Array-Based Comparative Genomic Hybridization Analysis in Children with Developmental Delay/Intellectual Disability

Developmental delay (DD) is a condition wherein developmental milestones and learning skills do not occur at the expected age range for patients under 5 years of age. Areas used for evaluating developmental stages are gross and fine motor skills, speech and language skills, cognition, and personal-social development. Intellectual disability (ID) is characterized by limited or insufficient development of mental abilities, including intellectual functioning impairments, such as learning and cause-effect relationship [1]. Intellectual disability cases are often diagnosed in the early school-age period. The incidence of DD is 1.0–3.0% in the general population, whereas that of ID is approximately 2.7% among early school-age children [2]. Some cases have DD or ID as the only finding and are called isolated cases. Conversely, cases accompanied by facial dysmorphism, autism spectrum disorder (ASD), epilepsy and congenital anomalies, are called syndromic DD/ID [3]. Recent studies have shown that biological signaling pathways causing DD/ID, ASD, and epilepsy phenotypes are common. Additionally, the relationship between signaling pathways involved in early brain development, synaptic plasticity, and neuronal migration and the formation of these phenotypes has been demonstrated [4].

Isolated and syndromic DD/ID cases show extreme genetic heterogeneity. Genetic etiology can be detected in approximately 40.0% of the cases, whereas chromosomal abnormalities are observed in 25.0% [5,6]. Conventional cytogenetic testing can be used for detecting ≥5 Mb chromosome abnormalities. Moreover, specific chromosomal abnormalities can be investigated using fluorescence in situ hybridization (FISH) techniques. The diagnostic yield of both techniques for detecting DD/ID cases is approximately 5.0–6.0% [7]. Array-based comparative genomic hybridization (aCGH) can detect copy number variations (CNVs) in the whole genome at higher resolution than conventional cytogenetic methods. Copy number variations are defined as changes >1 kb resulting in an increase and/or decrease in the genomic DNA [8]. The CNVs are divided into two groups: recurrent and non recurrent. Recurrent CNVs often arise during meiosis from non-allelic homologous recombination (NAHR) between low copy repeat elements (LCRs). Non recurrent novel microdele-tion/microduplication syndromes have been identified in recent years owing to the widespread application of aCGH in diagnosis [9,10]. In a review published by Miller et al. [11] in 2010, the diagnostic yield of aCGH was 12.2% in 21,698 DD/ID cases retrieved from 33 different studies. The aCGH is currently recommended as the first-tier genetic test for DD/ID cases worldwide [11]. The aim of this study was to discuss the clinical findings and aCGH analysis results of isolated and syndromic DD/ID cases in the context of genotype-phenotype correlation.

The study included 139 cases (77 females, 62 males) who met the patient selection criteria. The mean age was 6.3 ± 5.1 (range 1.0–25.0). Data analysis revealed 38 different CNVs in 35 cases. The average size of CNVs was 7.01 ± 11.38 Mb (range 0.215–50.379 Mb). Of the 38 CNVs, 19 were gains and 19 were losses. Additionally, 73.6% (28/38) of all CNVs were de novo, whereas 26.4% (10/38) were inherited. The CNVs were divided into three groups according to their pathogenicity: 21 were pathogenic, 10 were VUS (five VUS, likely pathogenic; four VUS, no subclassification; one VUS, likely benign), and seven were benign. Of the pathogenic CNVs, 10 (47.6%) were gains and 11 (52.4%) were losses. Pathogenic CNVs were further grouped by size: two (9.5%) were <1.0 Mb, five (23.7%) were 1.0–3.0 Mb, seven (33.4%) were 3.0–10.0 Mb, seven (33.4%) were >10.0 Mb. In addition, 66.6% (14/21) of the pathogenic CNVs were de novo, whereas 33.4% (7/21) were inherited. Two of the inherited pathogenic CNVs were inherited from a parent with a similar phenotype (one maternal and one paternal) and one from a healthy parent (maternal). Of the four inherited pathogenic CNVs, two resulted due to healthy carrier parents of balanced reciprocal translocation and two due to inversion. In two cases, more than one pathogenic CNVs was detected.

In this study, 19 cases with pathogenic CNVs (13.6%, n = 19) and five cases with likely pathogenic CNVs (3.5%, n = 5) were found in a total of 139 cases diagnosed with DD/ID. When all pathogenic and likely pathogenic cases were evaluated, the diagnosis rate was 17.1% (n = 24/139). The diagnosis rates for pathogenic/likely pathogenic CNVs in addition to DD/ID were as follows: i) 56.0% (n = 13) in cases with facial dysmorphism findings; ii) 50.0% (n = 12) in cases with congenital heart defects; iii) 45.8% (n = 11) in cases with epilepsy; iv) 41.6% (n = 10) in cases with microcephaly; v) 50.0% (n = 12) of cases with limb anomalies.

Of the 35 cases detected with CNVs, 16 had microcephaly, 15 had epilepsy, three had ASD, 14 had facial dysmorphism, 10 had short stature, two had congenital heart defect, and 10 had structural brain anomaly. The demographic and clinical findings and detailed neurological findings of the patients are summarized in Tables 1 and 2, respectively.

The aCGH analysis revealed pathogenic CNVs showing clinical features in 19 (13.6%) of the total 139 cases. The findings of karyotype analysis were normal in 29 (n = 29/35, 82.8%) of the 35 cases with abnormalities detected by the aCGH analysis. Recurrent microdeletion/micro-duplication syndrome was detected in eight of the 35 cases, whereas 11 had rare microdeletion/microduplication syndrome. The karyotype analysis, aCGH analysis, inheritance pattern of CNVs, and pathogenicity classification of the patients are summarized in Table 1.

Array-based comparative genomic hybridization is recommended as the first-tier test in unexplained DD/ID cases as it can detect submicroscopic deletions and duplications below 5.0 Mb that cannot be detected by conventional karyotype analysis [11]. The widespread use of aCGH technology in recent years has resulted in increased diagnostic rates for DD/ID cases and identification of new microdeletion/microduplication syndromes.

The diagnosis rates vary between 5.1–35.0% in the literature [14,15]. The variability in diagnosis rates may be related to differences in criteria for patient selection, resolution of the aCGH platform used, and classification of detected CNVs. With the use of aCGH as a first-tier test in DD/ID cases, the frequency of VUS variants also increases in addition to the increase in diagnosis rates, making it difficult to demonstrate the genotype-phenotype correlation. CNVs associated with recurrent/well-defined syndromes, inherited CNVs from parents with a similar phenotype, and CNVs containing defined morbid genes in the OMIM database were identified as pathogenic, whereas polymorphic CNVs frequently seen in population databases were considered benign [16]. However, clinical interpretations of unique non recurrent CNVs are not always easy. The low number of these CNV cases in the literature, the unclear dosage sensitivity status of the genes, and the difference in penetrance, make interpretation difficult. In this study, pathogenic CNVs were detected in 19 cases according to the ACMG criteria, of which eight were cases of recurrent microdeletion/microduplication syndrome: 22q11.21 deletion (DiGeorge) syndrome in one, 5p deletion (Cri-Du-Chat) syndrome in one, 4p16.3 deletion (Wolf-Hirschhorn) syndrome in one, Xp11.23-p11.22 duplication syndrome in one, 3q26 microduplication syndrome and 3p deletion syndrome in one, 15q13.3 deletion syndrome in one, 15q13.3 duplication syndrome in one, and 15q11-q13 duplication syndrome in one. Additionally, rare pathogenic CNVs were detected in 11 cases: 2q31.1 q31.3 deletion in one, 10p15.3p15.1 deletion in one, 19p13.3 duplication in one, 14q32.2q32.33 duplication in one, 16p13.11 deletion in one, 16q12.1q23.3 duplication in one, Xp22.2p21.3 duplication in one, 6q21q22.31 deletion in one, 8p23.3p23.1 deletion and 9p24.3p23 duplication in one, 8q24.21q24.3 deletion in one, and 17q21.32q 21.33 duplication in one. All five CNVs that were considered likely pathogenic have been previously reported in at least one case with DD/ID in the literature and contain morbid genes. The CNVs detected in all four cases in the VUS, no subclassification group were previously reported as VUS in cases diagnosed with DD/ID in the DECIPHER database. Cases of frequent CNVs in the general population were grouped as benign. According to the two-hit model proposed by Girirajan et al. [17] for DD/ID cases, large CNVs that are observed more frequently in patients compared to the general population, were defined as “susceptibility loci.” In these cases, it has been reported that there may be rare single nucleotide variations (SNVs) and small CNVs, which can be detected by whole-exome sequencing (WES) and whole-genome sequencing (WGS), that are responsible for the phenotype but cannot be detected due to the resolution of microarray platform [17]. Therefore, it is believed that investigating cases of VUS CNVs with next-generation sequencing methods, such as WES and WGS, will increase the diagnostic rates.

Of the 26 pathogenic/likely pathogenic CNVs, 13 were gains and 13 were losses. Additionally, 88.4% (23/26) of these CNVs were alterations larger than 1.0 Mb. It has been reported in the literature that microdeletion syndromes are more frequently observed, and microduplication syndromes are overlooked owing to their mild phenotype [18,19]. The more frequent detection of microduplication syndromes in this study can be attributed to the inclusion of cases with a mild phenotype. The interpretation of pathogenicity of microduplications is more difficult due to incomplete penetrance and unclear triplosensitivity status of the genes it involves. Microduplications that are not found in the general population, larger than 1.0 Mb, de novo, and contain morbid genes, are more likely pathogenic [20].

Some cases of pathogenic CNVs with novel clinical and radiological findings are rarely described in the literature. Patient 1, a 4-year-old male with DD, short stature, microcephaly and scoliosis findings, was diagnosed with a de novo 1554 kb duplication in the 17q21.32q21.33 region. Two cases with duplication detected by aCGH analysis in a similar region were reported in the literature and two cases were reported in the DECIPHER database (DECIPHER ID 997 and 356717) [21,22]. Developmental delay, short stature, microcephaly, scoliosis, micrognathia, upslanting palpebral fissures were common in all the reported cases. In our case, proximally placed thumbs are a novel finding. It has been reported that COL1A1, CHAD and SGCA genes located in the duplication region may be responsible for skeletal abnormalities, whereas the PPP1R9B gene may be responsible for the DD phenotype [22]. In Patient 2, a 6-year-old male with ID and epilepsy findings, a 597 kb maternal duplication in the Xq21.31q 21.32 region, was detected. Left temporal uncal dysplasia, which was not previously reported in this syndrome, was detected in the brain MRI of the patient, who had clinical findings similar to the cases reported in the literature. It has been reported that the PCDH11X gene located in the duplication region, may be responsible for the ID phenotype [23,24]. In Patient 4, a 6-year-old male with ID in addition to ventricular septal defect (VSD) and dysmorphic features, epilepsy 9985 kb de novo deletion, was detected in the 8q24.21q24.3 region. Two cases with deletion in the same region have been reported in the literature; additionally, it has been reported that the KCNQ3 gene may be responsible for the epilepsy phenotype [25]. Duplications were detected in patients 7 and 33 at 50.3 Mb in the 4q28.2q35.1 (137, 877, 879–188, 257, 773) region and 2.9 Mb in the 4q34.2q34.3 (177, 322, 096–180, 306, 130) region, respectively. Developmental delay, microcephaly, and epilepsy phenotype of both cases are in common with the cases reported in the literature [26]. The bifid thumb in patient 7 was a novel finding for this syndrome. Although the epilepsy types of both cases were different, seizures could be controlled by anti-epileptic treatment. More dysmorphic findings were observed in patient 7 who had a larger duplication. Chromosome 15q13.3 deletion and 15q13.3 duplication syndromes were detected in patients 11 and 27, respectively. In both cases, no significant facial dysmorphic findings were found except for ID. It has been reported that the OTUD7A and CHRNA7 genes may be responsible for the phenotype in both cases [27]. Patient 24, who was previously presented in a case report, had a de novo pure partial trisomy 16q and contributed to the literature with its novel dysmorphic findings [28].

The widespread use of aCGH analysis in DD/ID cases increases the diagnostic rate. However, karyotype analysis must also be considered in each case for evaluating the cases for balanced translocations, inversions, and low-level mosaicisms that cannot be detected by the aCGH method. Determination of the location, size, and involved genes of the chromosomal abnormality using aCGH is important in terms of genotype-phenotype correlations. Additionally, a clear presentation of the chromosomal abnormality is critical for prognosis, clinical follow-up, and rehabilitation program planning. In terms of the family of the index case, it becomes possible to present prenatal diagnosis and pre-implantation genetic diagnosis options by screening other family members for chromosomal anomalies and explaining the risk of recurrence in subsequent pregnancies to the family [29].

The first limitation of this study is the small number of patients. The small sample size may have resulted in the low detection rate of frequently observed microdeletion/microduplication cases. The second limitation of the study was that FMR1 and/or MECP2 gene analyses were not done before the aCGH analysis in these cases. Some guidelines recommend the analysis of these two genes in cases of DD/ID [30]. The strengths of this study are the detailed clinical, neurological, radiological and EEG findings of the cases.

In conclusion, the use of aCGH analysis as a first-tier test in DD/ID cases contributes significantly to the diagnosis rates and enables the detection of rare microdeletion/microduplication syndromes. The clear determination of genetic etiology contributes to the literature in terms of genotype-phenotype correlation.