Background Developmental coordination disorder is a common neurodevelopment disorder that frequently co-occurs with other neurodevelopmental disorders including attention-deficit hyperactivity disorder (ADHD). Copy-number variations (CNVs) have been implicated in a number of neurodevelopmental and psychiatric disorders; however, the proportion of heritability in developmental coordination disorder (DCD) attributed to CNVs has not been explored.
Objective This study aims to investigate how CNVs may contribute to the genetic architecture of DCD.
Methods CNV analysis was performed on 82 extensively phenotyped Canadian children with DCD, with or without co-occurring ADHD and/or reading disorder, and 2988 healthy European controls using identical genome-wide SNP microarrays and CNV calling algorithms.
Results An increased rate of large and rare genic CNVs (p=0.009) was detected, and there was an enrichment of duplications spanning brain-expressed genes (p=0.039) and genes previously implicated in other neurodevelopmental disorders (p=0.043). Genes and loci of particular interest in this group included: GAP43, RBFOX1, PTPRN2, SHANK3, 16p11.2 and distal 22q11.2. Although no recurrent CNVs were identified, 26% of DCD cases, where sample availability permitted segregation analysis, were found to have a de novo rare CNV. Of the inherited CNVs, 64% were from a parent who also had a neurodevelopmental disorder.
Conclusions These findings suggest that there may be shared susceptibility genes for DCD and other neurodevelopmental disorders and highlight the need for thorough phenotyping when investigating the genetics of neurodevelopmental disorders. Furthermore, these data provide compelling evidence supporting a genetic basis for DCD, and further implicate rare CNVs in the aetiology of neurodevelopmental disorders.
Statistics from Altmetric.com
Recent advances in genomic techniques such as high-resolution microarrays and next-generation sequencing have yielded significant insights into the genetic aetiology of neurodevelopmental disorders.1 In particular, structural genomic variation has been implicated in the aetiology of a proportion of children diagnosed with autism spectrum disorder (ASD), attention-deficit hyperactivity disorder (ADHD), intellectual disability, schizophrenia and cerebral palsy.2–5 Studies of ASD cohorts have identified an increased burden of rare genic copy-number variations (CNVs) and have characterised rare, usually de novo, recurrent CNV loci that are thought to contribute to the genetic risk.6 Specific genes within these CNV regions are implicated in the aetiology of ASD (eg, SHANK3, SYNGAP1 and DLGAP2) as well as other neurodevelopmental disorders.7–9 As the number of candidate genes and loci has increased, a striking recurrence of candidates identified in multiple disorders has been uncovered, which may account for a proportion of the significant comorbidity that has been noted among neurodevelopmental disorders.10 ,11 Identifying overlapping genes and pathways across disorders is critical to improve the understanding of their potential shared genetic aetiology and may inform future diagnostic standards and therapeutic interventions.
Developmental coordination disorder (DCD) is a neurodevelopmental disorder characterised by functional motor performance deficits in postural control, motor learning and sensorimotor processing that appear in early childhood and interfere with activities of daily living and academic achievement. These motor deficits are pervasive and cannot be explained by an intellectual disability, visual impairment or other identifiable neurological condition affecting movement.12 DCD affects 5–6% of school-aged children, making it one of the most common paediatric neurodevelopmental disorders.12 ,13 There is evidence that DCD is highly heritable with estimates approaching 70%.14 ,15 This estimate is similar to other neurodevelopmental disorders such as ASD (90%), ADHD (79%) and reading disorder (RD) (67%),15–17 suggesting that the aetiology of DCD is in part attributable to genetic variation.
Consistent with other neurodevelopmental disorders, DCD is phenotypically heterogeneous, with up to 69% of children meeting criteria for at least one other neurodevelopmental disorder,13 and approximately 50% of children with DCD being diagnosed with ADHD.14 ,18 The aetiology behind this extensive comorbidity is not well understood, but evidence suggests that there may be a shared genetic susceptibility underlying these neurodevelopmental disorders.15 ,19 To date, however, little is known about the genetic aetiology of DCD. This study investigates the impact of CNVs on the genetic aetiology of DCD by examining the CNV landscape in 82 Canadian school-aged children with DCD. The results, which identify an enrichment of rare genic CNVs that overlap known neurodevelopmental genes and loci, suggest that DCD should be added to the growing list of neurodevelopmental disorders whose aetiology can be partly attributed to rare CNVs.
Participants were recruited as part of a multidisciplinary study at the Alberta Children's Hospital in Calgary, Alberta, Canada, with approval from the Conjoint Health Research Ethics Board of the University of Calgary. This study included children between the ages of 8 and 17 who were recruited, assessed for and met diagnostic criteria for DCD with or without co-occurring ADHD and/or RD. Exclusion criteria were gestation <36 weeks, low birth weight, intellectual disability (IQ<70), seizure disorder and conditions that could affect cognitive or motor function. DNA from participants and available family members was extracted from peripheral blood leucocytes with informed consent. Children were identified with DCD if they scored ≤16th percentile on the Movement Assessment Battery for Children, Second Edition (MABC-2; Pearson Assessment Canada, Toronto, Ontario, Canada) and were reported by their parents as displaying motor difficulties that interfered significantly with daily functioning on the Developmental Coordination Disorder Questionnaire (DCDQ; table 1). Parents were also assessed for DCD using the DCDQ.20
The Wechsler Abbreviated Scale of Intelligence (WASI; Pearson Assessment, San Antonio, Texas, USA) was used to assess IQ. The Diagnostic Interview for Children and Adolescents-IV (DICA; Psych Press, Melbourne, Australia) and the Conners' Rating Scale—Revised (CPRS-R; Pearson Assessment Canada, Toronto, Ontario, Canada) were used to identify children with ADHD. The Conners' Adult ADHD Rating Scales-R (CAARS; Pearson Assessment Canada, Toronto, Ontario, Canada) was used to measure attention-deficit symptoms in parents. Finally, children were identified as having a RD if they scored ≥1.00 SD below the mean on the Wechsler Individual Achievement Test-Second Edition (Canadian Edition) (WIAT-2CND, Pearson Assessment Canada, Toronto, Ontario, Canada). Parents were assessed for RD using the Adult Reading History Questionnaire.21 The study cohort included 82 unrelated children: 25 with isolated DCD, 5 with DCD and RD, 36 with DCD and ADHD, and 16 with DCD, ADHD and RD. Two European control data sets (N=2988) were used in all case–control comparisons.22 ,23
Eighty-two participants and 2988 controls were genotyped on the Illumina HumanOmni2.5-Quad BeadChip SNP array (San Diego, California, USA) at The Centre for Applied Genomics (TCAG) in Toronto according to the manufacturer's protocol. Validation and segregation of CNVs was performed by qPCR using Life Technologies SYBR Green (Carlsbad, California, USA) on an ABI 7900HT Real-Time machine (Carlsbad, California, USA). Participant genotypes were clustered using 204 768 unlinked autosomal SNPs with a minor allele frequency (MAF) ≥5% in a multidimensional scaling analysis in PLINK to determine putative European ancestry. Outlier detection analysis was performed in PLINK in combination with self-reported ancestry for all 82 participants to identify a subcohort of DCD participants of European ancestry. Cycle sequencing was performed using BigDye Terminator V1.1 by Life Technologies (Carlsbad, California, USA), and analysed with the ABI 3130xl Genetic Analyzer (Carlsbad, California, USA). Variants were detected and analysed with Mutation Surveyor software from SoftGenetics (State College, Pennsylvania, USA).
CNV detection and analysis
CNV calling was performed as previously described.2 In brief, four calling algorithms were used to detect CNVs in cases and controls: iPattern, PennCNV, QuantiSNP and CNVPartition. Only stringent CNVs (those that were called by at least two algorithms and spanned at least five consecutive probes) were investigated. A CNV was described as rare if it was found in <0.1% of controls, or if <50% of its total length overlapped another CNV found in >0.1% of all samples, and was not found in the Database of Genomic Variants (DGV) or internal TCAG controls described by Lionel et al.2 To be consistent with previous studies, only rare, stringent CNVs that were at least 20 kb in length were used when performing burden and enrichment analyses, which allowed results presented here to be compared with previously reported data. Smaller individual CNVs, however, were investigated further if they overlapped genes previously implicated in other neurodevelopmental disorders.
Global burden and enrichment analyses of rare, stringent CNVs were performed using PLINK for subsets of CNVs, similarly to methods described by Pinto et al.24 Genic regions, which included all isoforms, were ascertained by RefSeq annotation (UCSC, V. February 2009, NCBI V37, hg19). To investigate whether there was an enrichment of certain genes relative to all genic CNVs, gene set enrichment analyses were performed in PLINK using methods described by Raychaudhuri et al.25 Gene lists for brain-expressed genes and genes implicated in other neurodevelopmental disorders were derived and described by Pinto et al.24 All analyses applied the permutation procedure to correct for multiple testing and yielded one-sided (enrichment in cases) empirical p values based on 10 000 permutations. This procedure provided a robust alternative for calculating corrected p values to the Bonferroni correction method. The permutation procedure randomly permuted the case–control groups and recalculated all test results for 10 000 permutations to calculate a distribution of p values. The permuted distribution was compared with the actual test p values to calculate a corrected sample p value.
The analysis of ancestry identified a subcohort of 67 participants of European ancestry that were ethnically matched to our control population (see online supplementary figure S1). This European subcohort (abbreviated as DCD-Euro herein) was used in all case–control genome-wide comparisons of CNVs to avoid confounds of population substructure. All individuals from the DCD cohort (N=82), however, were investigated for individual CNVs overlapping regions previously implicated in other neurodevelopmental disorders, which did not require a statistically robust case–control comparison. A smaller subcohort of 20 individuals of European ancestry with a diagnosis of isolated DCD (abbreviated as Isolated-DCD-Euro herein) was also investigated in case–control comparisons. A similar mean and median number of stringent CNVs >20 kb per genome, as well as a similar median size for all CNVs, was observed in our DCD cohort, DCD-Euro cohort, Isolated-DCD-Euro cohort and controls (see online supplementary table S1). In total, 3213 (1208 >20 kb) stringent CNVs, and 340 (106 >20 kb) rare CNVs were detected in the DCD-Euro cohort (see online supplementary tables S1 and S2). Out of the total CNVs identified, we chose 27 to validate by qPCR, all of which confirmed (see online supplementary figure S2 and table S3). This underscores the robustness of our CNV calling algorithm. Segregation analysis of rare CNVs was performed in 19 participants where parental samples were available. Six of the rare CNVs were maternally inherited, eight were paternally inherited and five were de novo. Of the 14 inherited CNVs, 9 were inherited from a parent who scored in the clinical range for at least one neurodevelopmental disorder.
Differences in CNV rate and the proportion of samples with a rare CNV between cases in the DCD-Euro cohort and controls were investigated for both duplications and deletions between and above 20 kb, 100 kb, 500 kb and 1 mb (see online supplementary table S4). Significant differences were observed in the number of rare CNVs per genome above 500 kb (p=0.011), and this difference was explained by CNVs in the range of 500 kb to 1 mb (p=0.018) (figure 1 and see online supplementary table S4). Between 500 kb and 1 mb, there was also a significant difference in the proportion of cases with at least one rare CNV compared with controls (p=0.009) (figure 1A, B). When deletions and duplications were analysed separately, there was a significant difference in the CNV rate for deletions above 500 kb (p=0.044), but not between 500 kb and 1 mb (p=0.054) (see online supplementary table S4). There were, however, significant differences in the number of cases with at least one deletion between 500 kb and 1 mb (p=0.045) or duplication above 500 kb (p=0.044) compared with controls (see online supplementary table S4). When considering the Isolated-DCD-Euro participants, results were similar to the DCD-Euro cohort with a significant difference in the number of rare CNVs per genome above 500 kb (p=0.039), though this difference could not be explained by either the CNVs between 500 kb and 1 mb (p=0.072) or above 1 mb (p=0.229) (see online supplementary table S5). Significant differences were also seen in the number of deletions per genome above 500 kb (p=0.009) and between 500 kb and 1 mb (p=0.029), as well as the number of cases with at least one deletion between 500 kb and 1 mb (p=0.023) (see online supplementary table S5).
PLINK was also used to perform global burden analysis of CNVs overlapping genes annotated in RefSeq in the DCD-Euro and Isolated-DCD-Euro cohorts (UCSC, V. February 2009, NCBI V37, hg19) (see online supplementary tables S6 and S7, respectively). Significant differences were observed for the number of genes included in CNVs above all size ranges (p<0.05), except 1 mb, in DCD-Euro cases compared with controls. These differences may be explained by CNVs between 500 kb and 1 mb CNV (4.91-fold increase (p=0.009)) (see online supplementary table S6). These results were mirrored when considering duplications alone but not deletions (see online supplementary table S6). A statistically significant difference was observed in the number of CNVs spanning at least one gene above 500 kb (2.97-fold increase (p=0.003)) and between 500 kb and 1 mb (3.06-fold increase (p=0.010)) in the DCD-Euro cases compared with controls (see online supplementarytable S6 and figure S1C, D). This difference was also observed for duplications above 500 kb (2.79-fold increase (p=0.022)) and between 500 kb and 1 mb (2.93-fold increase (p=0.033)), but not deletions. In addition, gene density in CNVs as represented by the average number of genes per kb was greater between 500 kb and 1 mb for duplications (3.16-fold (p=0.034) increase) and above 1 mb for deletions (3.68-fold increase (p=0.016)) in the DCD-Euro cases (see online supplementarytable S6). When the Isolated-DCD-Euro cases were considered, similar results were found, except that the numbers of genes spanned by CNVs were enriched between 100 and 500 kb (2.33-fold increase (p=0.038)) as well as above 1 mb (6.95-fold increase (p=0.044)), though the number of rare CNVs in each category was very small. No significant differences were found for the number of rare CNVs with at least one gene, except for deletions above 500 kb (7.3-fold increase (p=0.032)) (see online supplementarytable S7).
Since cases showed an increased rate of genic CNVs, enrichment analysis was performed in PLINK to investigate whether there was an over-representation of certain types of genes located within genic CNVs.25 There was enrichment in the number of brain-expressed genes (p=0.038) and genes involved in other neurodevelopmental disorders (p=0.045) included in the duplications between 500 kb and 1 mb in the DCD-Euro cohort compared with controls. In the same size range, the DCD-Euro cohort also showed an enrichment in the number of deletions that overlapped brain-expressed genes (p=0.039) and genes involved in other neurodevelopmental disorders (p=0.043) (table 2). This analysis was also performed for the Isolated-DCD-Euro cases, and no significant differences were detected, except for the number of genes contained within CNVs for both brain-expressed genes and genes previously implicated in other neurodevelopmental disorders between 20 and 100 kb (see online supplementary table S8). Summary statistics are provided in the online supplementary table S9.
A total of 15 rare CNVs in 11 participants overlapped regions and/or genes previously implicated in ADHD, ASD, epilepsy, anxiety disorders, schizophrenia and/or Tourette syndrome (table 3). Among five children with isolated DCD was a 300 kb deletion spanning two exons of RBFOX1, an RNA-binding protein that has been implicated in ADHD, ASD and epilepsy.26 Two other participants with isolated DCD had intronic deletions in FHIT, a protein involved in purine metabolism, which has previously been implicated in ASD.27 Though small, an approximately 19.8 kb duplication was identified in a male participant with isolated DCD that overlapped the final 2 exons of SHANK3, a well-known and characterised ASD gene.7
In the six participants with co-occurring DCD and ADHD (with or without RD), four large duplications and three large deletions were identified (table 3). These included a female child with a 16p11.2 duplication, for which ADHD is a commonly associated phenotype.2 This participant also had a 321 kb deletion within VIPR2, a neuropeptide which has been previously associated with schizophrenia that was inherited from her mother who screened positive for DCD.28 Three CNVs in PTPRN2 were identified in a male child with DCD, ADHD and RD; PTPRN2 has been previously associated with ADHD.2 ,26 A 1.3 mb de novo deletion was identified in another male child with DCD, ADHD and RD that spanned the 22q11.2 distal deletion syndrome region. This regional deletion has been described in other individuals with features that include motor delays, and behavioural, learning and cognitive difficulties.29 A 913 kb de novo deletion spanning GAP43 and LSAMP, both neuronal regeneration and outgrown-associated genes, was detected in a male child with DCD and ADHD. GAP43 deletions have been associated with schizophrenia,30 and recently deletions of 3q13.2–q13.31 (which include GAP43 and LSAMP) have been implicated in a syndrome characterised by hypotonia, and motor, language and cognitive delays.31
Previous studies have identified potential candidate genes first identified through the analysis of CNVs in individuals with ASD, schizophrenia and ADHD by screening cohorts for potentially pathogenic single-nucleotide variants (SNVs).2 ,30 Deletions encompassing GAP43 are rare, and this coupled with the known biological function of GAP43 led us to assess whether SNVs in GAP43 may be contributing to the aetiology of DCD. All exons and the core promoter region of GAP43 were sequenced in an expanded cohort of 100 children with DCD, which included the 82 children in our original DCD cohort, and 5 variants of unknown significance were identified (see online supplementary table S3 and figure S3).
Here, we have described findings that suggest that there is a genetic liability in DCD explained, in part, by CNVs, and that there may be a common genetic aetiology shared between DCD and other neurodevelopmental disorders. Previous studies investigating the impact of rare CNVs in neurodevelopmental disorders have demonstrated an increased rate of rare and large CNVs overlapping genic regions. It has been proposed that differences in the rate of genic CNVs may be important in understanding the aetiology of these neurodevelopmental disorders. An increased rate of rare CNVs containing at least one gene was identified above all size ranges, except 1 mb, as well as a threefold increased rate (p<0.05) between 500 kb and 1 mb. Within these genic CNVs, our data indicate a greater number of genes included in CNVs, and a greater number of CNVs that encompass at least one gene between 500 kb and 1 mb. These findings are consistent with research on other neurodevelopmental disorders and suggest that children with DCD possess greater genomic variation involving genic regions than controls. Furthermore, we detected significant differences for both analyses when considering duplications alone, but not deletions, suggesting that the overall differences seen may be accounted for by duplications (see online supplementary table S6). It has been suggested that the shared aetiology of various neurodevelopmental disorders may be genetic in origin, so we performed the enrichment test for shared genes that have been previously associated with these conditions (eg, ADHD, ASD and OCD). Though the sample size for analyses was small (N=67), by comparing with a large control set (N=2988) we demonstrated that large deletions in children with DCD show significant enrichment (p<0.05) for brain-expressed genes and genes previously associated with other neurodevelopmental disorders. Children with DCD were also found to have more duplications (p<0.05) that spanned brain-expressed genes and genes previously implicated in other neurodevelopmental disorders (table 2).
The brain-expressed enrichment analyses performed are subject to bias due to the relatively large size of brain-specific genes; however, by using the ‘cnv-enrichment-test’ function within the PLINK toolset we were able to bypass potential confounds related to gene size, as well as differences in CNV size, and CNV rate in cases and controls.25 To reduce the number of false-positive CNV calls, a lower CNV size limit of 20 kb was used in the case–control analyses. The analysis of small CNVs detected by next-generation sequencing in a simplex autism cohort identified a higher gene density in participants with autism, suggesting a role for small CNVs in understanding the genetics of neurodevelopmental disorders.32 As such, further analysis of CNVs smaller than 20 kb in a larger cohort of DCD participants is warranted. Previous studies that investigated the impact of CNVs on neurodevelopmental disorders may not have taken into account that upwards of 60% of individuals with neurodevelopmental disorders meet diagnostic criteria for another. As such, many individuals in these cohorts would likely meet diagnostic criteria for at least one other neurodevelopmental disorder.13 Thirty per cent of the participants in the extensively phenotyped cohort presented here had a diagnosis of isolated DCD. The results of the larger DCD-Euro cohort (N=67) are comparable with those seen previously in the literature, so we analysed this small cohort of European ancestry with isolated DCD (N=20) (see online supplementary tables S5, S7 and S8). While a number of statistically significant differences were found, and the general findings had similar trends to the larger DCD-Euro cohort, further investigation with a larger participant cohort is needed.
Of the genes that were identified as having overlap in other neurodevelopmental disorders, several cases sparked particular interest. One female participant with DCD and ADHD had an inherited deletion spanning VIPR2 from a mother who had symptoms associated with DCD. This participant also had a duplication in 16p11.2, which was either de novo or paternally inherited. The neuropeptide vasoactive intestinal peptide (VIP) and its receptor (VIPR2) have a range of functions, and when knocked-out can result in the disruption of locomotion in mice.33 Microduplications in 16p11.2 are frequently associated with ADHD,2 ,34 schizophrenia35 and other cognitive and behavioural phenotypes.36 It is possible that the single-gene deletion of VIPR2 in this child may be modifying the effects of the 16p11.2 duplication, which results in the more severe phenotype of DCD with ADHD when compared with her mother who screened positive for isolated DCD. In a participant with DCD, ADHD and RD, a distal 22q11 deletion between low copy repeats (LCRs) 4 and 5 was identified. Recently, a new deletion syndrome (MIM 611867) involving the same LCRs, which are immediately distal to the common 22q11 deletion syndrome region, has been described in individuals who display developmental and motor delays, learning difficulties, behavioural problems and ADHD.29
We also identified a number of interesting CNVs within children with isolated diagnoses of DCD, including a small, 19.8 kb duplication encompassing exons 22 and 23 of SHANK3 (NM_033517.1), which includes the sterile α motif, which is essential for synaptic targeting and may play a role in the organisation of SHANK3 at the postsynaptic density.37 ,38 In addition, CNVs and SNVs within this gene account for approximately 1% of ASD cases,7 and overexpression of Shank3 has been shown to result in manic behaviour and hyperactivity in mice.39 The rare nature of variants within SHANK3 and the previously reported pathogenicity of disruptions in this gene may indicate that the duplication may be contributing to the motor phenotype in this child.
A child with DCD and comorbid ADHD was found to have a 913 kb de novo deletion encompassing both GAP43 and LSAMP. GAP43 is a highly conserved neuronally expressed gene that has important roles in controlling path finding and branching during development, and neurotransmitter release in the adult central nervous system.40 ,41 Abnormal expression and mutations in GAP43 have been observed in individuals with schizophrenia.30 ,42 Furthermore, decreased stress responses and sensory impairments, learning problems and autistic-like features have been observed in heterozygous and null mice.43 Recently, individuals harbouring 3q13.2–q13.31 deletions that include GAP43 have been described as having motor and language delays, and a variety of cognitive and behavioural issues.31 By sequencing a cohort of individuals diagnosed with schizophrenia, Shen et al.30 demonstrated that novel and de novo SNVs in GAP43 might contribute to the disorder. Since CNVs in GAP43 are very rare, and the known function is potentially biologically relevant, we sequenced the gene and characterised five novel or rare variants of unknown significance. Due to the unknown impact of these variants on gene function, the results presented here may not suggest a relationship between SNVs in GAP43 and the DCD phenotype, though functional studies may be warranted.
The present study used extensively phenotyped participants and parents, strict inclusion and exclusion criteria, dense SNP arrays, a stringent CNV calling algorithm and a very large control cohort (∼1:40 case to controls ratio). This allowed for the identification of a number of rare CNVs in participants with DCD. Validation by qPCR revealed that five rare CNVs >20 kb arose de novo in the 19 participants for which DNA from both parents was available (26.3% if considering trios and 6.1% if considering all DCD participants in the cohort). While this rate is larger than the rates previously observed in ADHD cohorts (4.81% and 7.43% found in the CHOP and IMAGE cohorts, respectively, and 1.7% reported elsewhere), the true de novo rate is likely less than this as parental DNA was not available to confirm the inheritance of all identified CNVs.2 ,44
Overall, these data demonstrate that children with DCD have a high rate of large and rare genic CNVs, and that these CNVs are over-represented for brain-expressed genes and genes previously implicated in other neurodevelopmental disorders. We also showed that children with DCD (with or without co-occurring ADHD and RD) possess rare CNVs that overlap genes previously implicated in other neurodevelopmental disorders, supporting the hypothesis that these disorders share a common genetic aetiology. Although requiring validation in a large replication cohort, our results strongly suggest that the genetic component of DCD can be explained partly by CNVs, and that analysing CNVs is a viable approach to investigate the genetic basis of this common neurodevelopmental disorder.
We wish to thank the participants and their family members for their ongoing contributions to this project. This work is part of a collaborative effort between Dr Deborah Dewey's research group in the Behavioural Research Unit, Department of Pediatrics, Alberta Children's Hospital, University of Calgary, and Dr Francois Bernier's and Dr Jillian Parboosingh's research group in Medical Genetics at the Alberta Children's Hospital, University of Calgary, and Dr Stephen Scherer's group at The Centre for Applied Genomics, Toronto. This work was supported by the Canadian Institute of Health Research, MOP-86588.
Contributors FPB reviewed the suggested guidelines for authorship and confirmed each of the authors meet all four of the suggested criteria for authorship.
Funding Institute of Human Development, Child and Youth Health (MOP-86588).
Competing interests None declared.
Ethics approval Conjoint Medical Health Research Committee – University of Calgary.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Currently, unpublished raw SNP array data have not been shared in public database and remain the intellectual property of the investigators. The authors do favour data sharing and will endeavour to do once our research on the CNVs and DCD is completed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.