Recurrent benign copy number variants & issues in interpretation of variants of unknown significance identified by cytogenetic microarray in Indian patients with intellectual disability

Cytogenetic/cytogenomic/chromosomal microarray (CMA) has been recommended as a first-tier diagnostic test in the work-up of patients with intellectual disability (ID)/developmental delay (DD)/multiple congenital anomalies (MCA) and/or autistic spectrum disorders (ASDs)1. The diagnostic yield is estimated to be in the range of 15-20 per cent in cases with idiopathic ID/DD2. Along with causal pathogenic copy number variants (CNVs), CMA platforms also identify many other CNVs which are difficult to be categorized in benign or pathogenic variants. These variants are called as variants of unknown significance (VUS)234. These pose great dilemma in front of cytogeneticists as well as to clinicians in providing genetic counselling, prediction of risk of recurrence and providing prenatal diagnosis. In this study we describe various issues in interpretation of CNVs identified in CMA analysis in Indian patients with idiopathic ID/DD and report normal variants in Indian patients.

Material & Methods

This study was conducted in the department of Medical Genetics, Sanjay Gandhi Post Graduate Institute of Medical Sciences (SGPGIMS), Lucknow, India, from May 2012 to April 2013. All those patients with idiopathic ID/DD with or without malformation or dysmorphic features were included whose relevant clinical details were available and the family agreed to participate in the study and consented to provide the sample. Cytogenetic analysis by G banded karyotype at 450-550 band level was normal in all patients. CMA was performed in 86 cases with ID/DD with or without dysmorphic features in whom clinical examination and appropriate investigations had not provided aetiological diagnosis. CMA was performed in parents wherever consent of the parents and their blood samples were available. The present study protocol was approved by the institute ethical committee of SGPGI, Lucknow.

CMA analysis: CMA was performed by the Cytogenetics 2.7M Array (Affymertix ®, USA, 71 cases) and HumanCytoSNP-12 (Illumina, USA, 15 cases). Analysis was done by Affymetrix® Chromosomal Analysis Suite and Genome studio software (Illumina) as per manufacturers’ protocol. Cytogenetics 2.7M Array has density of 2.7 million markers covering the whole genome. It also includes 400,000 probes to detect single nucleotide polymorphisms (SNPs) to enable the detection of copy neutral changes (loss of heterozygosity, LOH). Illumina HumanCytoSNP 12 has 200,000 probes for SNP, providing genome coverage and 220,000 cytogenetic markers for 250 targeted genomic regions. Human genome version GRCh 37:Feb 2009 (hg 19) (http://genome.ucsc.edu/cgi-bin/hgGateway?db=hg19) was used in data annotation.

Copy number variants (CNVs): CNVs were reported only if copy number gain was >400 kb in size and copy number loss was more than >200 kb in size. CNVs were classified into benign/non-pathogenic, pathogenic/clinically relevant variants (which are associated with known microdeletion/microduplication syndrome and/or associated with clinical phenotype or large de novo variants with genes associated with phenotypes like autism, epilepsy, intellectual disability or other significant neurological dysfunction) and VUS (genomic variants which have not been previously reported in normal individuals and insufficient information regarding clinical significance)4. This delineation was made after looking into published literature and curetted databases5. The size of CNV, its gene content and its de novo or inherited status were also taken into consideration. VUS were further divided into possibly benign [inherited from either clinically normal parent and/or not reported in Database of Genomic Variants (DGV)6, no relevant Online Mendelian Inheritance in Men (OMIM) phenotype7, no relevant genes or a particular CNV was present in multiple patients in recurrent manner], possibly pathogenic (if it was de novo or OMIM loci associated with DD/ID/ASDs/other central nervous system disorders like ataxia and epilepsy) and possibly VUS (no definite central nervous system associated genes or phenotype and/or one or more genes associated with basic cell function, i.e. embryogenesis, cell migration) according to available evidence of published literature and databases34. Patients harbouring at least two large CNVs (>5 Mb) were designated to have double segment imbalances. Subtelomeric copy number gains or losses were further validated by multiplex ligation dependent probe amplification (MLPA) test8.

Results

A total of 86 patients with idiopathic DD/ID with or without malformation/dysmorphism were included in the study. Of these, nine (10.5%) were less than one year of age, 43 (50%) were between age 1 and 5 yr while 34 (39.5%) were more than 5 yr of age. Forty one (47.6%) patients were males while 45 (52.3%) were females.

Pathogenic CNVs: Pathogenic variants were found in 18 patients giving a yield of 20.9 per cent. Of these, 14 patients (13 deletions, 1 duplication) had variants which were already associated with known microdeletion/microduplication syndromes. The details of these patients are presented in Table I. Three of these 18 patients had double segment imbalances indicating the possibility of inherited/de novo chromosomal rearrangement. Of these three, one family (in extended pedigree) had three children affected with global developmental delay with facial dysmorphism suggesting a familial balanced chromosomal translocation. Details of cases with double segment imbalances are presented in Table II. One patient had de novo heterozygous copy number gain of 14 Mb size. This patient was a 22 yr old male born in non-consanguineous family with no significant family history. The clinical features included short stature, facial dysmorphism (maxillary hypoplasia) and brachydactyly (Figure). The patient was talkative and had friendly personality. This region was harbouring >75 genes [arr10q21.1q22.1(59168091-73319571)X3]. No gene was definitely associated with mental retardation/developmental disability or other related disorders (UCSC genome browser hg19 version http://genome.ucsc.edu/cgi-bin/hgGateway?db=hg19). Important genes in this region include NEUROG3 (transcription factor involved in neurogenesis) and TFAM (polymorphism has been reported in Alzheimer's disease and parkinsonism). Other genes were involved in various basic cellular functions including contact, motility, mRNA transport and metabolism. In DECIPHER (https://decipher.sanger.ac.uk) a few entries have been described in overlapping region associated with mental retardation. On the basis of large size and de novo nature, this CNV was interpreted as pathogenic.

Table I Pathogenic variants seen in patients with idiopathic DD/ID (n=14) with known pathogenic gains/losses

Table II Double segment imbalances in three patients with global developmental delay

Close

Figure

Photograph of patient, having de novo heterozygous 14 Mb gain on 10q21.1-22.1. Facial dysmorphism included maxillary hypoplasia and downslanting palpebral fissures. Hands showing brachydactyly.

Export to PPT

VUS: Twenty five (29%) patients did not have any CNV detected by CMA. On the other hand, in 26 (30.2%) patients all CNVs (total 41 CNVs, 13 losses, 28 gains) detected were interpreted as benign. Size of these benign CNVs was ranging in size from 226 kb to 3.3 Mb. Seventeen of 68 (25%) patients had one or more VUS (total 31) giving and average of 1.8 VUS per case.VUS, which were present in patients harbouring definitely pathogenic variants, were not included in this list. Almost half (9/17) of the patients were having multiple VUS. Maximum number of VUS in a single patient was four. Four out of 31 VUS (7.7%) were interpreted as possibly benign (2 gains and 2 losses, size range 233-1115 kb, Table III). Eleven CNVs (35.2% of all VUS), seen in 10 patients were interpreted as possibly VUS (all gains, size range is 422-2399 kb, Table IV). Sixteen CNVs (51.6% of all VUS) in 10 patients (1-2 per case) were interpreted as VUS, possibly pathogenic (6 losses, 10 gains, size range 206- 2284 kb, Table V).

Table III Possibly benign variants of unknown significance

Table IV Variants of unknown significance (Possibly VUS).

Table V Possibly pathogenic variants of unknown significance.

Of the 15 patients with single definite pathogenic variant, nine were also having possibly pathogenic VUS or possibly VUS at unrelated parts of genomes. Three patients had single VUS. Rest of them were harbouring 2-5 VUS. One of the three patients with double segment imbalances had VUS at different chromosomal region (1.7 Mb loss at 10q21.1) apart from two primary gains/losses.

Recurrent benign CNVs: Five CNVs including 4 gains and 1 loss (size range 301-927 Kb, Table VI) were present as recurrent benign CNVs in our patients. The size of each CNV was much larger than those variants which were reported in DGV (hg19) (Database of Genomic Variant; http:projects.tcag.ca/variation/).

Table VI Recurrent benign copy number variants

LOH regions: We analyzed LOH regions in 36 patients in whom CMA was performed by Affymetrix2.7 M array and no definite pathogenic variant was identified. Laboratory cut-off for analysing these LOH regions was kept as 5Mb and X chromosome was not included in the analysis. This 5Mb cut-off was decided on the basis of study done by Sund et al9. Of these 36 patients, two were born by consanguineous parentage and in another patient there was history of similarly affected sibling but there was no consanguinity. In consanguineous (between first cousins) families, the number of LOH regions (>5Mb size) was 3 and 12, respectively. Total region of homozygosity was 91 and 235 Mb, respectively (3.1 and 8.1% of total autosomes). In 34 non-consanguineous families, 27 (84%) had no significant LOH regions. Three patients had single LOH region (5-6Mb) on an autosome. In four families (11.7% among non-consanguineous families), 2-24 LOH regions (32-188 Mb) were found, which were corresponding to 1.1 - 6.5 per cent of total autosomes.

Discussion

The diagnostic yield of CMA in our patients with idiopathic ID/DD was 20.9 per cent which was in accordance with other studies showing the diagnostic contribution of CMA in the range of 15-20 per cent1011. Of the 18 pathogenic variants, five were located in subtelomeric region. These subtelomeric gains/losses can be identified by MLPA using probe set for subtelomeric regions. Also MLPA can be used to diagnose cases with known microdeletion and microduplication syndromes. At present MLPA probe set for common microdeletion contains probes for 21 regions. In a study done at our centre the diagnostic yield of MLPA using subtelomeric and common microdeletion probe set in patients with idiopathic developmental delay was 9.3 per cent8. MLPA can be acceptable substitute to CMA in those families who can not afford CMA.

We also found one novel pathogenic copy number gain of 14 Mb size in one patient with DD and facial dysmorphism. Though not described in literature, various genes in this region are involved in basic cellular metabolism including neurogenesis. There were three patients with double segment imbalances. In these patients, possibilities can be interchromosomal exchange of segments representing the possibility of chromosomal imbalance or separate chromosomal events12. The risk of recurrence in the former case will be up to 50 per cent if inherited in comparison to <1 per cent in the later events as most of these pathogenic variants are de novo in origin. In all these cases karyotype of patients/parents or fluorescent in-situ hybridization analysis will be essential for accurate risk prediction of recurrence in family.

Interestingly, 60 per cent patients who were having at least one definite pathogenic variant were also having clinically important CNVs at other genomic location. These VUS in patients may contribute towards modulation of clinical features leading to phenotypic differences of the patients. In a study conducted by Girirajan et al13, in 32,587 children with developmental delay, prevalence of second additional genetic variant was 10 per cent. They have hypothesized that these CNVs may be responsible for phenotypic variations in microdeletion/microduplication syndromes.

In this study, we found 31 VUS in 17 patients with no definitely pathogenic variants. Pyatt et al10 in their study on 1998 samples found 563 abnormalities in 490 patients. The size range of these VUS was 33 kb to 2.9 Mb. Similar to this study, frequency of duplication variants were much more than deletion (66 vs 33% in our study and 63 vs 36% in their study). In the present study, of the 31 VUS, 27 CNVs had to be interpreted as either possibly pathogenic VUS or possibly VUS. The various reasons for these VUS can be different CMA platforms, unavailability of stringent guidelines for interpretation, wide variation in phenotype of a particular CNV, rapidly expanding databases of benign as well as pathogenic variants, genes of unknown function, non availability of family members for genetic testing and reduced penetrance of various pathogenic CNVs310.

We reported five benign recurrent CNVs in our patients. The presence of these variants indicates towards the possibility of ethnic variation of benign variants. Also, there is some evidence that certain variants may predispose a particular population to abnormal phenotype and provide protection to other population1415.

The limitation of our study was small number of patients. Also parental CMA analysis could not be done in many cases with VUS, mainly because of unavailability of parents’ samples. Initially de novo variants were thought to be more significant in terms of its pathogenicity and inherited benign variants were considered to be more benign. According to recent published literature13, penetrance of such variants can range from 10-60 per cent. Girirajan et al14 proposed two hit model for variability of phenotype in recurrent CNVs or for those inherited from either parent. We found 91-235 Mb regions of homozygosity in consanguineous families and 32-188 Mb region of homozygosity in 11.7 per cent of non-consanguineous families. Percentage of shared genome and patients with LOH regions were more than published literature. This may be due to inbreeding over many generations as there is custom of marrying amongst specific caste group. In a previous study, the detection rate of LOH regions was present in 4.2 per cent patients14. In that study, discrepancies between clinical documentation of parental consanguinity/illegal parental relationship were raised14. However, being at a clinical genetics centre we ourselves have taken detailed family history. Hence there is definite documentation of consanguinity.

In conclusion, this study of CMA from Indian patients with ID/DD with diagnostic yield of 20.9 per cent highlights the difficulty in interpretation of CNVs identified by CMA. Our study also highlights the importance of MLPA as an acceptable substitute of CMA for those families who cannot afford CMA due to cost constraints. There is a need for more Indian data about recurrent benign CNV in the population, as it will further help us in categorization of CNVs into benign vs VUS.