Diagnostic Yield of Chromosomal Microarray Analysis in Autism Spectrum Disorder

Background: The genetic underpinnings of Autism Spectrum Disorder (ASD) are very heterogeneous and can range from monogenic and syndromic cases to oligogenic/polygenic forms. Rare and common copy number variants (CNVs) can contribute to the aetiology of the disorder and differences in CNV burden can contribute to the substantial clinical heterogeneity observed in ASD. For this reason, array-CGH have become a first-tier diagnostic test in the medical work-up of autistic individuals.

Objectives: The aim of this study is to identify pathogenetically-relevant CNVs in 254 ASD families (217 simplex and 37 multiplex) and to verify rates of de novo vs inherited pathogenic CNVs.

Methods: A total of 275 ASD subjects from 217 simplex and 37 multiplex families were consecutively admitted to the Service for Neurodevelopmental Disorders at Campus Bio-Medico University Hospital (Rome, Italy) and to the Interdepartmental Program “Autism 0-90” at the “G. Martino” University Hospital (Messina, Italy), between the years 2015 – 2017. Blood was drawn from patients and first-degree relatives, DNA was extracted from leukocytes and array-CGH analysis was performed using the Human Genome CGH SurePrint G3 Microarray 4 x 180K Kit (Agilent). Array-CGH outcomes were blindly classified by four authors according to five different categories ranging from “absence of functionally relevant CNVs” to “certainly causal CNV”. Data were imported on a MySQL RDBMS instance and analyzed through SQL queries.

Results: “Certainly causal CNVs” and “probably causal CNVs” were detected in 22/275 (8.0%) and in 44/275 (16.0%) ASD patients, respectively. The frequency of duplications and deletions is comparable. However, “certainly causal CNVs” are significantly more often de novo (Chi sq. 8.46; 2 d.f., P<0.05), and in particular de novo deletions, as compared to inherited deletions (Chi sq. = 14.28; 2 d.f., P<0.001) or to de novo duplications which are not significantly increased in frequency compared to inherited duplications. On the other hand, rare duplications are significantly more often “certainly causal” than “probably” or “possibly causal” (Chi sq. 10.97; 2 d.f., P<0.01), regardless of being de novo or inherited. Furthermore, paternal transmissions of rare “certainly causal CNVs” are significantly enhanced compared to maternal transmissions of rare “certainly causal CNVs” and to paternal transmissions of less pathogenic CNVs (Chi sq. 7.58; 2 d.f., P<0.05). Finally, CNVs recurrent in ASD are present in 25/66 (37.9%) patients carrying “probably” or “certainly causal CNVs”.

Conclusions: Array CGH identified certainly or probably pathogenic CNVs in 66/275 (24.0%) ASD patients in our sample, confirming their high diagnostic yield. Significant contributions to ASD risk come from de novo deletions, from rare recurrent CNVs, and from paternally transmitted CNVs, supporting the variable penetrance of many pathogenic CNVs, especially duplications, in the ASD spectrum. Recurrent CNVs, though explaining a minority of cases, may provide more reliable opportunities to elucidate disease-relevant mechanisms.