Array-comparative genomic hybridisation (array-CGH) is a relatively new test that permits close scrutiny of chromosomal structure to detect genomic microdeletions and microduplications that are invisible in a conventional karyotype. Array-CGH is now the ‘first-line’ genetic test in the investigation of early developmental impairments and learning difficulties, especially if the clinical picture includes dysmorphism, abnormal growth, congenital anomalies, epilepsy and autism, alone or in combination. However, due to the array-CGH report's technical content and the uncertain clinical significance of many genomic findings, the results of array-CGH studies need careful interpretation. Array-CGH trebles the frequency of diagnosis compared with conventional karyotyping, but collaborative working, involving paediatricians, clinical geneticists and clinical scientists, is most important for interpretation of the results of new genomic investigations in everyday clinical practice.
Statistics from Altmetric.com
Microarray-comparative genomic hybridisation (array-CGH) has replaced conventional microscopy-based chromosome analysis as the first-line genetic investigation of children and adults with developmental disorders.1 In traditional karyotyping, blood lymphocytes yield chromosomes that are examined under a microscope. With array-CGH, sometimes called molecular karyotyping, DNA is extracted from nucleated cells, and after a series of steps, results are examined by the clinical scientist on a computer. The resolving powers of these two techniques can differ by several orders of magnitude and with the identification of hitherto invisible chromosome deletions and duplications, diagnoses can be achieved in older children and in adults who have had negative chromosome studies in the past.2 ,3 This article describes how the paediatrician should use array-CGH to diagnose genetic causes of congenital malformation, dysmorphic syndromes, disorders of growth and neurodevelopmental disorders.
The karyotype represents a very low-power examination of the 3.1 GB (3100 million base pairs) of DNA sequence that make up the human genome. The first reference human DNA sequence (genome) was obtained in 2001 and has undergone continuous revision ever since. A quite unexpected finding from genome analyses is the large amounts of variation in important DNA sequence in healthy individuals.
The most common type of genetic variation is structural copy number variation: copy number variants (CNVs) consist of segmental deletions and duplications of DNA sequence, arbitrarily >1000 bp in length. CNVs spontaneously occur on every chromosome at any location by different genetic mechanisms,4 but there are CNV ‘hotspots’ that, by virtue of the surrounding DNA sequences, are the location of recurrent deletions and duplications. Of these, certain large CNVs cause characteristic dysmorphic and malformation syndromes such as Prader–Willi syndrome or Di George syndrome, but many smaller CNVs, such as deletion or duplication of 16p11.2, predispose to less recognisable neurodevelopmental disorders such as autism, epilepsy, schizophrenia and intellectual disability. Many of these smaller CNVs are considered to be susceptibility loci because they exhibit incomplete penetrance, meaning that not everyone harbouring one of these CNVs has the associated disability or phenotype. Table A (adapted from reference 5) in the online supplementary material gives some further examples of common CNVs associated with intellectual disability, including penetrance estimates.
Healthy individuals have CNVs affecting variable lengths of protein coding and non-coding DNA sequence. When a CNV is detected by array-CGH, it is not always apparent whether the CNV is pathogenic (disease causing or associated with a developmental disorder) or if it is a harmless or benign variant. In general, small CNVs that affect fewer or no genes are less likely to have pathogenic effects than large CNVs that contain scores of genes. DNA sequence duplications tend to have less severe clinical effects compared with deletions, but there are many exceptions. For certain genes, it is essential for health to have two functioning copies. Thus, deletion of a single gene can be disease causing. When one copy of a gene is absent and the remaining copy of the gene is not sufficient to produce a normal phenotype, that gene is described as haploinsufficient. An individual with deletion of a haploinsufficient gene on one chromosome is likely to be affected by the associated developmental disorder and he or she will have a 50% chance of passing on the phenotype to each child as a dominant trait.
Our understanding of chromosomal variation and the interpretation of array-CGH results is improving with time, as information about different CNVs and their pathogenicity is being accumulated. The Database of Genomic Variants represents an ongoing effort to record all the variants present in healthy people,6 while the Database of Chromosomal Imbalance and Phenotype in Humans Using Ensembl Resources (DECIPHER) and European Cytogeneticists Association Register of Unbalanced Chromosome Aberrations (ECARUCA) databases collect CNVs that are identified in individuals with phenotypic abnormalities; box 1 provides further detail about DECIPHER.7
Aids to interpretation of array results: Database of Chromosomal Imbalance and Phenotype in Humans Using Ensembl Resources (DECIPHER)
DECIPHER is a security-assured, interactive web-based database (figure 1) designed to record chromosomal imbalances and plausibly pathogenic DNA sequence changes. DECIPHER links a variety of bioinformatics resources relevant to the imbalance found in each patient. Known and predicted genes within a copy number variant are listed, along with common copy number changes in healthy populations. Genes recognised to be clinically important are highlighted. With consent from patients, anonymised clinical data can be viewed by registered members of the DECIPHER consortium, permitting identification of patients with similar genetic changes and facilitating discussion with other clinicians.
The array-CGH process and samples required
Array-CGH requires a 2–5 mL EDTA blood sample from the patient for DNA extraction. Stored DNA or good quality DNA extracted from another tissue may also be suitable after a quality check in the laboratory.
Some laboratories also request that heparinised blood is sent with the EDTA sample as abnormal results are often validated by another method such as fluorescent in situ hybridisation (FISH). FISH is also useful to determine whether any CNV found is associated with a chromosome rearrangement such as a translocation, which confers a risk of recurrence in families.8
EDTA and heparinised samples are stable at room temperature for several days, which permits non-urgent transport to the laboratory. Details of how the test is performed are given in box 2. The clinical scientist interprets the findings in the light of clinical information from the referring doctor, so high-quality clinical data on the request form are essential.
The array comparative genomic hybridisation (CGH) testing process
An illustration of the steps involved in performing array-CGH testing in the laboratory. DNA sequences from the patient are compared with a set of purified DNA sequence fragments deposited on a solid support (the microarray platform). The sequence set is selected to represent chromosomal regions of greatest clinical interest. The patient's sample and a reference (control) DNA sample are differentially labelled (green and red, respectively). These two samples are mixed in equal proportion, their double-stranded DNA molecules are separated and the single strands are rehybridised (‘comparative’ = competitive reannealing) to the fixed DNA sequences on the microarray. The array-CGH analysis software uses algorithms to identify deletions and duplications in the patient's sample and produce an image of the red-green colour balance at each point in the array: in figure 2, excess red indicates patient DNA deletion and excess green indicates patient DNA duplication.
In replacing conventional karyotyping, the aim is to provide the requesting clinician with an array-CGH result within an equivalent timeframe. Results can and have been reported within a working week, but for non-urgent cases 4–8 week turnaround times are more typical. Urgent requests should be discussed in advance with the laboratory because in some cases an alternative technique such as conventional karyotyping, FISH or a molecular test might be the preferred option.
How to use array-CGH: clinical questions
In a child with unexplained intellectual disability and a normal karyotype by conventional testing, should array-CGH be performed?
Yes, it is appropriate to reinvestigate undiagnosed children with normal appearing 46,XX or 46,XY karyotypes since the abnormality detection rate is much higher (3× greater) after array-CGH/molecular karyotyping.1 ,3 Developmental disorders where array-CGH studies have proven superior are listed in table 1. The evidence base comprises very large case–control studies,9–11 large prospective studies from clinical service laboratories12 ,13 and small clinical case series and family studies (see online supplementary material for additional references).
Early developmental impairments with and without congenital malformations and dysmorphisms are the main indications for molecular karyotyping. Results can be summarised thus: large CNVs (>1 000 000 bp) positively correlate with early developmental impairments, congenital malformations and intellectual disability with epilepsy and/or dysmorphism. Large deletions that are shown to be de novo (a new mutation absent in parents) tend to have more severe clinical effects than duplications, but deletion of a single haploinsufficient gene can have a profound effect on phenotype.
As mentioned previously, certain recurrent CNVs predispose to or confer ‘susceptibility’ to neurodevelopmental diagnoses. It is important to note that the genetic penetrance of these conditions is incomplete, typically between 10% and 60%, meaning many carriers are unaffected.14 Patients should be counselled regarding the possible outcomes of the test; these are highlighted in box 3.
What are the potential findings that should be discussed when obtaining consent to perform array-comparative genomic hybridisation in a child with unexplained intellectual disability?
It is helpful for parents to know in advance that although a definite diagnosis might emerge, an inconclusive result is not infrequent, for example, a novel genetic variation of unknown clinical significance. A copy number variant (CNV) is likely to be classified in this way if it has been identified rarely in the past, there is little evidence from previous family studies and little is known about the function of the gene or genes involved in the CNV. In this case, parental DNA samples might be requested to help determine whether the CNV is de novo or whether it has been inherited from a similarly affected or an apparently unaffected parent. A much less common outcome is a result that reveals an unexpected health risk or gene carrier status. This might be welcome if early genetic diagnosis prevents unnecessary investigation, permits treatment or reveals a genetic risk in a future pregnancy. However, such a result might not be welcome if it reveals predisposition to an untreatable, adult-onset condition. A further discussion regarding the uncertainties that may arise is set out in relation to genomic testing by Wright et al.15
Should every child with an abnormal finding on array-CGH be referred to clinical geneticists?
When the report describes a definite pathogenic finding such as a well-known chromosomal syndrome, the paediatrician may be confident in their explanation and happy to take any additional samples that may be required to rule out a predisposing inherited chromosome rearrangement.
If a possibly pathogenic CNV is discovered in the patient, it is best to discuss this with a geneticist. Examination of parents’ DNA to determine whether the CNV has been inherited or arisen de novo is recommended, but new mutations can be benign and inherited CNVs may be pathogenic in parents (although to a lesser extent in most cases).12 Inheritance studies might not be possible if one or both parents are absent. Interpretation of the results can be more difficult if one or both parents have learning difficulties or suffer an associated mental health problem and potentially could result in a delicate situation.
Rare CNVs that are inherited from a healthy parent pose a particular challenge: thus, a moderately or severely affected child brings the family to medical attention and a mildly or unaffected carrier parent is secondarily identified by family screening. These circumstances cause doubt about the extent to which the child's more severe phenotype is attributable to the inherited CNV, and also diminish the utility of genetic screening in the extended family. Box 4 discusses how other aspects of a child's care may be impacted by array-CGH findings, while table B in the online supplementary material gives some examples of possible scenarios that may be encountered in practice.
In a child who has intellectual disability, how will performing array-comparative genomic hybridisation affect other aspects of their care?
Anecdotally, having a diagnosis is mostly a good thing,17 and may point towards health screening and early intervention. Removing guilt, clarifying issues for education and social services (including child protection issues), informing the legal process and assisting the understanding of diminished responsibility are other possible benefits.3 Better informed parents tend to be less vulnerable, for example, to harm caused by false research findings such as those implying vaccination causes autism. Finding an unambiguously causative de novo mutation underlying severe autism is reassuring when parents are fearful of having a second affected child.
However, in our experience, some families do not welcome the news that an unsuspecting parent has passed on a likely pathogenic copy number variant. Another sensitive issue we have encountered concerns young, ‘looked after’ children who have developmental impairments. In this situation, where the child's family history might have contributed to their need to be ‘looked after’, there is a significant chance of an abnormal or uncertain array finding that might prejudice their chance of an optimal permanent placement.
Unique (http://www.rarechromo.org) is an international support group for families where there are children and adults clinically affected by chromosome deletions and duplications. Unique publishes extremely useful pamphlets on chromosome tests and syndromes.
In summary, complex reports should be discussed with colleagues from Clinical Genetics, ideally before they are given to families, a recommendation that most UK geneticists would support. A brief explanation followed by the offer of an appointment at the genetic clinic is often the best option. It can be helpful for families to know what issues will be discussed at the genetic clinic.16
Generally the issues are
Inheritance studies: as mentioned, de novo CNVs are more often considered causal, whereas CNVs inherited from a healthy parent are more often considered benign variants.
Genomic characteristics of the CNV: its size, whether it is a duplication or a deletion, its gene content and whether there is any known disease association.
Clinical examination and comparisons: obtaining information about other patients from databases that hold genomic and phenotype information, for example, ECARUCA (http://www.ecaruca.net) and DECIPHER (http://decipher.sanger.ac.uk).
Genetic risks: for example, to relatives and to future pregnancies.
In a child with unexplained developmental disorder, does a normal array result rule out a causative chromosome abnormality?
No. Array-CGH will not rule out all pathogenic CNVs. Neither can it detect genetically balanced chromosome rearrangements such as translocations or inversions (very uncommon causes of developmental disorders), rare genetic changes such as uniparental disomy (eg, causing some cases of Prader–Willi syndrome) or chromosome methylation abnormalities such as those causing Beckwith–Weidemann or Silver–Russell syndrome.17
Topics for further research
Qualitative research is needed to assess the consequences of making genomic diagnoses, especially in families with limited understanding of complex genetic testing.
The environment, de novo and inherited mutations
Genetic and environmental factors almost certainly increase the risk for new and deleterious CNVs. There has been limited study of risk factors for de novo CNV production. In understanding developmental disorders, the environment, de novo and inherited mutations, as well as epigenetic factors, should all be considered.
Molecular karyotyping presages genome and exome sequencing
Next-generation DNA sequencing is a far more detailed level of genetic investigation that was discussed recently in Archives.18 This technique will soon be adapted to diagnose CNVs and may well supersede array-CGH. Box 5 gives more details regarding the Deciphering Developmental Disorders project, which is using exome sequencing technology. The UK has a plan to sequence 100 000 genomes of National Health Service patients over the next 3–5 years. This might benefit many but will produce data that will be challenging to interpret. The involvement of front-line clinicians who are skilled in integrating family history data with clinical and pathological findings is essential.
The Deciphering Developmental Disorders (DDD) project
In the UK, DDD arose out of the Database of Chromosomal Imbalance and Phenotype in Humans Using Ensembl Resources project and offers state-of-the-art molecular testing on a research basis. Recruitment is coordinated by clinical geneticists and testing is carried out at the Sanger Centre in Cambridge. Over 8000 UK families have been recruited who have a child with unexplained, abnormal development where standard genetic tests have not provided an explanation. Using whole exome sequencing of parent and child trios, new genes that cause intellectual disability have been elucidated. The results are providing an online catalogue of genetic changes linked to symptoms that enable clinicians to diagnose developmental disorders. More details can be obtained from UK Regional Clinical Genetics Departments.
Standardised reporting: dealing with incidental findings
There are differences in the quality, the interpretation and the reporting of genome data and ‘incidental’ findings among laboratories. For example, the American College of Medical Genetics and Genomics has suggested that all clinical genomes should be screened for a specific set of conditions, namely ‘actionable’ conditions, where there is prospect of preventing harm. This could include screening children for some adult-onset conditions. The European Society of Human Genetics has taken a more conservative stance and a vigorous debate has ensued. Whatever happens, it will be helpful to seek a more standardised approach to reporting genetic test results.16
Clinical bottom lines
Array-CGH studies should be considered in the investigation of any undiagnosed developmental disorder.
When obtaining consent for array-CGH investigations, tell parents that the result might not be clear-cut and very rarely it will reveal something unanticipated.
Inheritance studies are often required to help determine the significance of an array-CGH finding.
It is best to discuss all but the most straightforward abnormal results with the laboratory and/or a clinical geneticist before speaking to the parents. In general, caution should be exercised in attributing phenotypic effects to novel or rare genetic lesions. Close collaboration with genetics colleagues is advised.
Test your knowledge
Which of these changes will be picked up by microarray-comparative genomic hybridisation?
DNA copy number variation
Low-level chromosome mosaicism
All pathogenic deletions within single genes
Balanced chromosome translocation
In which of these conditions are you most likely to get a positive array-CGH result?
Severe intellectual disability
Which of these statements is false?
A positive array result assists accurate genetic counselling
A positive array result can influence medical management
A negative array result rules out chromosome imbalance as the cause of a developmental disorder
Study of parents’ DNA often assists interpretation of their child's array-CGH report
Which of these is the first-line genetic test to diagnose causation of intellectual disability?
Traditional cytogenetics with fluorescence in situ hybridisation studies
Fragile X gene test
Approximately, what percentage of children with severe intellectual disability will have an unambiguous, positive finding on array-CGH?
Answers to the questions are on page 29.
Literature review was conducted using PubMed and multiple search terms including ‘developmental disorders’, ‘microarray genomic hybridisation’, ‘Array-CGH’, ‘Copy Number Variation’.
Thanks go to Signature Genomics for allowing the use of their diagram of array-CGH testing and to Dr Norma Morrison for valued advice. We would like to dedicate this article to Dr John Tolmie, who sadly passed away during the publication process. John contributed a huge amount to the specialty of clinical genetics and was especially interested in neurodevelopmental disorders. He had a wealth of experience with these patients, and so we are glad to be able to share some of his insights in this article. He was a kind and generous person, who is greatly missed.
Answers to the multiple choice questions
(1) A. (2) B. (3) C. (4) B. (5) B.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors All co-authors contributed equally to this article.
Competing interests None.
Provenance and peer review Commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.