Vineland adaptive behavior scales to identify neurodevelopmental problems in children with Congenital Hyperinsulinism (CHI)

Background Congenital Hyperinsulinism (CHI) is a disease of severe hypoglycaemia caused by excess insulin secretion and associated with adverse neurodevelopment in a third of children. The Vineland Adaptive Behavior Scales Second Edition (VABS-II) is a parent report measure of adaptive functioning that could be used as a developmental screening tool in patients with CHI. We have investigated the performance of VABS-II as a screening tool to identify developmental delay in a relatively large cohort of children with CHI. VABS-II questionnaires testing communication, daily living skills, social skills, motor skills and behaviour domains were completed by parents of 64 children with CHI, presenting both in the early neonatal period (Early-CHI, n = 48) and later in infancy (Late-CHI, n = 16). Individual and adaptive composite (Total) domain scores were converted to standard deviation scores (SDS). VABS-II scores were tested for correlation with objective developmental assessment reported separately by developmental paediatricians, clinical and educational psychologists. VABS-II scores were also investigated for correlation with the timing of hypoglycaemia, gender and phenotype of CHI. Results Median (range) total VABS-II SDS was low in CHI [-0.48 (-3.60, 4.00)] with scores < -2.0 SDS in 9 (12%) children. VABS-II Total scores correctly identified developmental delay diagnosed by objective assessment in the majority [odds ratio (OR) (95% confidence intervals, CI) 0.52 (0.38, 0.73), p < 0.001] with 95% specificity [area under curve (CI) 0.80 (0.68, 0.90), p < 0.001] for cut-off < -2.0 SDS, although with low sensitivity (26%). VABS-II Total scores were inversely correlated (adjusted R2 = 0.19, p = 0.001) with age at presentation (p = 0.024) and male gender (p = 0.036), males having lower scores than females in those with Late-CHI [-1.40 (-3.60, 0.87) v 0.20 (-1.07, 1.27), p = 0.014]. The presence of a genetic mutation representing severe CHI also predicted lower scores (R2 = 0.19, p = 0.039). Conclusions The parent report VABS-II is a reliable and specific tool to identify developmental delay in CHI patients. Male gender, later age at presentation and severity of disease are independent risk factors for lower VABS-II scores.


(Continued from previous page)
Conclusions: The parent report VABS-II is a reliable and specific tool to identify developmental delay in CHI patients. Male gender, later age at presentation and severity of disease are independent risk factors for lower VABS-II scores.
Keywords: Glucose, Insulin, Vineland, Development, Cognitive assessment, Neurodevelopment, Developmental delay Background Congenital Hyperinsulinism (CHI) is a significant disorder of hypoglycaemia caused by excessive and unregulated insulin secretion [1,2]. CHI usually presents early in the neonatal period (Early-CHI), but later presentation (Late-CHI) is also well recognised [3][4][5]. A significant proportion of children with Congenital Hyperinsulinism (CHI) have adverse neurodevelopmental abnormalities in spite of improvements in medical care [3,4,6]. Identifying neurodevelopmental outcomes is a priority for children in follow up care; children with developmental needs may require additional support for physical and learning disabilities. The Vineland Adaptive Behavior Scales II© (VABS-II; Pearson Education Incorporated, San Antonio, Texas) parent report questionnaire is a tool that has been standardised for factors including gender, race, age and parental education, to identify children with developmental delay in the domains of communication, daily living skills, social skills, motor skills and behaviour. VABS-II is useful as an adaptive functioning inventory that could be completed at home without time consuming hospital visits and assessments. VABS-II has been used in a few children with CHI [7] but its reliability as a general screening tool to assess developmental delay in this population has not been assessed. VABS-II could be a credible tool to screen for adverse neurodevelopment in children with CHI, particularly at a younger age before formal time consuming cognitive testing is feasible. In this study, we have investigated the utility of VABS-II questionnaires as a parent report screening tool to identify developmental abnormalities in a relatively large population of children with CHI.

Aims
We aimed to 1. investigate performance of VABS-II to identify developmental delay in CHI and 2. identify patient factors correlating with VABS-II scores.

Methods
Parents of a cohort of children with CHI (n = 64), presenting consecutively between 2013 and 2015 to a specialist CHI treatment centre, completed the VABS-II questionnaire following consent. The diagnosis and treatment of CHI was based on established criteria and clinical practice [1,2]. Medical and surgical treatment was individualised for each child. Patients' characteristics and clinical outcome data were obtained from a database of patients. CHI was considered early (Early-CHI) if hypoglycaemia presentation was in the first month of life. Children who presented with hypoglycaemia later than one month had Late-CHI. In such children, neonatal records did not provide evidence for persistent hypoglycaemia. Following hospital discharge, children with Early and Late-CHI were assessed in the outpatient department by the clinical team comprising of clinicians, specialist nurse practitioners, dieticians, speech and language therapists and one clinical psychologist. VABS-II was discussed with parents as a routine screening tool for development after the age of 1 year.
VABS-II is a validated measure of intellectual and developmental functioning, and has been used in children with neonatal conditions [8], in neurological problems [9] and in children with genetic problems [10]. VABS-II has also been used in a small cohort of children with CHI, but its reliability as a screening tool has not been assessed [7]. Although VABS-II can be applied in children from birth, milder forms of developmental delay may not be apparent until an older age when clear progress in several developmental domains is obvious. Therefore, the minimum age for using the VABS-II was chosen at 18 months. No upper limit was specified; however as scores for motor skills in children > 6 years are estimates, analysis of VABS-II scores were run both with and without children > 6 years.
The VABS-II questionnaire was posted out to parents by the clinical psychologist (JN) who was trained and accredited to use and interpret the VABS-II. Where necessary she contacted parents by telephone to discuss queries about VABS-II responses. Populated questionnaires were returned to her for analysis in each of the domains of Communication, Daily Living Skills, Social Skills and Motor Skills [http://www.pearsonclinical.com/]. Domain scores were then compounded to derive the Adaptive Behaviour Composite (Total) score. For each VABS-II domain, scores were converted to Standard Deviation Scores (SDS) based on mean (SD) 15 (3). Total VABS-II scores were also  converted to Total VABS-II SDS (VABS-II Total) based on mean (SD) 100 (15), (total scores are not additive). For each VABS-II domain and for composite scores, SDS < -2.0 was indicative of significant developmental delay. The Behaviour component of VABS-II was independently scored for internalising, externalising and total maladaptive behaviour scores, with raw scores having inverse correlation with behaviour outcomes. High scores corresponded to poor behaviour outcomes; for total Behaviour, scores ≥ 20 were considered unsatisfactory.
VABS-II was assessed for repeat variability by comparing initial report with a second report in a volunteer group (n = 7) after at least 1 year. The repeat assessment was performed to investigate if VABS-II demonstrated variability in relation to the timing of the test that could affect the interpretation of results. VABS-II was also analysed in a group of children with idiopathic ketotic hypoglycaemia (IKH) with normal neurodevelopment (n = 9) to assess test performance in an alternative condition of hypoglycaemia without significant adverse neurodevelopment.
VABS-II scores were compared with objective developmental assessment performed within 6 months of reporting. This assessment was performed by developmental paediatricians, clinical psychologists and educational psychologists who were unaware of VABS-II performance scores. The parents of children with CHI were unaware of the VABS-II scores and report until after the objective developmental assessment was performed. However, this testing was not centralised to the CHI centre; instead objective developmental assessment relied on methods specific to the local health authority, and were blinded to the results of VABS-II scores. However, derogation to local services meant that uniformity of formal testing was not maintained although the choice of formal developmental assessment allowed flexible testing in children of all abilities. The following developmental assessments were utilised -Wechsler Preschool and Primary Scale of Intelligence for Children -UK 4th Edition (WPPSI-IV), Wechsler Intelligence Scales for Children, 4th edition (WISC-IV UK), Movement Assessment Battery for Children -UK Second Edition (MAS-2), Bayley Scales of Infant and Toddler Development -Third Edition (Bayley-III) and Griffiths Developmental Scales. Objective assessment reports were available in 15 children, 6 from our centre and 9 from elsewhere. In the rest, information describing cognitive and developmental outcomes was obtained from patient clinical correspondence and reports obtained from community paediatricians and school assessments by educational psychologists. Formal or informal developmental testing was performed independent of VABS-II testing in all children in the cohort; therefore the study design did not control for the severity of neurodevelopmental outcome. As these tests varied in their reporting styles, no attempt was made to achieve uniformity of output, except for recording the presence or absence of delay in one or more domains of childhood development in the following categoriesgross motor, fine motor, social and adaptive, communication and language. Brain neuroimaging was not performed routinely but reserved for clinical need.
VABS-II scores were also investigated for correlations with the timing of hypoglycaemia presentation, gender and phenotypes of CHI which included focal (solitary hyperfunctioning lesion in the pancreas), diffuse (hyperfunction in all islets in the pancreas) and transient (resolving hypoglycaemia, not requiring surgical treatment or long term medical therapy) forms, and treatment response. Transient and persistent CHI were defined as per previous descriptions [3,11] with persistent forms indicating requirement for medication or need for pancreatic surgery. Genetic mutation status was determined by testing for known genes associated with CHI as previously described using standard methods [11]. Mutation status was positive if any pathologic mutation was present, in heterozygous or homozygous form, regardless of the mode of inheritance. CHI gene mutation was utilised as a proxy for greater severity, with known genetic forms having a greater requirement for medical or surgical therapy and less likely to achieve resolution of disease [11]. However, it is accepted that severity can be variable within individuals with the same genotype, between heterozygous, homozygous and compound heterozygous mutations and between K-ATP channel genes and non K-ATP channel genes. It is also possible that children without genetic mutations may have severe disease. While genetic mutation status is not an ideal severity marker, other markers such glucose infusion rates were not obtainable in patients with mild forms of CHI and those presenting late. We also utilised other severity markers such as response to diazoxide, transient or persistent CHI and requirement for surgery, although we accept that such markers are not validated, may be non-concordant, may represent a disproportionately severe end of the spectrum of disease and introduce bias in statistical correlations.
VABS-II was also tested in 9 children with IKH (age range 3.00 to 5.40 years), to assess performance in children with hypoglycaemia not due to CHI. These children presented with ketotic forms of hypoglycaemia in 2014-2015 and underwent investigations to exclude known causes of hypoglycaemia including CHI. Children who were recruited to the study did not have formal developmental assessment. However they were reviewed by the clinical psychologist who ensured normal neurodevelopment. Their hypoglycaemia was not treated by regular medication/ food supplements but instead emergency hypoglycaemia prevention protocols incorporating additional carbohydrate intake during illness episodes were adopted.
Statistical analysis was performed by SPSS IBM© version 23.0 (IBM, New York, USA). VABS-II SDS between groups were compared by non-parametric tests. For repeat samples, paired tests with unequal variances were utilised. Probability of group membership was assessed by odds ratio, while sensitivity and specificity of tests were checked by receiver operating characteristic (ROC) curve analysis. The study was supported by the North West Research Ethics Committee, Project Reference Number 07/H1010/88.  Fig. 1]. VABS-II scores were in the normal population range [-0.33 (-1.73, 1.13)] for 9 children with IKH, in keeping with prior normal objective developmental assessment. VABS-II was repeated in 7 children (age range 3.00 to 9.30 years) with CHI; individual domain and total scores remained similar [paired samples test, p = 0.18 to p = 0.95], suggesting that the VABS-II was valid on repetition without significant deviation with advancing age.

VABS-II scores in CHI correlate with developmental delay
VABS-II scores were correlated with developmental delay (affected in at least one developmental domain) identified by objective developmental assessment [ Table 1]. Correlation was also sustained for involvement of one or more developmental domains regarded as a yes/no binary variable [odds ratio, OR (confidence interval, CI) 0.28 (0.13, 0.61), p = 0.001]. As motor skills scores were estimated in a group of children > 6 years old (n = 14), the analysis was re-run in the subgroup of children    Fig. 3].
(2)Linking VABS-II scores to CHI severity: VABS-II scores were analysed for correlation with disease severity. Various phenotypic characteristics describing severity of CHI in our cohort has been provided in Appendix. Carriage of CHI gene mutations serving as a possible proxy for greater severity correlated with lower VABS-II Total scores in analysis of covariance with age at presentation as the covariate [adjusted R 2 = 0.19, p = 0.039] with a modest effect size of 10%. In keeping with mutation status as a marker of severity, responsiveness to diazoxide was also positively correlated with VABS-II, i.e. unresponsiveness to diazoxide was associated with lower VABS-II Total scores (p = 0.019); however, no significant effects were noted for the following factors: transient or permanent CHI (p = 0.413), focal CHI (p = 0.742) and pancreatic surgery (p = 0.132).

Discussion
In this study, we have assessed the reliability of VABS-II as a screening tool for developmental delay in children with CHI. VABS-II and maladaptive behaviour scores individually correlated with developmental delay. VABS-II had high specificity, indicating accuracy of screening investigations for formal developmental assessments. Male gender and late age of hypoglycaemia presentation were risk factors for lower VABS-II scores. Diazoxide unresponsiveness and carriage of CHI genetic mutations, proxy markers of severity of CHI, were also associated with lower VABS-II scores.
Although VABS-II is applied widely to children with different conditions, its utility and accuracy in children with CHI has not been investigated previously. Our study has reviewed the performance of VABS-II in a relatively large cohort of children with the rare disease of CHI, over a two year period and correlated parent reports with objective clinical developmental examination. Further, our study examined VABS-II in IKH, where hypoglycaemia is not usually associated with significant brain injury; as expected VABS-II showed normal variation in IKH individuals. Our study also found consistency in repeating VABS-II in later life, thereby eliminating age dependent bias. The strength of association of VABS-II with developmental delay and the observation that a third of children have abnormal neurodevelopment [3] makes a compelling case to use VABS-II routinely in children with CHI with any clinical concern about neurodevelopment.
Although VABS-II demonstrated strong correlation with developmental delay and had high specificity, formal assessment methods to diagnose developmental delay were not uniform. However as cognitive testing using inventories such as WISC-IV UK can be time consuming, require trained psychologists and may be unsuitable for younger children, flexibility of objective developmental assessment was inevitable. Nonetheless, the non-uniformity of formal developmental testing remains a weakness, which may account for the skewed sensitivity values of VABS-II. It does reflect the variability that exists within clinical practices, however.
The study design did not specify formal developmental testing for all patients. It is possible that more patients who had severe neurodevelopmental delay were tested by formal methods than those with mild delay. However, in those tested formally in our cohort, nearly half were tested routinely. In such patients, testing was not biased by the severity of adverse neurodevelopment. It is possible that the sensitivity of VABS-II for milder adverse neurodevelopment is higher than that observed. An important corollary from our observations is the need for a more rigorous test of performance of VABS-II in the detection of milder forms of developmental delay. We recognise that mild abnormalities due to early life hypoglycaemia may not be reversed by early detection by VABS-II; however it is possible that early therapeutic interventions and adaptions in home and learning environments could be beneficial. The identification of relatively subtle neurodevelopmental abnormalities and interventions on quality of life could be a logical follow on study.
In this study, brain imaging was not performed as routine. Instead brain magnetic resonance imaging was prioritised for clinical need as in a previous observational study [3]. While the absence of brain imaging could be construed as a weakness in the study design, the value of routine brain imaging, a resource intense investigation, has not been substantiated for neurodevelopmental screening of CHI other than to identify the topography of lesions [12].
VABS-II had high specificity but low sensitivity for developmental delay. Therefore in the context of screening, VABS-II may not be relied upon as a primary tool for developmental referral in CHI. However, in clinical practice, developmental concerns are routinely discussed in outpatient follow up, which would obviate the requirement for VABS-II to be a highly sensitive instrument. The clinical suspicion of delayed development could be construed as a first line screening test, followed by VABS-II as a next step screening tool. The high specificity of VABS-II in the context of clinical concerns should generate sufficient concern to trigger referral for formal developmental assessments.
It is not clear why males with CHI have lower VABS-II scores than females. Frequency of males was greater in our cohort; however, gender as a variable was controlled in analysis at several levels. Our study has raised the interesting question whether gender difference could be an independent predictor over hypoglycaemic brain injury to cause intellectual disability. This question cannot be answered within the remit of our study design and needs to be examined in larger cohorts. Recent observations suggest modulation of sensitivity of arcuate nuclei to hypoglycaemia by suprachiasmatic nuclei in male rats [13]; it remains to be seen if similar mechanisms apply to male brains in children with CHI.
Our study shows that children presenting later with CHI have lower VABS-II scores, correlating with adverse developmental outcomes. This observation remains even after adjustment for male gender, another risk factor for lower VABS-II scores. CHI usually presents in the neonatal period; it is possible that patients with Late CHI had neonatal hypoglycaemia but that the diagnosis was achieved later. However, examination of neonatal records makes this less likely, although not impossible. A previous observation of later presentation has been associated with long term neurological disabilities [4] suggesting recurrent and unrecognised hypoglycaemia impacting on developmental outcomes. In our cohort, it is likely that recurrent hypoglycaemia was missed prior to the eventual diagnosis of CHI. Such hypoglycaemic episodes could be responsible for greater severity of adverse neurodevelopment in those with Late CHI. The inverse correlation with age at presentation indicates the need for early recognition and treatment of hypoglycaemia in early life [1,14]. As expected, diazoxide unresponsiveness and carriage of genetic mutations, proxy markers of severity of CHI, were associated with lower VABS-II scores. It therefore follows that treatment response and gene mutation testing in the initial phase of clinical management may provide prognostic markers to determine neurodevelopmental trajectories in CHI. It is well recognised that patients undergoing pancreatectomy for CHI have a high frequency of neurobehavioural deficits [15]. Our study adds further evidence to the impact of severity phenotyping on long-term outcomes.

Conclusions
We have evaluated the performance of Vineland Adaptive Behavior Scales, 2nd edition (VABS-II) in children with CHI and noted lower scores correlating with the presence of developmental delay with high specificity. Male gender, late age at presentation and severity of CHI are risk factors for adverse outcomes. VABS-II can be reliably used in neurodevelopmental follow up of CHI patients to trigger formal developmental assessment.