Clinical Management of Children and Adolescents with Neurofibromatosis Type 1 Like Phenotypes and Complex Behavioural Manifestations: A Multidisciplinary and Dimensional Approach

Introduction. Cognitive and behavioural problems associated with Neurofibromatosis type 1 (NF1) are common sources of distress and the reasons behind seeking help. Here we describe patients with NF1 or NF1-like phenotypes referred to a Tier 3 Child and Adolescent Psychiatry Department and highlight the benefits of a multidisciplinary assessment. Methods Prospective data were gathered from NF1 patients aged 7–15 years, referred by the NF1 Referral Centre due to additional difficulties either in management or diagnosis. For the selected cases, we performed a psychiatric assessment, a tailored neuropsychological evaluation based on clinical demands and history, broad speech and motor skills evaluations if there were concerns regarding language, motor abilities and/or learning difficulties and autism specific evaluations, if clinically relevant. No exclusion criteria were applied. Results Complex NF1 cases represented only 5% of the patients (11/224). Assessments revealed the complexity of NF1 phenotype and a variety of problems including learning difficulties, emotional problems and autism spectrum disorders. Specific evaluations of language, motor, attentional and neurovisual domains were essential to guide tailored intervention strategies. Conclusions In terms of clinical implications, the heterogeneity of NF1 phenotypical manifestations needs to be considered when developing assessment and remediation approaches for children with complex NF1.


Background
Neurofibromatosis type 1 (NF1) affects approximately 1/3,000 individuals worldwide [1]. Diagnostic criteria are varied and include café-au-lait spots, neurofibromas, freckling of the axillary or inguinal regions, optic glioma, Lisch nodules, distinctive osseous lesions (such as sphenoid dysplasia or thinning of long bone cortex), as well as having a first-degree relative evidencing NF1 symptoms as detailed above [2].
Although NF1 is an autosomal dominant condition (familiar type), de novo mutations account for up to 50% of the cases (sporadic type). NF1 is completely penetrant, nonetheless it exhibits variable clinical expressivity, even between family members with the same NF1 mutation [3]. Most NF1 mutations result in reduced intracellular levels of the protein neurofibromin, leading to excessive cell proliferation, including development of neurofibromas and tumours and diminished cognitive capacity [4].
In clinical practice, cognitive and behavioural problems associated with NF1 and their impact on academic performance is a common source of distress and the reason behind seeking help in child psychiatry. Indeed, between 30% and 70% of individuals with NF1 have learning disabilities concerning speech, reading, writing, spelling and mathematics. ese problems represent the most significant cause of lifetime morbidity associated with the disease [5,6]. Even though some phenotypic patterns have been suggested in the past, a more in-depth analysis reinforces the need to tailor the diagnosis and the treatment of these patients. e global cognitive functioning of NF1 patients is usually preserved, although somewhat lower when compared to unaffected siblings or peer groups [7]. Intellectual disability, defined clinically as an intelligence quotient (IQ) lower than 70 in cognitive evaluations, is estimated to be around 4-8% in NF1 patients vs. 2-3% in the general population [8,9]. However, the exclusion of patients with central nervous system (CNS) pathology from cohort studies of NF1 prevalence reduces the number of patients with clinically moderate to severe NF1 [8] and may underestimate the prevalence of intellectual disability amongst this population.
For many years the Child and Adolescent Psychiatry Department of Pitié-Salpêtrière has been treating paediatric patients who present neuropsychiatric symptoms of rare diseases [10][11][12] as examples. As a specialised team, the department receives complex patients which present significant management issues. NF1 patients are referred to this department. Here we describe patients with NF1 received in our Departement and highlight the benefits of a multidisciplinary assessment.

Patients.
Prospective data were gathered from NF1 patients (NIH, 1988) aged 7-15 years, referred by the Neurofibromatosis Referral Centre at Trousseau Paediatric Hospital for evaluation at the Department of Child and Adolescent Psychiatry at University Hospital La Pitié-Salpêtrière. e data spans from 2009 until 2016. e procedures regarding the assessments and follow-ups at the Referral Centre are described elsewhere [13]. Referrals were made due to additional difficulties either in management or diagnosis of patients and represented only a small part of the cohort (11/224-approximately 5%). No exclusion criteria were applied as we wanted to describe all complex aspects of this clinical sub sample.

Procedures and Instruments.
Patients were assessed in medical consultations and further comprehensive evaluations. For the selected cases, we extracted from the prospective database: sociodemographic data (age, sex, academic level); all relevant information from the semi-structured medical interview to evaluate the patient's personality and family history of psychiatric and medical disorders, including NF1 clinical features, complications and follow-ups; and all relevant biological (e.g. genetic testing), physiological (e.g. electro-encephalography) and imaging (e.g. magnetic resonance imaging) information. In addition, we performed (1) a psychiatric assessment; (2) a tailored neuropsychological evaluation (e.g. executive function, attention, memory, neurovisual) based on clinical demands and history; (3) a broad speech evaluation, if there were language (oral and/ or written) concerns or learning difficulties; (4) a global and fine motor skills evaluation, if there were concerns regarding motor abilities and/or learning difficulties (such as difficulties to write), and (5) autism specific evaluations, if clinically relevant. e list of testing is given in Tables 1 and 2. Evaluations were performed because they had clinical relevance. ey were adapted to the needs and characteristics of each patient. Medical files containing clinical data were thoroughly explored (AM, AJ, MP) and relevant data were included (AM, AJ, AT, MP). Psychiatric diagnoses were made according to DSM-5 criteria. Finally, the scores obtained in neuropsychological, speech and fine motor skills assessments were converted into standard deviations in order to have unique common statistics and to ease presentation.

Results
Eleven patients were included. All, except one, were boys with a mean age of 10.6 (range 7-15) years. Five patients had a first degree relative with NF1 (father, mother or sibling). Two patients lived in foster care and five with one parent (due to divorce). One had Legius syndrome, a genetic condition associated with sprouty related EVH1 domain containing 1 (SPREAD1) gene mutation, which phenotype overlaps that of NF1 [14]. For clarity, we present the patients in two separate tables. Table 1 summarises the clinical profile of four NF1 patients with an intellectual deficiency and Table 2 presents the profile of patients with subnormal IQ.

Reasons for Referral.
Patients came with multiple and diverse complaints. Reasons for referral included learning difficulties (due to the consequences of instrumental difficulties, suspicion of attention deficit hyperactivity disorder (ADHD) or intellectual disability; 푁 = 7), or emotional problems or suspicion of emergent psychopathology (푁 = 6). A few (푁 = 4) were referred because of autism spectrum disorder (ASD) (doubts regarding diagnosis or difficulties in clinical management).
treatments such as different types of psychotherapy were proposed and appeared to be helpful.

NF1 and Autistic Spectrum Symptomatology.
Four patients presented clear autistic features a er clinically comprehensive/ developmental assessment and additional assessment tools (patients 1, 2, 4; Table 1 and patient 9, Table 2). ree patients with autistic features had concurrent intellectual disability (Table 1). A fi h patient with an ongoing history of social difficulties had a pragmatic communication disorder (patient 5; Table 2). Complementary psychological tests were helpful for the diagnosis (CARS [19], ADI [15], ADOS [16]). Treatment strategies addressed co-morbidities as well as core features of autism (e.g., social strategies in group) in outpatient settings or in the context of specific schooling models.

Discussion
is case series reflects the variety of problems and the complexity of severe NF1 paediatric patients. e gender bias reflected in our sample may result from the overrepresentation of boys in child and adolescent psychiatry consultations and therefore will not necessarily reflect the epidemiology of NF1 [30]. Also, the morbimortality of the disease, is mild to moderate in most cases. Nonetheless, clinical complications of NF1 can present a serious burden. is was the case for most of the patients reported here, who were referred to a special psychiatric clinic for rare diseases. e large heterogeneous phenotypic expression of NF1 is likely to be a consequence of the stochastics chain of events associated with NF1. At a molecular level, the reduced intracellular levels of neurofibromin found in NF1 patients induce impairments in learning and memory through imprecise, i.e. abnormally high or low RAS modulation and consequential gamma-aminobutyric acid (GABA)-mediated excessive inhibition in the hippocampus [4,31]. Also, patients with NF1 may show subcortical unidentified bright objects (UBOs) as we found in four of our patients. In all but one, UBOs were present in patients with intellectual disability and a multitude of somatic complications of the disease. UBOs are present in approximately 60-70% [32,33] of the children with NF1 but tend to disappear with age. Histopathologic studies have shown that UBOs correspond to areas of myelin vacuolization with increased water content [34] and therefore, could reflect disordered myelination [35]. e presence of UBOs has been associated with lower intellectual ability [36,37] and also visuospatial impairments when UBOs were located in the cerebellum [38]. e latter seems to be the case for patients 1 and 2 ( Table 1) and patient 8 ( Table 2) in our case series.

NF1 Cognitive Impairment and Learning Disabilities.
In this case series, the patients presented either a degree of intellectual disability (Table 1) or an average but heterogeneous cognitive profile with significant functional impairments ( Table 2). Many studies have now made clear how intelligence is only mildly affected in the vast majority of NF1 patients. However, specific impairments in cognition are very common (up to 80% of children in NF1 clinics [8]) and have a negative intertriginous freckling, few neurofibromas and/or Lisch nodules, were present. ese patients all had subnormal IQ.

NF1 and Learning Disabilities.
All patients had learning disabilities that appeared as a primary or secondary cause of referral. For the patients who had subnormal IQ scores (Table 2), it was common to find discrepancies in IQ subscales. Within each subscale, scores could vary widely (data not shown). Even if the patients received an average/mean IQ, their global intellectual efficiency was weaker when compared to the general population.
Specific learning disabilities were also common. ey included specific language developmental delays (patients 5, 6, 9, 11; Table 2), graphomotor delays (patients 5, 6, 7, 8; Table  2) and developmental coordination disorders (dyspraxia) (patients 8, 9; Table 2). Some patients also had comorbid ADHD (patients 5, 6, 8; Table 2). e extensive developmental evaluation was crucial to disentangle diagnosis and also to guide further therapeutic propositions. Such specific propositions included school adaptations (such as having more time to finish evaluations, the use of computer to write), speech and reading therapies once or twice per week, and motor skills remediation once per week, that lead to improvement (nonpharmacological treatment; Table 2).
Patients with intellectual disabilities (Table 1) arguably presented learning difficulties. Standard tests for cognitive evaluation (such as the Wechsler intelligence scale for children, WISC [29]) were difficult to perform in such patients. In these cases, clinical evaluation and developmental scales were helpful (Brunet Lezine [18]), Terman Merrill [25]), as well as some psychosocial scales (Psychoeducational Profile (PEP) [23], Vineland [28]) that estimate performance in more ecologic day to day activities. In this context, specific evaluations of language, motor, and attentional domains are equally important to guide tailored intervention strategies. Intervention strategies were generally more global in the context of schooling for disabled children.

NF1 and Attentional
Issues. Four of our patients had an ADHD diagnosis, obtained a er clinical assessment and a specific neuropsychological evaluation (Test of Everyday Attention for Children, TEA-Ch [24]). A fi h patient (patient 4; Table 1) diagnosed with ADHD was evaluated through clinical observation and functional scales (Conners [20]), because the neuropsychological evaluation was difficult to perform given the intellectual deficit. In all cases, an ADHD diagnosis was made along with other comorbidities and was never isolated. Four patients were treated with methylphenidate and one was given specific cognitive training addressing attention. ree patients receiving methylphenidate improved and one stopped the treatment due to side effects.
e most frequent diagnosis was anxiety disorder and/or NF1 related stress (patients 1, 4, 7, 10, 11). Two visited for the evaluation and treatment of mood disturbances (patients 8, 11; Table 2). Finally, another patient visited for psychiatric evaluation in the context of sexual assault (patient 3; Table 1). Medication along with nonpharmacological Case Reports in Psychiatry 8 than the general population (up to 40% in some samples [56]). Whether or not this is enough to state that ASD is increased in NF1 patients is another question. First, ASD screening tools are not meant to provide a diagnosis of ASD, but rather identify children who should be further evaluated. Second, NF1 patients with high SRS scores show mild to moderate ASD symptomatology (e.g. latter diagnosis during school years; no stereotyped behaviours) [57]. We believe that research should explore further the meaning of ASD symptomatology in NF1. Given the rarity of stereotyped behaviours in NF1, perhaps the DSM-5 diagnosis of social communication disorder will be more appropriate in some cases? Also, is it legitimate to use SRS score as a continuous variable? Indeed, when re-analysing the largest cohort of NF1 so far that reported SRS scores [59] we found that the best model to predict data distribution using admixture analysis was a bimodal distribution (Figure 1), with a large group of patients with a mean SRS score equal to 55 (red curve), and a smaller group of patients with a higher mean SRS score equal to 70 (green curve). is means that considering SRS as a continuous variable showing a normal distribution may be inappropriate.
is also means that NF1 phenotype heterogeneity might be better understood separating subgroups of patients as it has been shown in other genetically complex neurodevelopmental disorders.
However, we do believe that rare typical cases of ASD can also be encountered in NF1. ree patients in this case series presented an unequivocal profile of ASD comorbid with intellectual disability. Several authors postulated that autism associated with neurogenetic syndromes can be classified as "complex autism" or "syndromic autism" [60][61][62]. Complex autism is characterised by lower IQ, higher rates of comorbidities both psychiatric and somatic, and higher rates of epilepsy [63,64]. For these severe cases, this classification may be useful to better explore their clinical profile and tailor treatment to the child's needs [65].

NF1 and Emotional
Problems. NF1 has also significant social and psychological consequences on individuals and their families [66,67]. Most patients of this case series were referred with emotional problems. While some (푛 = 7) had behavioural difficulties (sometimes secondary to ADHD or to learning disabilities), others expressed internalising symptoms such as anxiety or depression. ese conditions can o en be found in NF1 [67]. e emotional impact of cosmetic deformities, the fear of malignancy, and the medical complications, such as the management of hypertension or the need for surgical interventions should be carefully evaluated and not be neglected. Learning difficulties can also generate feelings of low self-esteem and inefficiency that can last in time. In addition, the impact of attention deficits and behaviour problems in children with NF1 o en leads to lower social acceptance and lack of self-confidence. As mentioned previously, when emotional problems are secondary to ADHD, effective treatment of ADHD can improve them both [46]. Besides, parents of NF1 children experience greater stress than other parents [68], sometimes associated with feelings of guilt regarding the genetic transmission of NF1. e parents should also receive proper care.
impact in the quality of life. Some specific deficits have been reported to NF1: visual-spatial deficits [39], speech and language deficits, motor skill deficits [40], social skill deficits [5,41] and attentional deficits [8,42]. All may lead to learning disabilities and further emotional suffering. e presence of visual-spatial and attentional deficits has been robustly replicated [7,43]. However, research has revealed contradictory results regarding motor skills and language performance in patients with NF1 [7]. Moderate to severe intellectual disability and severe clinical cases seemed to be the focus of interest of early works in neurofibromatosis [44] but severe cases of NF1 have received little attention recently.

NF1 and ADHD.
In all our NF1-ADHD patients, the ADHD diagnosis was made along with other comorbidities and was never isolated. Patients with NF1 present more symptoms and are more o en diagnosed with ADHD than the general population. Prevalence estimates range from 30% to 50% [45,46], which are higher than those expected in the general population, i.e. about 5% in children and 2.5% in adults [47,48].
Clinical and cognitive profiles of both ADHD patients with and without NF1 are heterogeneous. Individuals are affected in different domains of attention, impulsivity, hyperactivity and executive functioning, and to different degrees [46,49]. Visual attention seems consistently impaired in NF1 patients [50,51] and would lead to instability in focusing attention and lower resistance to interference in controlled tasks but also to inattentive and impulsive behaviour in natural environments.
Importantly, ADHD-NF1 patients seem responsive to methylphenidate [42,46]. is improvement has been also reported for children with IQs lower than 80 [46]. In our case series, four of the patients diagnosed with ADHD were treated with methylphenidate, three showing a great improvement and one dropping out of treatment due to side effects. e high frequency of ADHD in children with NF1, as well as the demonstrated significant comorbidity of ADHD with literacy learning disabilities [8] and social skills problems [52] indicates the need for thorough screening of ADHD symptomatology in all children with NF1 [45]. Besides, the impact of attention deficits and behaviour problems in children with NF1 o en leads to lower social acceptance and lack of self-confidence/esteem. When emotional problems are secondary to ADHD, effective treatment of ADHD can improve them both [46].

NF1 and Autistic Spectrum Symptomatology.
A er some early work evoking associations between autism and NF1 [53] there is a re-enacted concern regarding ASD and NF1. Since Huijbregts and de Sonneville [54] showed that NF1 impairments in cognitive control (i.e. a combination of processing speed, working memory, inhibitory control, and emotional processing functions) is associated with the presence of autistic traits, several studies have explored the prevalence of ASD symptoms in NF1 populations.
Studies focusing on social impairments as a continuous variable within the social responsiveness scale (SRS) [55][56][57][58] showed that NF1 patients present more social impairments Phenotype-genotype correlations tend to subscribe the idea that behavioural manifestations present themselves with continuous variation. is type of pattern was proposed for NF1 autistic symptoms [59]. However, we believe that more complex phenotype/genotype association occurs in NF1 for psychiatric comorbidity including ASD symptomatology. Similarly to other genetic conditions such as juvenile myotonic dystrophy [73], or tuberous sclerosis [74], bimodal phenotypic patterns of intelligence/cognitive performance seem to adapt better to the distribution of IQ scores in these populations. In terms of clinical implications, the heterogeneity of NF1 phenotypical manifestations needs to be considered when developing assessment and remediation approaches for children with NF1.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Treatment Modalities and Outcome.
Strategies that decrease either RAS activity or GABA-mediated inhibition have been suggested to specifically treat learning and emotional deficits associated with NF1. Pharmacologically targeting this pathway with an HMG-CoA reductase inhibitor (an anti-RAS agent), such as lovastatin [69] evidenced some encouraging results in animal models. However, to date, clinical trials with NF1 patients have shown modest results [70][71][72]. erefore, current treatment options remain symptomatic. e heterogeneity of NF1 phenotypical manifestations needs to be considered when developing assessment and remediation approaches for children with NF1. e outcome is therefore variable, different for each patient.

Conclusions
In severe cases of NF1 with important psychiatric morbidity, it can sometimes be difficult to disentangle specific domains of impairment, as cognition as a whole is complex. In day-today activities, more than one cognitive domain is needed to perform adequately, making the task of disentanglement even more challenging. is is why a multidimensional and longitudinal clinical evaluation is warranted when assessing patients. ). e admixture analysis shows that the best fitting model for SRS scores in NF1 founds 2 subgroups with low and high scores (mean SRS = 55 (red curve) and 70 (green curve), respectively). Data from Morris et al. [59].