Association testing of copy number variants in schizophrenia and autism spectrum disorders.

Bernard J Crespi,Helen J Crofts

doi:10.1186/1866-1955-4-15

Bernard J Crespi, Helen J Crofts

Open Access

https://doi.org/10.1186/1866-1955-4-15

Copy DOI

Abstract

BackgroundAutism spectrum disorders and schizophrenia have been associated with an overlapping set of copy number variant loci, but the nature and degree of overlap in copy number variants (deletions compared to duplications) between these two disorders remains unclear.MethodsWe systematically evaluated three lines of evidence: (1) the statistical bases for associations of autism spectrum disorders and schizophrenia with a set of the primary CNVs thus far investigated, from previous studies; (2) data from case series studies on the occurrence of these CNVs in autism spectrum disorders, especially among children, and (3) data on the extent to which the CNVs were associated with intellectual disability and developmental, speech, or language delays. We also conducted new analyses of existing data on these CNVs in autism by pooling data from seven case control studies.ResultsFour of the CNVs considered, dup 1q21.1, dup 15q11-q13, del 16p11.2, and dup 22q11.21, showed clear statistical evidence as autism risk factors, whereas eight CNVs, del 1q21.1, del 3q29, del 15q11.2, del 15q13.3, dup 16p11.2, dup 16p13.1, del 17p12, and del 22q11.21, were strongly statistically supported as risk factors for schizophrenia. Three of the CNVs, dup 1q21.1, dup 16p11.2, and dup 16p13.1, exhibited statistical support as risk factors for both autism and schizophrenia, although for each of these CNVs statistical significance was nominal for tests involving one of the two disorders. For the CNVs that were statistically associated with schizophrenia but were not statistically associated with autism, a notable number of children with the CNVs have been diagnosed with autism or ASD; children with these CNVs also demonstrate a high incidence of intellectual disability and developmental, speech, or language delays.ConclusionsThese findings suggest that although CNV loci notably overlap between autism and schizophrenia, the degree of strongly statistically supported overlap in specific CNVs at these loci remains limited. These analyses also suggest that relatively severe premorbidity to CNV-associated schizophrenia in children may sometimes be diagnosed as autism spectrum disorder.

Highlights

Autism spectrum disorders and schizophrenia have been associated with an overlapping set of copy number variant loci, but the nature and degree of overlap in copy number variants between these two disorders remains unclear
An alternative yet non-exclusive hypothesis, originally suggested by Eliez [18] in the context of 22q11.2 deletions, is that for some loci, sharing of copy number variants (CNVs) between schizophrenia and autism is more apparent than real, because of false-positive diagnoses of relatively severe, CNV-associated premorbidity to schizophrenia as autism or autism spectrum disorder [19]. The presence of such false positives is predicated on the supposition that autism or autism spectrum disorders and schizophrenia represent distinct conditions with partially overlapping childhood psychological deficits and abnormalities, which are due to different neurodevelopmental causes
CNVs were included for analysis if they met three criteria: (1) previous evidence for statistical association of the CNV with schizophrenia, autism, or both disorders; (2) reports of the CNV in both conditions, from case control, family-based, or case series studies (CNV studies based on ascertainment of a range of childhood conditions, followed by focused analysis on one or more specific CNVs); and (3) sufficient information on phenotypes associated with the CNVs, especially intellectual disability and developmental, speech or language delay, to assess its effects on childhood development, bearing in mind how individuals were ascertained

Summary

Introduction

Autism spectrum disorders and schizophrenia have been associated with an overlapping set of copy number variant loci, but the nature and degree of overlap in copy number variants (deletions compared to duplications) between these two disorders remains unclear. Such findings implicate overlapping genetic risk factors and shared etiology, which has been postulated to help explain shared phenotypes, mainly deficits and abnormalities in social development and language This interpretation, runs counter to a large body of non-genetic evidence that differentiates autism spectrum disorders from schizophrenia [14,15,16,17], including an exclusionary condition with regard to autism in DSM-IV. An alternative yet non-exclusive hypothesis, originally suggested by Eliez [18] in the context of 22q11.2 deletions, is that for some loci, sharing of CNVs between schizophrenia and autism is more apparent than real, because of false-positive diagnoses of relatively severe, CNV-associated premorbidity to schizophrenia as autism or autism spectrum disorder [19] The presence of such false positives is predicated on the supposition that autism or autism spectrum disorders and schizophrenia represent distinct conditions with partially overlapping childhood psychological deficits and abnormalities, which are due to different neurodevelopmental causes

Methods

Results

Discussion

Conclusion