Genet Epidemiol 33:559–568. doi:10.1002/ 48. Curatolo P, Napolioni V, Moavero R (2010) Autism spec- gepi.20408 trum disorders in tuberous sclerosis: pathogenetic pathways 42. Preacher KJ, Hayes AF (2008) Asymptotic and resampling strat- and implications for treatment. J Child Neurol 25:873–880. egies for assessing and comparing indirect effects in multiple doi:10.1177/0883073810361789 mediator models. Behav Res Methods 40:879–891. doi:10.3758/ 49. O’Connor K, Kirk I (2008) Brief report: atypical social cognition BRM.40.3.879 and social behaviours in autism spectrum disorder: A different 43. Preacher KJ, Kelley K (2011) Effect size measures for mediation way of processing rather than an impairment. J Autism Dev Dis- models: quantitative strategies for communicating indirect ee ff cts. ord 38:1989–1997. doi:10.1007/s10803-008-0559-5 Psychol Methods 16:93–115. doi:10.1037/a0022658 50. Overwater IE, Verhaar BJH, Lingsma HF et al (2017) Interde- 44. Gallagher A, Madan N, Stemmer-Rachamimov A, Thiele EA pendence of clinical factors predicting cognition in children with (2010) Progressive calcified tuber in a young male with tuber - tuberous sclerosis complex. J Neurol 264:161–167 ous sclerosis complex. Dev Med Child Neurol 52:1062–1065. 51. Hirota T, Veenstra-Vanderweele J, Hollander E, Kishi T (2014) doi:10.1111/j.1469-8749.2010.03792.x Antiepileptic medications in autism spectrum disorder: a system- 45. Koh S, Jayakar P, Dunoyer C et al (2000) Epilepsy surgery in chil- atic review and meta-analysis. J Autism Dev Disord 44:948–957. dren with tuberous sclerosis complex: presurgical evaluation and doi:10.1007/s10803-013-1952-2 outcome. Epilepsia 41:1206–1213. doi:10.1111/j.1528-1157.2000. 52. Tuchman R (2004) AEDs and psychotropic drugs in children with tb00327.x autism and epilepsy. Ment Retard Dev Disabil Res Rev 10:135– 46. Gallagher A, Grant EP, Madan N et  al (2010) MRI find- 138. doi:10.1002/mrdd.20026 ings reveal three different types of tubers in patients with 1 3

Journal

European Child & Adolescent PsychiatrySpringer Journals

Published: Oct 23, 2017

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off