Background
ID (intellectual disability) and ASD (autism spectrum disorder) are life-long conditions with deficits in cognitive functioning (IQ<70) and adaptive skills that affects 1–3 % of children worldwide [
1]. Array-CGH (aCGH) offers a high diagnostic yield, ranging from 14–20 %, for individuals with unexplained ID, ASD or multiple congenital anomalies (MCA) [
2‐
5]. Available evidence suggests a change in the diagnostic approach for children with neuropsychiatric disorders and/or congenital anomalies, indicating the aCGH as the first-tier cytogenetic diagnostic test [
5]. In 2014 the SIGU (Italian Society of Human Genetics) suggested that ID, of all severity, and/or ASD and/or epilepsy, hypotonia, dysmorphisms, growth alteration, congenital malformations might be associated with pathogenic aCGH results.
The chromosomal imbalances detected by aCGH are defined copy number variations (CNVs) that are referred as: microdeletions and microduplications of clear clinical relevance or pathogenic, variants of uncertain significance (VOUS) and benign polymorphisms [
6‐
8]. The advances in molecular methodology of aCGH technology,
along with its broader application, facilitated the detection of novel pathogenic CNVs. The significance of many VOUS still remains uncertain causing serious problems in defining their contribution in patients affected by ID, MCA and ASD [
9‐
13].
The aim of this study was twofold:
1.
To determine phenotypic clues associated to pathogenic CNVs, outlining criteria for selecting patients to be studied with aCGH as first-line test.
2.
To identify cytogenetic criteria of VOUS to be taken into account for their potentially pathogenicity.
These aims allow us to depict an integrative flow-chart applying for children with ID.
Discussion
ID is a developmental disability, which presents in infancy or early childhood characterized by impaired intellectual functioning and adaptive behavior. The prevalence is estimated between 1 and 3 % of children [
5,
20‐
22]. ASDs are clinically heterogeneous disorders that include: autism, Asperger syndrome, pervasive developmental disorders not otherwise specified, and childhood disintegrative disorder. ASD ha
s been estimated to affect 1/100 to 1/150 children [
3]. Many children with ASD also have ID representing together the most frequent referral to geneticists for a diagnostic workup. American Academy of Pediatrics Committee has discussed the importance of early identification of the cause of neurocognitive phenotype identifying several benefits [
3]. Since even small chromosomal anomalies have been established as a major cause of ID and ASD, aCGH test has become an important diagnostic tool for patients at least sharing neuro-phenotypes [
2‐
5].
Despite a very high number of studies describing genetic findings of CNVs, we identifed a high rate of pathologic CNVs, more frequently deletions, in a cohort of patients with ID and/or ASD. High gene content is also found in patients with VOUS and ID.
We also demonstrated that a positive family history for ID/ASD/MCA (cardiologic, renal, intestinal anomalies) and ASD were two good independent indicator of pathologic CNVs.
On the basis of our data we suggest that aCGH should be used as first tier diagnostic test in the presence of ID/ASD/MCA. We also underline the importance of family historical recall and of parents’ clinical observation in order to evaluate the presence of ID/ASD in the parents.
Among VOUS, higher gene density was found in patients affected by ID.
A diagnosis of metabolic disorders has been reported in 1–5 % of patients with ID [
3,
19,
23,
24]. Diagnostic evaluation for inborn errors of metabolism (IEMs) was performed in 126 patients giving normal data. Notwithstanding, metabolic studies should be thoroughly considered in patients with ID due to the potential treatability of IEMs.
ACGH was performed in patients included in the retrospective part of this study because of negative results of other genetic studies, while it was applied as first tier test in the prospective part of the study because growing number of papers highlighted its diagnostic power [
25‐
27]. We identified 65 (30 %) and 25 (12 %) patients with pathogenic CNVs and VOUS respectively. The detection rate of clinically significant CNVs is about 30 % therefore higher than the yields (from 14.2–21.1 %) obtained from studies that used similar platforms [
2,
3,
5,
10,
28‐
30]. We speculate that the higher prevalence of pathological CNVs in our study is potentially ascribed to the careful selection of patients. Among patients with pathological CNVs, the deletions were more frequent than duplications (65 vs 42) being the former more commonly interpreted as pathogenic [
10]. The pathogenic CNVs have a higher size (7 Mb) and gene density (35 genes) than VOUS and benign ones in agreement with other reports [
29,
31,
32]. 29 patients with pathological CNVs have multiple rearrangements, which are known to exacerbate neuro-developmental phenotypes (Table
3) [
5,
32]. In these patients parental chromosomal study did not reveal balanced translocations.
We diagnosed 47 patients (22 %) with OMIM syndromes on the basis of either overlapping described well- microdeletion/microduplication syndromes or known causing-genes mapped within chromosomal rearrangements. ACGH detected CNVs scattered throughout the genome, but the chromosome 1, 8, 22, X resulted most frequently involved in line with other reports [
33]. The non-random involvement of specific chromosomal segments could be the results of non-allelic homologous recombinant mutational mechanism [
29]. Other recurrent pathogenic CNVs involved 1q21.1, 1q41q42, 2p15, 16p13.1, 16p11.2, 17q21.31 allowing us to characterize the phenotypes associated to chromosomal rearrangements in these specific regions [
30,
34‐
39]. A specific chromosomal abnormality does not always correspond to a specific or suggestive phenotype. In such cases, the detection of genomic aberration precedes the definition of specific phenotypes [
30]. In our dataset patients with 22q11.21 deletion and Cri-du-chat syndromes showed an atypical phenotype making the clinical diagnosis challenging [
40].
The potential limitations of aCGH application regard: delayed turnaround time, the impossibility of the detection of balanced translocations and low-level mosaicisms, the high costs, so that clinical criteria for selection of patients with higher probability of pathogenic CNV are desirable.
The selection of patients who are most likely to have a diagnosis by aCGH, minimizing the number of benign CNVs or negative results, remains an attractive goal [
41]. This study represents the largest collection of specific clinical and instrumental data for which an association with pathologic CNVs has been investigated. From previous studies, the same rate of pathologic chromosomal imbalances by aCGH was found in unselected and selected patients ([
18,
41] respectively). Other studies found the higher frequency of pathogenic CNVs in patients with congenital anomalies, unspecified dysmorphisms, growth anomalies, heart defects, primary microcephaly and familial occurrence of ID [
22,
42‐
44]. The diagnostic yield among patients with more severe ID would be expected to be higher than in patients with milder ID [
45‐
47]. In our cohort, 47 patients were enrolled because of MCA and they did not show ID. Indeed pooling data from patients with different ID degree and without ID, we conclude that more severe ID is not statistically related to pathogenic CNVs [
4,
5,
29]. Among the consistent number of clinical and history data analyzed, positive family history for ID/MCA/ASD and isolated ASD were found to be associated to pathological aCGH results. We would underline that other congenital anomalies as ocular dysmorphisms (
p=0.062), hearing loss (
p=0.127), neurological signs (
p=0.103), cutaneous dyscromia (
p=0.08) and endocrinological system involvement (
p=0.128) are potentially predictors of pathological CNVs.
In this cohort, 37 patients were affected by ASD. The overall diagnostic yield of aCGH for patients with ASD ranges from 18.2–22 % [
45,
48‐
50]. In our cohort the diagnostic yield is consistently greater (around 44 %, 16 patients out 37). Among ASD patients of this case-study, pathogenic CNVs are mostly located at chromosome 1, 4, 6, 8, 21 and 22, that partially confirm the previous results from the literature [
45,
51]. In the present study we found a low frequency of abnormal FRAX-A test results as previously described [
19].
Some hesitations in using aCGH in clinical setting diagnostic test derive from the difficulties in the efficient discrimination between benign, VOUS and pathogenic CNVs [
2,
11,
52]. CNVs can be interpreted as abnormal (pathological CNVs), VOUS and benign. We interpreted CNVs as pathogenic when contained: critical regions of microdeletion/microduplication known syndromes, genes associated with autosomal dominant inherited diseases and when cases with similar phenotypes and overlapping CNVs have already reported. CNVs are likely to be benign if they are reported in controls databases (similar CNVs in at least three healthy individuals in the same “sense”, with an overlap of more than the 50 % and the not-overlapped part less than 100 Kb), if they do not contain genes and/or known regulatory elements. Comparative analysis, with data listed in available large datasets, guide toward the clarification of CNVs clinical impact and interpretation. Multiple sources were considered as level of documentation. All the identified CNVs have been compared to those listed in: the Database of Genomic Variants (DGV,
http://projects.tcag.ca/variation) that includes healthy individuals, the pathogenic CNVs databases for patients with ID, ASD and MCA: as the International Standard Cytogenomic Array Consortium Databases (ISCA,
https://www.clinicalgenome.or), as well as the Database of Chromosomal Imbalance and Phenotype in Humans using Ensemble Resources (DECIPHER, https://decipher.sanger.ac.uk/). The Database of Genomic Structural Variation (dbVar,
http://www.ncbi.nlm.nih.gov/dbvar) including structural variation from both normal control population and disease population has been consulted as well. The genes, involved in the chromosomal region of interest, and their functions have been checked by UCSC Genome Browser (http://
http://genome-euro.ucsc.edu/cgi-bin/hgGateway) and Ensamble Genome Browser (
http://www.ensembl.org/index.html). In the interpretative process, each gene, within the CNV as well as neighboring genes, was studied for its potential role in neurological development, by all the available evidence along sources as OMIM, Genereviews, PubMed. The CNVs not associated with previously reported pathogenicity or benignity criteria were estimated as VOUS. The potential pathogenicity of VOUS is reported to be determined by many factors: the “sense” of the rearrangement (deletion or duplication, as the penetrance of duplications is considered lower than of deletions), the size (pathogenic imbalances tend to be larger than benign) and the gene content [
7]. We detected and analyzed 25 de novo VOUS and evaluated some cytogenetic indicators: overall size, gain vs loss, presence of multiple rearrangements (complex rearrangements involving several CNVs) and gene content. Only gene content had a significant correlation with ID. The gene content should be evaluated in order to speculate the pathogenicity of VOUS [
7,
10,
53].
Due to presence of VOUS, incomplete penetrance, and variable expressivity of CNVs the role of genetic counseling in aCGH testing and CNVs interpretation complements the diagnostic testing. Moreover pre-test counseling cannot be underestimated and should review potential benefits and limitations of the test.
Competing interests
The other authors have indicated they have no potential conflicts of interest to disclose. The other authors have indicated they have no financial relationships relevant to this article to disclose.
Authors’ contributions
Drs GC and DM conceptualized and designed the study, drafted the initial manuscript, and approved the final manuscript as submitted. Drs FV, AC, PF and VMG collected the data samples, carried out the initial analyses, reviewed and revised the manuscript, and approved the final manuscript as submitted. Drs RG and AM performed the genetic tests. Dr DB performed the statistical analysis. Drs LN and GA designed the data collection instruments, and coordinated and supervised data collection, critically reviewed the manuscript, and approved the final manuscript as submitted. All authors read and approved the final manucsript.