Neuropsychiatric Functioning in CDLS: A Detailed Phenotype and Genotype Correlation

Behavioural phenotype and autism-related traits of 38 patients affected by Cornelia de Lange syndrome (CdLS) were assessed using a specific neuropsychiatric protocol. Subsequently,we search for possible genotype–phenotype correlations comparing individuals with NIPBL variants and patients with negative molecular results. Firstly results showed a higher percentage of subjects with normal intellectual quotient (IQ) and borderline IQ; adaptive skills were lower than expected for age in all participants. 39.5% of the sample presented with autism spectrum disorder (ASD), NIPBL mutated individuals demonstrated a worse trend in comparison with the clinical diagnosis group. non-truncating individuals displayed no ASD and better communication abilities than truncating individuals. Findings increase our awareness of the strengths and weaknesses points in CdLS individuals.


Background
Cornelia de Lange syndrome (CdLS, OMIM, #122470, #300590, #610759, #614701, #300882) is a rare genetic disorder characterized by peculiar facial dysmorphisms, major malformations, growth retardation, and developmental delay/intellectual disability with an estimated prevalence between 1 in 10,000 and 1 in 30,000 live births (Kline et al., 2007). Classic (or typical) CdLS phenotype is easily recognized, but clinical expression is widely variable. A recent International Consensus Statement defined the whole CdLS phenotype as a spectrum and conceived a diagnostic score based on a combination of cardinal and suggestive features to uniform the definition of classic and non-classic CdLS (Kline et al., 2018). CdLS spectrum (CdLSp) belongs to Coesinopathies because a mutation in one cohesin functionrelevant gene (NIPBL, SMC1A, SMC3, RAD21, HDAC8, ANKRD11, BRD4) is identified in most patients, although the causal variant remains unknown in 15-20% of patients (Huisman et al., 2013;Kline et al., 2018).
In the past, different studies searched for genotype-phenotype correlations in CdLS; the general consensus suggests that NIPBL mutations are likely to be associated with a more "classic" clinical presentation and a more severe impaired cognitive function (Gillis et al., 2004;Mannini et al., 2013). According to the diagnostic score proposed by the International Consensus Statement, most patients with NIPBL mutation have a high score and meet the criteria for the definition of classic CdLS (Kline et al., 2018). A kind of genetic gradient of severity for the other genes associated with CdLSp was presumed with HDAC8 gene at the top with a more classic and severe clinical presentation, and SMC3 gene at the bottom with a mild presentation. None of these correlations, however, are based on statistical analysis of a large cohort of patients.
Literature reports intellectual disability (ID) ranging from mild to profound; most patients have a moderatesevere ID, but some individuals have a normal IQ. Adaptive behaviour is compromised, and communication abilities vary widely but, typically, major difficulties are present in expressive language skills which are more compromised than receptive ones (Ajmone et al., 2014;Goodban, 1993;Sarimski, 1997). Subsequent studies demonstrated that CdLS individuals showed difficulties in verbal comprehension and in explanation of concepts (Mulder et al., 2019).
The prevalence of ASD symptomatology in CdLS ranges between 27 and 82% (Basile, 2007;Mulder et al., 2016;Richards et al., 2015). Moreover, restricted and repetitive Behaviours occur in patients with more marked ID and with ASD and usually they are associated with increased risk of self-injurious Behaviours (Arron et al., 2011).
It is important to point out that many researches have focused their attention on particular features of the Behavioural phenotype (Bruserudu et al., 2016;Cochran et al., 2015;Fisher et al., 2016;Moss et al., 2017) but few of them have used complete evaluations' protocols, characterized by the mix of direct and indirect tools, specifically chosen in order to obtain a broad and more appropriate assessment in these children. A direct assessment made by clinicians with experience in genetic syndromes and neurodevelopmental disorders combined with an indirect evaluation which takes into consideration information given by the parents reduces misdiagnosis and allows for a better intervention plan.
Despite the growing interest on genotype-phenotype correlations and on behaviour phenotype of genetic syndromes, specific studies in CdLS cohorts evaluating the correlations between genotype and neurodevelopmental characteristics, Behaviour and communicative aspects are limited (Ajmone et al., 2014;Bhuiyan et al., 2006;Huisman et al., 2017;Moss et al., 2017;Mulder et al., 2019); most of these studies are descriptive and there is a lack of specific assessment protocols.
These studies suggest that CdLS individuals with NIPBL mutations show lower levels of cognitive functioning and adaptive Behaviour skills, as well as more self -injurious Behaviour (SIB) compared with patients with the SMC1A variant (Huisman et al., 2017;Mulder et al., 2019). Furthermore, NIPBL truncating mutations usually cause more pronounced Behaviour problems, including autism, but similar problems also occur in patients caring missense mutations (Bhuiyan et al., 2006). Finally, no studies have shown a clear association between the type of NIPBL variant and level of ID Wulffaert et al., 2009;Yan et al., 2006), only one study indicated that missense and in frame deletions of NIPBL were associated with mild cognitive phenotype (Ajmone et al., 2014).
In this study, we aim to outline the Behavioural phenotype of CdLS based on a broad functional description of a cohort of 38 patients evaluated using a tailored neuropsychiatric assessment. Moreover, we try to bring out some genotype-phenotype correlations, supported by statistical analysis. The results emerged from this study should increase our knowledge on specific neuroBehavioural strengths and weakness points of CdLS individuals to guide appropriate targeted clinical interventions and management for these individuals and their families in order to improve their quality of life.

Participants
38 CdLS italian patients (19 males and 19 females), with an age at evaluation ranging between 7 months and 15 years (mean: 6.05 ± 4.13 with a median of 5 years), referred to our clinic for assessment and follow-up, were enrolled in the study. All patients had a clinical diagnosis of CdLS according to diagnostic criteria proposed by Kline et al. (2018).
Molecular tests included NIPBL sequencing on blood and buccal smear and NIPBL MLPA analysis, SMC1A, HDAC8, RAD21 and SMC3 sequencing analyses were performed. Before the introduction of next generation sequencingbased cohesin gene panels in 2015, the molecular workflow included the sequence of single gene Sanger sequencing started with NIPBL, based on their frequencies, completed with NIPBL MLPA analysis.
Among 38 patients in the study cohort, 17 individuals were part of a previous study evaluating communication, cognitive level, development and Behaviour between 2009 and 2011 (Ajmone et al., 2014); the other 21 were evaluated between 2011 and 2020.
All patients were assessed at Child and Adolescent Neuropsychiatric Service, Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico, Milan. The assessment protocol was administered to all subjects during outpatient visits.

Procedures
Neuropsychiatric phenotype of all the patients was assessed using a specific protocol, concerning Intellectual Quotient (IQ), General Quotient of Development (GQ), communicative skills, behavioural aspects and adaptive Behaviour based on direct and indirect evaluation.
Firstly, we analysed and described each developmental area considering the whole cohort of patients in order to obtain a detailed and in-depth description of CdLS Behavioural phenotype. Moreover, we focused on identifying phenotype-genotype correlations comparing individuals with NIPBL variants (CdLS NIPBL mutated group) and patients with negative molecular results (CdLS clinical diagnosis group). Finally, a more specific comparison between two different types of NIPBL mutations (truncating versus missense) was performed. This comparison was possible because the sample was homogeneous in age and number.

Cognitive and Developmental Assessment
Two different scales were used to assess Intellectual Quotient (IQ) and General Quotient of Development (GQ): the Leiter International Performance Scales Revised-Leiter-R (Roid & Miller, 1997), and the Griffiths' Scale (Griffiths, 1986). We used the Griffiths' Scale (Griffiths, 1986) to evaluate the general developmental quotient (GQ) in patients from 0 to 8 years old. The GQ is composed of six sub-quotients, one for each area investigated (locomotor, personal social, language, eye and hand coordination performance, and practical reasoning). The GQ identifies how children perform across developmental areas.
The Leiter International Performance Scales Revised-Leiter R (age range 2-21 years) (Roid & Miller, 1997), is a non-verbal cognitive test. It is useful because it has a short version that does not require high attention skills. Moreover, it does not require verbal communication abilities, and it can be administered when communication difficulties are present or to non-speaking children. We decided to consider the IQ obtained using Leiter scale because it may better highlight the real cognitive abilities of CdLS patients (Ajmone et al., 2014).
Although the two measures do not perfectly correspond, we considered both as indices of children's development. Both tests yielded a standardised quotient with M = 100 and SD = 15.
In literature, the use of a combined IQ/GQ index is quite common if participants are at significantly different stages with respect to the level of functioning (e.g., Noyola et al., 2001;Winterkorn et al., 2007;Van Schooneveld & Braun, 2013). In addition, the Griffiths GQ appeared to be a good predictor of later IQ (e.g., Bowen et al., 1996).

Adaptive Behaviour Assessment
The Vineland Adaptive Behaviour Scale (VABS) in its Italian adaptation and validation (Balboni & Pedrabissi, 2003) was used to assess adaptive Behaviour. The VABS is a semistructured interview with caregivers and allows us to assess global adaptive behaviour skills (Adaptive Behaviour Composite) and ability on four specific domains (Communication, Daily Living Skills, Socialisation, Motor Skills).

Communication and Language Evaluation
For the assessment of expressive and receptive skills we used the Communication domains of VABS scale (expressive and receptive language). It provides a Language ageequivalent score.

Behavioural Assessment and Autism Spectrum Disorders
Behavioural characteristics of the participants were assessed using the Child Behaviour Checklist-CBCL (Achenbach and Rescorla, 2001), while ASD (autism spectrum disorders) symptomatology was evaluated with the Childhood Autism Rating Scale (CARS) (Schopler et al., 1988). Child Behaviour Checklist (CBCL) (Achenbach and Rescorla, 2001) was used to assess children's Behavioural characteristics. The CBCL is a 100-item questionnaire completed by parents reflecting their point of view of the child's Behaviour at the time of administration and for the preceding 3 months. It provides a child Behaviour profile into eight subscales: withdrawn behaviour, somatic complaints, anxiety/depressed behaviour, opposite behaviour, aggressive Behaviour, social problems, thought problems, and attention problems. Single sub-scales can also be scored in terms of two broad grouping of symptoms: internalizing (consisting of anxious/depressed, withdrawn, emotionally reactive, somatic complaints) and externalizing (consisting of attention problems, aggressive Behaviour, rule-breaking).
The Childhood Autism Rating Scale (CARS) is a Behaviour rating scale used as a screening measure of Autism Spectrum Disorder in clinical and research studies. This scale investigates the presence of Behavioural, cognitive, and communicative characteristics associated with autism and provides a final score, produced by the clinicians at the end of the evaluation. Fifteen areas are considered, and for each area, the child is rated on a scale from 1 (normal behaviour) to 4 (severely abnormal behaviour). Thus, the total score ranges from 15 to 60. In compliance with the CARS manual, we adopted a cut-off score of 30. For the purposes of the present study, children were divided into two categories, namely, children without autistic traits, i.e., children with scores ranging between 0 and 30, and children with mild, moderate or severe autistic traits, i.e., children with scores ranging between 30 and 60.

Statistical Analysis
Independent sample t-tests were used to evaluate differences in IQ/GQ indexes, CARS scores and CBCL internalizing, externalizing and total scores between CdLS NIPBL mutated group and CdLS clinical diagnosis group. The same type of analysis was performed between truncating NIPBL mutations and non-truncating NIPBL mutation groups.
Age equivalent VABS scores were compared using ANCOVAs controlling for age. This practice is debated (Glass et al., 1972;Miller & Chapman, 2011) even though it is common in different research fields involving human participants (see, for example, recent papers from Flesher et al., 2020;Gaubatz et al., 2020).
A correlation between IQ/GQ indexes and CARS scores was performed to investigate a general relationship between developmental/cognitive levels and the characteristics associated with autism. A similar analysis was conducted using dichotomized variables of intellectual functioning (average or borderline functioning vs. mild, moderate, severe or profound disability) and autistic traits (with or without). Due to small sample size, P-values were not adjusted for multiple comparisons, but we included power calculations (for unbalanced ANCOVAs we used the average group size).

Molecular Analysis
Participants were divided into two groups according to clinical and molecular diagnosis: 31 participants (81.6%) carried one heterozygous pathogenetic or likely pathogenetic variant in NIPBL gene (CdLS NIPBL mutated group) while 7 (18.4%), with a clinical diagnosis of CdLS, were negative for all CdLS associated genes analyzed (CdLS clinical diagnosis group).
Among individuals with NIPBL mutations, we divided patients on the basis of the truncating effect of NIPBL mutation to find some correlations with the behavioural and cognitive data. Most patients had a mutation with truncating effect (n: 17; 44.7%), while 10 individuals (26.3%) had a non-truncating, missense mutation. In 4 patients (3 or 7.9% with splicing mutation and 1 or 2.6% with in-frame deletion) it was not possible to define the mutation effect without functional studies. No significant age difference was present between the truncating NIPBL group and the non-truncating NIPBL group (mean age 4.71 years; SD 3.82; mean age 5.30 years; SD 3.47).
Subsequently, a between-groups comparison was performed on the participants' cognitive characteristics and their genotype.
The CdLS NIPBL mutated group showed a worse trend in cognitive level despite the fact that a significant difference was not found (IQ/GQ Mean index in NIPBL positive: 56.14, SD 28.97; IQ/GQ Mean index in NIPBL negative: 42.81, SD 29.71). A between-group comparison between the Truncating and missense NIPBL mutations group did not show any significant difference either.

Adaptive Behaviour Evaluation
Participants were evaluated for Adaptive Behaviour (VABS Adaptive Behaviour composite) and an adjusted developmental age was obtained (age equivalent score). Adaptive skills were lower than expected for age in all participants (n = 36, mean adaptive age: 28.42 months range from 2 months to 9.67 years; SD 23.57 months).
We obtained this profile in the main areas composing  (Table 1)

Communication and Language Evaluation
Participants were evaluated with the VABS Communication scale (n = 32; M: 31.97; SD 30.85) and the Communication Expressive and Receptive subscales were considered.
Moreover, differences between NIPBL mutated group and CdLS clinical diagnosis group were found in the Communication Expressive subscale (with a generalized eta squared of 0.336) while no significant differences were found in receptive abilities; this finding confirms the expressive ability as a typical weakness in NIPBL mutated individuals.

Autism Spectrum Disorders
Children were divided into two categories by CARS cut-off score of 30: 18 (47.5%) were classified having ASD, while 15 (39.5%) were found to be without autistic traits (5, 13.2%, were not evaluated because they were too young to be evaluated with this tool). Independent sample T-tests were conducted to test differences in Behavioural assessment scores between CdLS NIPBL mutated group and CdLS clinical diagnosis group. A difference (± 2.512; p < 0.05) was found in CARS scores where the NIPBL mutated group showed higher scores for ASD symptomatology (36.615 ± 12.878) than clinical diagnosis group (27 ± 7.61). A significant difference was found also between the Truncating group and Missense at CARS assessment (41.429 ± 12.113 vs. 29.312 ± 11.622) (Fig. 1). The correlation between IQ/GQ indexes and CARS scores was strong and highly significant (r = − 0.83, p < 0.001) and the chi-squared test confirmed an association between the dichotomized variables of intellectual functioning and autistic traits (χ 2 1 = 7.16, p < 0.01).

Behavioural Assessment
In this study we considered the three broad scales of CBCL: Internalizing (M: 61.03 ± 10.58), Externalizing (M: 57.19 ± 9.89) and Total Problems (M: 61.47 ± 10.76). CBCL was used according to the chronological age of participants (n = 32). Differences in CBCL scores had medium (for externalizing symptoms and total score) and large (for externalizing symptoms) effect sizes. T-tests comparison did not show significant group difference both between clinical diagnosis group and NIPBL mutated group and between Truncating and Missense group. This may be due to missing one CBCL evaluation of a participant with clinical diagnosis (Tables 4, 5).

Discussion
The aim of the study was twofold. First of all, we delineated in detail the neurodevelopmental and Behavioural phenotype of individuals with CdLS using a specific direct and indirect neuropsychiatric evaluation protocol; subsequently, we searched for possible genotype-phenotype correlations within the collected data. The first part of the study, describing the general CdLS behavioural phenotype in the sample, confirms the preliminary results of our previous data (Ajmone et al., 2014) and those found in literature (Basile et al., 2007;Kline et al., 2018;O'Brien, 2006;Oliver et al., 2008;Moss et al., 2008;Sarimski, 2002) showing typical points of strengths and weaknesses of these children. Results show a higher percentage of subjects with normal IQ (n = 5, 13.2%) and Borderline Intellectual Functioning (n = 6, 15.8%), as compared with previous literature [Basile et al., 2007;Kline et al., 2018]; this confirms the notion that using more appropriate instruments for cognitive evaluation (short and nonverbal tools like the Leiter Scale) may better highlight the real cognitive abilities of these individuals.
In accordance with the literature, adaptive skills were lower than expected for age in all participants (mean adaptive age: 28.42 months range from 2 months to 9.67 years; SD 23.57 months) (Basile et al., 2007;Kline et al., 2018): the weakest areas were Socialization, Motor skills, and Communication. Moreover, previous researches showed that Adaptive Behaviour skills in CdLS can decrease with age (Bhuiyan et al., 2006;Kline et al., 2018;Srivastava et al., 2014), therefore early support for children and family in this area is needed.
With regards to language and communication, expressive language is more compromised than receptive language, in line with previous reports [Ajmone et al., 2014;Goodban, 1993;Sarimski, 2002;Mulder 2019;Kline et al., 2018]; nevertheless, receptive abilities are also impaired, showing difficulties in verbal comprehension as recently described in a research too (Mulder et al., 2019). Direct evaluation showed the presence of ASD in 39.5% of the sample, which is consistent with previous studies that showed a range between 27 and 82% (Basile et al., 2007;Mulder et al., 2016;Richards et al., 2015). Furthermore, consistent with previous studies, our data confirm a strong correlation between ID and ASD (Basile et al., 2007).
The second aim of the study was to define a specific genotype-phenotype correlation in CdLS patients. According to the consensus conference (Kline et al., 2018) we focused on the NIBPL variant and its different mutations (truncating and missense). The literature describes a more severe phenotype in patients with a truncating mutation than in those with a missense mutation (Bhuiyan et al., 2006), but to the best of our knowledge, the literature lacks specific and standardized neuropsychiatric assessment studies to confirm it. By stratifying CdLS phenotypes by genetic cause, significant differences in neurodevelopmental levels, and Behavioural phenotypes were observed. NIPBL mutated individuals demonstrated a worse trend in cognitive functioning, in communication expressive skills, in motor skills, and in ASD symptoms compared to the Clinical Diagnosis Group. Due to the heterogeneity of our sample for age and number, our considerations are not to be considered absolute.
In agreement with previous research (Ajmone et al., 2014;Bhuiyan et al., 2006), the severity of the behavioural phenotype in patients with truncating mutation was significantly higher if compared with patients with missense mutation. The Missense group showed higher adjusted age equivalent means in Communication scale (36.333 ± 6.111 vs. 18.026 ± 4.778) and in Expressive subscale (32.933 ± 4.767 vs. 19.108 ± 3.216) and fewer scores in ASD symptomatology (29.312 ± 11.622 vs. 41.429 ± 12.113) than Truncating mutation group regardless of cognitive level (no ID significant difference was found between the different groups). The homogeneity of number, age and cognitive level of this sample allowed us to clearly define two different neuropsychiatric phenotypes: truncating and Missense phenotype. Patients with missense mutation showed better abilities in Communication (especially in expressive language) and in relational aspects with no ASD while patients with truncating mutations showed worse abilities in Adaptive behaviour, Daily Living Skills, Socialization, Motor Skills, and Communicative abilities. The absence of a significant difference in IQ level between the two groups underlines that the ASD seems to be a peculiar characteristic of Truncating phenotype. This is one of the first genotype-phenotype studies in a relatively large cohort of CdLS individuals evaluated with a tailored neuropsychiatric assessment. Our purpose was to investigate different molecular mutations and their neurodevelopmental correlates, while taking prognostic and rehabilitative aspects into accounts.
This study showed that NIPBL mutated individuals demonstrated a worse trend in cognitive level, expressive communication, motor area, and ASD symptoms if compared with the Clinical diagnostic group. Literature and everyday clinical practice show how in CdLS individuals impaired communication ability is prominent and impacts both on socialization and Behaviour (Goodban, 1993;Kline et al., 2018;Moss et al., 2013;Sarimski et al., 2002), but this is the first work that defines expressive language as a typical point of weakness in individuals with NIPBL pathogenetic variant and specifically in patients with truncating mutation. Furthermore, receptive language seems to be a typical fragility of all CdLS individuals regardless of the genotype: these data are consistent with other previous findings (Ajmone et al., 2014;Mulder et al., 2019) and confirm the need for an early rehabilitation and augmentative communication approach in all communicative areas. Moreover, it is well known that communication disorders play a very significant role in the occurrence of Behavioural disorders such as oppositional and withdrawal traits, in line with our results showing a significant ASD symptomatology in NIPBL mutated group and in truncating mutations group. These findings guide us to consider the communicative aspect as one of the main intervention priorities in these patients. As in the case of any child with developmental delay, early intervention on communication and language is highly recommended to produce favorable outcomes on development and reduce the risk of Behavioural disorders. Early Augmentative and Alternative Communication (AAC) interventions and speech therapy are therefore of paramount importance from the outset to prevent the onset and aggravation of challenging Behaviours (Kline et al., 2018). Finally, particular attention is paid to motor skills, which are more compromised in patients with NIPBL Truncating mutation involving motor delay, clumsiness and coordination impairment. These data are consistent with literature that describes major limb defects in individuals with truncating NIPBL variants Gillis et al., 2004;Mehta, 2016). These malformations have a great impact on motor and adaptive skills. Early neuromotor rehabilitation is a priority for all Cdls children, and above all for patients with NIPBL truncating variants.
In conclusion, in this study we delineated in detail the neurodevelopmental and behavioural phenotype of individuals with CdLS with a standardized and tailored assessment. Furthermore, although truncating mutations were generally found in literature to cause a more severe phenotype, our findings allow us to to define two different neuropsychiatric phenotypes: truncating and Missense phenotype with specific strengths and weaknesses. The results allow the clinicians to develop a tailored and effective management with a preventive approach and rehabilitative priorities. Although this study focused on patients referred to our clinic from the entire national territory, future studies should consider extending the sample in a multicentre way in order to avoid a possible ascertainment bias towards individuals with more significant impairments.