Comparative Genomic Advanced article

Hybridization in the Study . Introduction

Article Contents

of Human Disease . CGH Technology

. Copy Number Variation (CNV)

. Clinical Application of CGH

Sau Wai Cheung, Baylor College of Medicine, Houston, Texas, USA
. Summary
Amber Nolen Pursley, Baylor College of Medicine, Houston, Texas, USA
Online posting date: 15th November 2011

Microarray-based comparative genomic hybridization CGH Technology

(CGH) has made a significant impact on the ability to
diagnose human constitutional disease by detecting CGH is a technique that was first developed by mixing a
genomic copy number changes that were previously patient’s genome with a reference genome and co-hybrid-
undetectable by other types of cytogenetic and molecular izing to metaphase chromosome spreads for analysis
technologies. Not only can hundreds of well-character- of chromosomal imbalances (Kallioniemi et al., 1992).
ized genetic syndromes be detected in a single assay, but Metaphase chromosome spreads were later replaced by
new genomic disorders and disease-causing genes are also arrays using deoxyribonucleic acid (DNA) spots for
being discovered through the utilization of this technol- hybridization targets immobilized on glass slides. This
ogy. Clinical implementation of array CGH Hybridization technique is referred to as array CGH (aCGH) and enables
detection of copy number changes throughout the genome
has been extended to the prenatal setting, where it is also
with high resolution and provides a basis for high
proving to enhance the diagnostic capabilities in the
throughput analysis of genomic imbalances (Solinas-
perinatal period. However, the clinical interpretation of Toldo et al., 1997; Pinkel et al., 1998; Cai et al., 2002).
the increasing number of copy number variations See also: Comparative Genomic Hybridization
detected as the resolution of the microarrays is improved In aCGH, equal amounts of labelled genomic DNA
still poses a formidable challenge to laboratorians, health from a patient and reference sample are mixed and cohy-
care providers and families. bridized to an array containing the DNA targets (Figure 1a
and b). The reference sample is typically gender matched to
the patient and from an individual male or female,
although some laboratories use pooled male and female
DNA. The slides are scanned into image files using a
microarray scanner and the spot intensities are measured
Introduction (Figure 1c). The resulting ratio of the fluorescence intensities
is proportional to the ratio of the copy numbers of DNA
The research development and clinical application of sequences in the patient and reference genomes, which is
comparative genomic hybridization (CGH) in the past 10 converted to a log2 ratio and displayed for copy number
years has revolutionized the diagnostic work-up in clinical analysis (Figure 1d) (Cheung et al., 2005; Lu et al., 2007).
practice and facilitated the identification of the molecular Once a genomic imbalance has been identified, it can be
basis of many genetic diseases. Although the technology validated with other cytogenetic and molecular methods
was first applied to the field of cancer research, where it is that include metaphase or interphase fluorescence in situ
still utilized today both for research and clinical diagnosis hybridization (FISH) analyzes (Figure 1f), long-range and
and prognosis, its application to the diagnosis and dis- quantitative PCR methods, customized multiplex ligation-
covery of constitutional human genetic disease will be dependent probe amplification assays and other aCGH
discussed herein. See also: Microarrays in Disease Diag- platforms with higher resolution (Lee et al., 2007). See also:
nosis and Prognosis Fluorescence in situ Hybridization; Studying Genomic
Aberrations by Microarray Profiling
eLS subject area: Genetics & Disease The level of genomic resolution depends on the num-
ber, size and distance between the interrogating probes.
How to cite: The initial clinical arrays primarily used large genomic
Cheung, Sau Wai; and Pursley, Amber Nolen (November 2011) clones such as bacterial artificial chromosomes (BAC)
Comparative Genomic Hybridization in the Study of Human Disease. and P1 artificial chromosomes (PAC) clones (size of 75–
In: eLS. John Wiley & Sons, Ltd: Chichester.
200 kb) as interrogating probes, which were later replaced
DOI: 10.1002/9780470015902.a0005955.pub2
by oligonucleotides (60–85 mers) of which thousands to

eLS & 2011, John Wiley & Sons, Ltd. www.els.net 1

 Comparative Genomic Hybridization in the Study of Human Disease

Experimental procedure

Patient Control

Image from laser scanner

Mix

Duplication Deletion
(a) (b) Hybridization to array (c)

Array profile Fish

1.5 del 17
10

11

12

13
14
15
16
17
18
19
20
21
22
X
1

7
8
9

1.0
0.5
0.0
–0.5
10

11

12

13
14
15
16
Median 17
18
nProbes 19
20
21
22
–1.5 X
1

7
8
9

Min. start Min. stop

Mean
Value

Total
Chr Max. start Max. stop RefSeq genes (max. 50) Status
Min. size Max. size
probes
14,069,275 15,363,282 Abnormal
17:p12

CDRT15, HS3ST3B1, MGC12916

–0.84

–0.91

92

(Positive
Loss

92
14,014,324 15,492,482 CDRT7, PMP22*, TEKT3, CDRT4, nl 17
1.294 1.478
FAM18B2 finding)

(d) (e)

Figure 1 Array comparative genomic hybridization. A schematic of the array CGH process is shown at the top of the slide. The bottom left portion of the
slide shows a typical representation of the array data generated by in-house software for copy number analysis with an image of the FISH confirmation on the
right. (a) Genomic DNA from the patient is labelled with a green fluorescent dye (Cy5) and genomic DNA from a normal control is labelled with a red
fluorescent dye (Cy3). (b) The two samples are mixed and co-hybridized to the array of DNA fragments. (c) A laser scanner reads the fluorescent signals and
the intensities of each color are quantified using special software. A copy number loss is indicated by a red spot (more control and less patient DNA) and a
copy number gain is indicated by a green spot (more patient and less control DNA). (d) The graph is arranged so that chromosomal data are presented in
order from chromosome 1 on the left to chromosome Y on the right. The log2 ratio scale is shown on the Y-Axis. The centre line represents neutral copy
number, and gains and losses are plotted above and below this line, respectively. In this example, a loss in copy number on chromosome 17p12 is shown in
red. The table below shows the location, size of deletion, log ratio as well as the number of the oligos within this region. (e) FISH confirmation shows lack of
signal (red oval) for target probe on one chromosome 17, confirming the deletion (green signal is control probe, red signal is target probe) Deletions in this
region have been reported in patients with hereditary neuropathy with liability to pressure palsies (HNPP, OMIN 162500).

millions can be synthesized on one glass slide with the The most significant advantages of aCGH over tradi-
level of genomic resolution doubling every year over the tional cytogenetic methods include higher resolution and
last few years (Lu et al., 2008; Ou et al., 2008b; Boone more precise mapping of chromosomal aberrations. In
et al., 2010). Some array platforms use single-nucleotide addition, the technology is more robust, simplistic, repro-
polymorphism (SNP) allele-specific probes, which are ducible and is automated resulting in higher throughput.
generally 25–50 oligomers (Aradhya and Cherry, 2007). Since there is no need for cell culture, the turn around time
SNP arrays demonstrate a lower signal-to-noise ratio per is shorter than in cytogenetic methods. Most clinical aCGH
probe than oligonucleotide arrays, which can result in less platforms require only a few micrograms of genomic DNA,
sensitivity for detecting CNVs. However, in addition to and whole-genome amplification procedures enable fur-
the log2 ratio, SNP arrays provide data about the B ther reduction of the amount needed for analysis. Also,
allele frequency which is useful in detecting polyploidy, aCGH is superior to chromosome and FISH analysis for
uniparental disomy and identity by descent (Alkan detecting genomic duplications. As with other clinical
et al., 2011). Initially, the clinical aCGH platforms diagnostic methods, there are limitations with aCGH
were targeted microarrays designed to detect aneuploidies, technology. aCGH is not able to identify balanced
well-characterized microdeletion or microduplication syn- rearrangements such as translocations and inversions, nor
dromes and subtelomeric or other unbalanced chromo- can it provide information about where in the genome a
somal rearrangements. Commercially available clinical duplicated chromosomal segment is located. FISH is
arrays have evolved to provide whole-genome coverage as typically employed to determine if an identified copy
low as one probe per 2 kb. number gain is translocated or inserted into another

2 eLS & 2011, John Wiley & Sons, Ltd. www.els.net

 Comparative Genomic Hybridization in the Study of Human Disease

chromosome or represents a marker chromosome. Only ac.uk/PostGenomics/decipher). Careful consideration is

SNP-based arrays are able to detect uniparental disomy given to the size of the CNV, genomic position, gene(s)
and polyploidy. Even with the highest resolution arrays, involved and patient’s reported phenotype. If the impli-
small scale genomic alterations (5500 bp), such as point cated region contains gene(s) with functions compatible
mutations, cannot be detected by aCGH and the exact with the abnormal clinical findings or previously described
breakpoints of the copy number change cannot be deter- patients with a similar genomic imbalance and a similar
mined at the single base-pair level. Low level mosaicism phenotype, then the CNV should be regarded as likely
may not be detected although it has been proven that pathogenic. If after consulting these resources, the clinical
aCGH can detect chromosomal mosaicism that was not significance is still unclear, investigating the parents and
detected by cytogenetic analysis (Ballif et al., 2006; Cheung additional family members by FISH analysis or CMA
et al., 2007; Shinawi et al., 2008). Furthermore, platforms (depending on the size of the CNV) may often be necessary
that cover the entire genome at a very high resolution are to interpret and clarify the results. Although the presence of
more expensive and likely to detect genomic imbalances of the CNV in healthy family members suggests it is benign,
unclear significance. See also: Mosaicism low penetrance and variable expressivity of the phenotype
can complicate the interpretation. A larger, rare CNV that
is determined to be de novo in origin is more likely to be
Copy Number Variation (CNV) pathogenic. The de novo occurrence of a CNV is, however,
not an absolute evidence of its pathogenicity and caution
Array CGH can uncover numerous variations in DNA must be exercized for possible nonpaternity. Interpretation
copy number scattered throughout the human genome. of the clinical significance of CNVs remains challenging
Analyzing the DNA of 270 individuals from the HapMap and requires time-consuming extensive searches of the
project collection using aCGH and SNP genotyping arrays databases as well as literature and collaboration between
detected 1447 submicroscopic copy variable regions rep- the laboratory and referring clinician (Stankiewicz et al.,
resenting 12% of the genome (Redon et al., 2006). The sizes 2009). See also: Relevance of Copy Number Variation to
of these CNV regions are typically greater than 100 kb. Human Genetic Disease
These initial studies highlighted the incredible number of
CNVs observed in healthy individuals. However, the
breakpoints of these alterations were not well-defined to Clinical Application of CGH
allow an accurate assessment of the proportion of the
genome that was altered. This led to an overestimation of Detection of known genetic disorders
the extent of copy number polymorphism using large-insert
BAC clones. Recently, the same DNA samples were ana- Variations in chromosomal copy number detected by tra-
lyzed by high-density oligoarrays (24–42 M probes) which ditional chromosome analysis (karyotype) involves dupli-
enabled the discovery of CNVs as small as 500 bp (Conrad cated or deleted chromosomal material that is greater than
et al., 2010). Therefore, with increasing resolution, aCGH 3–5 Mb, although larger imbalances may be missed
platforms will be detecting more variations. depending on the chromosomal banding resolution and
Interpretation of CNVs in an affected individual is more genomic region involved. Examples of syndromes that can
of a challenge. Currently, the analysis of genomic variation be readily identified by routine cytogenetic studies include
is an extremely fast moving field due to the rapid increase in those caused by the presence of an extra chromosome;
array resolution, with the understanding of the clinical Down syndrome (trisomy 21), Edwards syndrome (trisomy
relevance of CNVs lagging behind. Initially, de novo CNVs 18), Klinefelter syndrome (XXY); or a missing chromo-
were interpreted as likely to be clinically significant whereas some; Turner syndrome (45,X); and missing or extra parts
familial CNVs are likely benign, but such an interpretation of chromosomes; Cri-du-chat (5p-), Wolf–Hirschhorn (4p-)
has become less definitive. It was also generally accepted and unbalanced chromosomal rearrangements. Chromo-
that duplications result in a milder phenotype than dele- some abnormalities are detected in approximately 1 in 200
tions. However, there are clear exceptions to these rules live newborns (Jacobs, 1977). The detection resolution for
now requiring a more systematic approach for investi- deleted and duplicated chromosomal material was
gating the clinical significance of CNVs. When a CNV is improved by the application of FISH technology, detecting
identified by aCGH in a clinical laboratory, various changes typically between 100 and 250 kb. Examples of
internal and internet-based databases are consulted to genetic disorders detectable by FISH but not usually by
determine if the CNV is polymorphic or potentially chromosome analysis include: Prader–Willi/Angelman
pathogenic. Examples of such databases are the Data- syndrome, 22q11 deletion syndrome, Williams syndrome
base of Genomic Variants (http://www.projects.tcag.ca/ and Smith–Magenis syndrome. However, there has to be a
variation/, http://www.genome.ucsc.edu), Online Men- clinical suspicion of these syndromes for a specific FISH
delian Inheritance in Man ((http://www.omim.org), Gene analysis to be utilized. FISH also improved the detection of
Tests (http://www.ncbi.nlm.nih.gov/sites/GeneTests) and subtelomeric abnormalities, which are associated with
Database of Chromosomal Imbalance and Phenotype developmental disabilities (Ravnan et al., 2006). However,
in Human using EnsemblResources (http://www.sanger. FISH is more limited in detecting microduplications than

eLS & 2011, John Wiley & Sons, Ltd. www.els.net 3

 Comparative Genomic Hybridization in the Study of Human Disease

1.5
1.0
0.5
0.0
–0.5

–1.5 del 22q11.2

22q11.2 Deletion syndrome

(a1) (a2)
Chromosome 22

1.5

1.0

0.5

0.0

–0.5

–1.5
(b1) (b2)
Microduplication 22q11.2 syndrome Fish confirmation
3 red signals indicates duplication

Figure 2 Detection of genomic disorders. Detection of 22q.11.2 microdeletion syndrome and reciprocal 22q11.2 microduplication syndrome by array
CGH with FISH confirmation. (a1) Array CGH showing a loss in copy number of chromosome band 22q11.2 involving the 22q11.2 deletion syndrome region
(red circle). (a2) FISH analysis shows lack of signal (red oval) for target probe on one chromosome 22, confirming the deletion (green signal is control probe,
red signal is target probe). Insert-G-banded chromosome analysis showing the deletion on one chromosome 22 (black arrow). (b1) Array CGH showing a
gain in copy number of chromosome band 22q11.2 involving the 22q11.2 duplication syndrome region (red circle). (b2) FISH analysis shows three signals
for the target probe, confirming the duplication (green signal is control probe, red signal is target probe). Insert-G-banded chromosome analysis showing
the duplication on one chromosome 22 (black arrow).

microdeletions. In a single assay, aCGH can simultaneously Utilizing a variety of arrays, the detection rate for copy
detect genome-wide deletions and duplications identified by number changes among patients with intellectual disability,
chromosome analysis and FISH, but with even greater dysmorphic features and multiple congenital anomalies has
resolution. In fact, some of the first new disorders detected by an average diagnostic yield of 12.2% as compared to about
aCGH were the reciprocal duplications of well-known 3% for G-banded chromosome analysis (Miller et al., 2010).
microdeletion disorders such as 22q11.2 microduplication Even greater detection rates (17.1%) have been observed
syndrome (reciprocal of 22q11.2 microdeletion syndrome, using high density arrays in select populations (Lu et al.,
Figure 2) and 7q11.23 microduplication syndrome (recip- 2008). As a result, the International Standard Cytogenomic
rocal of Williams–Beuren syndrome) (Ou et al., 2008, b; Array Consortium recommends that aCGH be the first-tier
Berg et al., 2007). See also: Chromosomal Genetic Disease: cytogenetic diagnostic test for patients with unexplained
Structural Aberrations; Chromosomal Syndromes and developmental delay/intellectual disability, autism spectrum
Genetic Disease; Chromosome Analysis and Identifica- disorders (ASDs) and multiple congenital anomalies (Miller
tion; Clinical Molecular Cytogenetics; Cytogenetic Tech- et al., 2010).
niques; Down Syndrome; Fluorescence in situ Whereas arrays were first designed to target genomic
Hybridization; Genetic Imprinting in the Prader–Willi disorders that result from submicroscopic deletions/
and Angelman Syndromes; Microdeletion Syndromes; duplications involving multiple genes, further increase in
Numerical Chromosomal Aberrations in Human Diseases; aCGH resolution has facilitated the diagnosis of single
Sex Chromosome Abnormalities; Smith-Magenis Syn- gene disorders due to haploinsufficiency resulting from
drome; Trisomy; Turner Syndrome; Williams Syndrome intragenic copy number changes, affecting even a single
Array CGH has made a significant impact in the diagnosis exon (Boone et al., 2010). Intragenic or whole gene
of individuals with mental retardation/development delay deletions are known to represent a varying portion of
and congenital anomalies without an obvious syndrome. the disease-causing mutations for conditions such as

4 eLS & 2011, John Wiley & Sons, Ltd. www.els.net

 Comparative Genomic Hybridization in the Study of Human Disease

Neurofibromatosis, type 1 (NF1, approximately 5%), Discovery of new genomic disorders

Rubinstein–Taybi syndrome (CREBBP and EP300, 10–
20%) and Pelizaeus–Merzbacher disease (PLP1, 50– The expansion to genome-wide coverage for current arrays
60%). It is estimated that microdeletions/duplications may has lead to the discovery of new microdeletion/duplication
comprise up to 15% of all mutations underlying single gene syndromes. Whereas historically, syndromes were
disorders (Vissers et al., 2005). Therefore, aCGH studies described by recognizing a certain constellation of clinical
should be considered for patients with a suspected auto- findings in multiple individuals and then searching for the
somal dominant disease and negative sequencing results. genetic alteration they all share, aCGH is describing
Not only does an exon-targeted array increase the diag- new syndromes in the reverse, that is, identifying first
nostic sensitivity for known disease-causing genes, but individuals with a common CNV and then, determining
detection of small changes can also serve as a powerful tool the clinical phenotype. However, many of these newly
for identifying new disease genes. For example, it was described syndromes lack cardinal features that would
discovered that mutations in the gene CHD7 cause allow the clinician to diagnose the condition based on
CHARGE syndrome by identifying a de novo 2.3 Mb clinical features alone. Greater variable expressivity is
overlapping microdeletion encompassing this gene in two being seen with these conditions as well as reduced
affected patients (Vissers et al., 2004). aCGH also has a role penetrance.
in the diagnosis of autosomal recessive genetic diseases as Advances in the understanding of the underlying geno-
some patients will harbor both a heterozygous copy num- mic architecture revealed that there are unstable regions
ber change too big to be detectable by sequencing in add- located throughout the genome that are prone to DNA
ition to a heterozygous single base change (Zhang et al., rearrangements, causing recurrent deletions and dupli-
2010). Furthermore, aCGH can confirm hemizygous cations. This in turn has led to the discovery of an
deletions in males and homozygous deletions where increasing number of human diseases known as genomic
sequencing assays fail to amplify a portion of or the whole disorders. These unstable regions are characterized by the
gene. See also: Identification of Disease Genes by CGH presence of low-copy repeats (LCRs) that serve as sub-
Microarrays strates for nonallelic homlogous recombination (NAHR).

Table 1 Examples of novel recurrent reciprocal deletion/duplication genomic disorders identified by array CGH
Genomic disorder Phenotype OMIM PMID references
a
1q21.1 deletion MR/developmental delay, microcephaly, 612474 16906162, 18784092, 19029900
cardiac abnormalities, cataracts
a
1q21.1 duplication Mild-to-moderate developmental delay, 612475 18784092, 19029900
macrocephaly, autism
a
15q13.3 deletion MR/developmental delay, epilepsy, mild 612001 18278044, 19289393, 19372089,
dysmorphic features 19898479, 20236110
a
15q13.3 duplication MR/developmental delay, neuropsychiatric - 19372089, 20506139
disorders, hypotonia
15q24 deletion Variable-MR/developmental delay, hypotonia, 613406 17360722, 19557438
short stature, digital anomalies, joint laxity,
genital anomalies, dysmorphic features
a
15q24 duplication MR/developmental delay, hypertonia, joint 613406 18755302, 19557438
limitations, short stature, digital anomalies,
genital anomalies, dysmorphic features
a
16p13.11 deletion MR/developmental delay, autism, epilepsy - 18550696, 19843651, 21150890
a
16p13.11 Developmental delay, behavioural - 18550696, 21150890, 21614007
duplication abnormalities (autism, ADHD)
a
17q12 deletion Renal abnormalities including renal cyst and 137920 17924346, 19844256, 21055719
diabetes syndrome, developmental delay,
autism
a
17q12 duplication Developmental delay, behavioral - 17924346, 19844256
abnormalities, epilepsy
17q21.3 deletion Variable-MR/developmental delay, hypotonia, 610443 16906162, 16906163, 16906164,
dysmorphic features, friendly behavior 18628315
17q21.3 duplication Variable-developmental delay, autism, 613533 17576104, 19502243
dysmorphic features
a
Phenotype is variable with reduced penetrance.

eLS & 2011, John Wiley & Sons, Ltd. www.els.net 5

 Comparative Genomic Hybridization in the Study of Human Disease

For NAHR to occur, 300–500 bp of perfect DNA sequence spaced at approximately 1 Mb intervals and identified
identity is the minimal efficient processing segment clinically relevant rearrangements in 8 out of 27 patients
required to mediate meiotic NAHR between intrachro- with ‘syndromic’ autism. See also: Autism; Molecular
mosomal LCRs (Reiter et al., 1998). LCRs of 410 kb in Genetics of Autism
size and with 495–97% DNA sequence identity have been The genetic cause of nonsyndromic autism or idiopathic
shown empirically to result in the most common recurrent autism presents more of a challenge since there is a lack of
NAHR-mediated interstitial or intrachromosomal geno- defining features besides the neurobehavioral phenotype
mic rearrangements (Lupski, 1998). The distance between and the majority of cases are simplex (one affected in a
LCRs is another factor apparently influencing NAHR, family). De novo copy number variants account for 5–8%
since larger-sized genomic rearrangements utilizing LCRs of simplex ASD by two independent studies using prac-
located further apart often correlate with large LCRs tically the same cohort of patients, thereby confirming the
(Stankiewicz and Lupski, 2002). See also: Chromosome- role of de novo CNVs in the etiology of idiopathic autism
specific Repeats (Low-copy Repeats); Genomic rearrange- (Levy et al., 2011; Sanders et al., 2011). Gilman et al. (2011)
ments: Mutational Mechanisms; Microdeletions and identified a large biological network of genes affected by
Microduplications: Mechanism; Segmental Duplications these rare de novo CNVs. These networks link molecules to
and Genetic Disease biological functions such as synaptogenesis, axon guidance
By searching the genome for regions with these specific and neuronal motility which led the authors to support the
characteristics, Sharp et al. (2006) predicted regions that hypothesis that autism is primarily a disease of synaptic
could lead to novel genomic disorders. Simultaneously, and neuronal connectivity malfunction (Zoghbi, 2003).
microarrays were specifically designed to interrogate these With the advances in the resolution of array technology,
regions and in fact, multiple patients were identified with hundreds of rare CNVs were identified in 996 ASD indi-
deletions and duplications of these predicted regions viduals as compared to 1287 controls, many of which
including (Table 1): 1q21.1, 15q13.3, 15q24, 16p13.11, correspond to the syndromic ASD patients indicating that
17q12 and 17q21.3 (Mefford and Eichler, 2009; Stankie- both syndromic and idiopathic ASDs have common
wicz and Lupski, 2010). etiologies (Pinto et al., 2010). In support of this, Sakai et al.
Although many of the previously described microdele- (2011) developed a protein interaction network that iden-
tion/duplication syndromes are de novo in origin, it is not tified hundreds of new interactions among proteins
unusual for one of the parents to also have the microdele- encoded by ASD-associated genes. They discovered
tion/duplication for these more newly described genomic unexpectedly high connectivity between SHANK and
disorders. Careful evaluation of the parent is needed, as TSC1, which are also previously implicated in syndromic
they may present with more subtle features than the child, autism. In addition, three of the identified network genes
underscoring the greater range of variable expressivity (MVP, ALDOV and KCTD13) are located at chromosome
observed. Also of upmost clinical importance is the much 16p11.2, the recurrent hot spot for structural variation in
higher recurrence risk associated with these inherited ASDs. A recurrent de novo 593 kb microdeletion and de
microdeletion/duplication syndromes than those previ- novo or inherited reciprocal microduplication at 16p11.2
ously well-characterized microdeletion/duplication syn- has been found in 1% of patients with autism and in 1.5%
dromes that are primarily de novo events. of patients with developmental delay (Weiss et al., 2008).

Autism Detection of uniparental disomy and

identity by descent
Array CGH technology is also being applied to determine
the genetic components of complex neuropsychiatric dis- SNP-based chromosome microarrays provide information
orders. ASDs are among the most common neu- about the distribution of allele homozygosity (HZ)
ropsychiatric disorders, with an estimated world-wide throughout the genome in addition to copy number
prevalence of 1–2.6% (Kim et al., 2011). It has been well- analysis. This information is useful for detecting uni-
established that ASDs represent a heterogeneous group of parental disomy (UPD) and consanguinity, without the
disorders that are highly heritable with heritability indices need for parental samples. Uniparental disomy occurs
estimated at 85–92% (Miles, 2011). Before the intro- when both copies of a chromosome or part of a chromo-
duction of clinical aCGH, a definitive etiology could be some are derived from only one parent and is suspected
identified in only approximately 10% of individuals with when long contiguous stretches of homozygosity (LCSH)
ASDs (Muhle et al., 2004) including chromosomal are present in a single chromosome (Papenhausen et al.,
abnormalities visible with cytogenetic methods (e.g. sex 2011). Clinically, UPD is significant when involving a
chromosome abnormalities, duplications involving chromosome or chromosomal segment containing
chromosome 15q11q12) (Vorstman et al., 2006), single imprinted genes, depending on the parent of origin.
gene disorders (e.g. fragile X syndrome, Rett syndrome, Examples of genomic imprinting disorders that can be
tuberous sclerosis or mutations in the SHANK3, NLGN3 caused by UPD include Prader–Willi (UPD15mat)/
and NLGN4X genes), or metabolic conditions. Jacque- Angelman syndrome (UPD15pat) (Figure 3a), Russell–
mont et al. (2006) applied aCGH with large insert clones Silver syndrome (UPD7mat) and Beckwith–Wiedemann

6 eLS & 2011, John Wiley & Sons, Ltd. www.els.net

 Comparative Genomic Hybridization in the Study of Human Disease

Maternal UPD 15 resulting in Angelman syndrome (isodisomy)

1.00
B Allele freq

0.50

0.00

2
Log R ratio

–2
p13 p12 p11.2 p11.1 q11.2 q12 q14 q15.1 q21.1 q21.2 q21.3 q22.2 q22.31 q23 q24.1 q25.1 q25.2 q25.3 q26.1 q26.2 q26.3
(a)

Regions of AOH containing a homozygous mutation resulting in an autosomal

recessive disease
GALT

1.00
B Allele freq

0.50

0.00
p24.1 p23 p21.3 p21.1 p11.2 p11.1 q11 q12 q21.13 q31.1 q31.3q32 q31.1 q33.2 q34.3
(b)

Figure 3 Detection of UPD and IBD by SNP array analysis. SNP array analysis showing evidence of uniparental disomy (UPD) and identity by descent (IBD).
(a) UPD for the entire chromosome 15 is indicated by absence of heterozygosity (top panel-lack of signal at 0.5 B allele frequency (BAF) which represents
genotype A/B) and no change in copy number (bottom panel-all signals are at 0 Log ratio). (b) Blocks of absence of heterozygosity (AOH) of the proximal
regions of chromosome 9p and 9q as demonstrated by lack of signals at the 0.5 BAF. Within the block of AOH at 9p (red oval) is the gene for galactosaemia,
GALT. The patient is affected with galactosaemia due to a homozygous mutation in the GALT gene. The parents are consanguineous, which is consistent with
the multiple blocks of AOH.

syndrome (UPD11p15.5pat) (Ledbetter and Engel, 1995). typically lower than postnatal peripheral blood samples.
UPD can also be associated with autosomal recessive dis- Additionally, without the need to culture cells, aCGH can
eases when the involved chromosome or chromosome offer a much more rapid turn around time, which is para-
segment contains the mutation transmitted by a hetero- mount in this particular clinical setting. This is also relevant
zygous carrier parent, resulting in a homozygous state in to cases of fetal demise where often the cells fail to grow in
the child. LCSH observed throughout the genome repre- cultures and thus, a diagnosis by conventional cytogenetics
sents identity by descent and is most pronounced in cases of cannot be provided. A limitation of employing the use of this
consanguinity (Schaaf et al., 2011). Consanguineous mat- technology in the prenatal setting is difficulty with the clinical
ings are at an increased risk for autosomal recessive dis- interpretation of variants of unknown significance. Parental
orders due to the loss of heterozygosity throughout the testing can resolve most of the CNVs detected, as usually
genome. Identifying LCSH through SNP array analysis of they are inherited and presumed benign if the parent dem-
an affected individual may assist with recognizing the dis- onstrates a normal phenotype. See also: Prenatal Diagnosis
ease-causing gene by narrowing the list of potential can- Van den Veyver et al. (2009) reported on 300 prenatal
didate genes to those located within the LCSH regions cases that were analyzed using a targeted array with a total
(Figure 3b). See also: Evolution of Imprinting: Imprinted of 19.3% of cases having a CNV. Of these, 13.3% were
Gene Function in Human Disease; Genetic Imprinting in interpreted as likely benign (primarily due to the presence of
the Prader–Willi and Angelman Syndromes; Imprinting the CNV in a healthy parent), 5% were defined abnormal-
Disorders; Single Nucleotide Polymorphism (SNP); ities whereas 1% were of uncertain clinical significance. In
Uniparental Disomy 2%, aCGH contributed important new information
whereas for two cases, the abnormality would not have been
Prenatal diagnosis detected without aCGH analysis. An example of the latter is
a case that was evaluated by aCGH due to a mosaic finding
Array studies are increasingly being used in the prenatal on routine chromosome analysis from CVS. Unexpectedly,
setting, but are usually restricted to less dense, targeted a deletion in the thrombocytopenia-absent radius (TAR)
arrays in an effort to minimize the detection of variants of region on 1q21.1 was detected and follow-up ultrasound
uncertain significance. Again, aCGH offers much higher and amniocentesis for clarification of the CVS results was
resolution than standard G-banded chromosome analysis, recommended. Ultrasound performed at the time of the
but it is even greater in the prenatal setting since the reso- amniocentesis revealed bilateral absence of the radii con-
lution of chromosome analysis for prenatal samples is sistent with a diagnosis of TAR and the de novo deletion

eLS & 2011, John Wiley & Sons, Ltd. www.els.net 7

 Comparative Genomic Hybridization in the Study of Human Disease

1.5

1.0
Uncultured amniocytes Del
0.5 1q

0.0

–0.5
Loss in copy number
1q21.1,~1.7 Mb, TAR region

–1.5

(a) (b)

(c) Ultrasound [17 weeks] (d) Ultrasound [19 weeks]

Figure 4 Prenatal diagnosis of TAR syndrome. Prenatal diagnosis of TAR syndrome by array CGH, FISH and ultrasound. (a) Array CGH showing a loss in
copy number of chromosome band 1q21.1 involving the TAR syndrome region. (b) FISH analysis shows lack of signal for the target probe on one
chromosome 1, confirming the deletion (green signal is control probe, red signal is target probe). (c) Ultrasound performed at 17 weeks showed bilateral
absence of radii with the hands attached directly to the humeri (indicated by red arrows). (d) Ultrasound performed at 19 weeks showed humeri appeared
markedly shortened and more curved but symmetrical and hands include thumbs bilaterally (indicated with red arrow).

involving the TAR region was confirmed on the amnio- rearrangement was actually imbalanced, providing more
centesis sample (Figure 4). More recent data (unpublished) accurate genetic counseling (Van den Veyver et al., 2009)
from over 1000 prenatal cases, including the previously It is the opinion of The American College of Obstet-
reported 300 cases, shows CNVs in 20.8% cases: 12.9% ricians and Gynecologists (ACOG Committee Opinion
likely benign, 7.3% abnormal and 0.6% uncertain clinical No. 446, 2009) that aCGH not currently replace classic
significance. Of the abnormal findings, 2.3% were not cytogenetics for prenatal diagnosis, but that targeted
detectable by karyotype analysis. The increased detection aCGH can be offered as an adjunct tool in prenatal cases
rate for abnormal cases is a reflection of the increased probe with abnormal anatomical findings and normal karyotype,
coverage on the arrays used for the more recent cases. This as well as in cases of fetal demise with congenital anomalies
suggests that higher resolution arrays may also have a place and the inability to obtain a conventional karyotype. Pre-
in prenatal diagnosis since the abnormal detection rate is test and post-test genetic counseling should be provided to
increased without increasing the detection of CNVs that couples elected to undergo such testing (Darilek et al.,
are likely benign or of unclear clinical significance over the 2008). The National Institute of Child Health & Human
rate seen with lower resolution targeted arrays. Development has funded a study to further evaluate the
The most common indications for prenatal aCGH efficacy of utilizing this technology routinely in prenatal
studies include advanced maternal age and abnormal diagnosis.
ultrasound findings. However, because of the increased
resolution compared to routine cytogenetic analysis,
aCGH is also frequently requested to clarify prenatal Summary
karyotype findings. Array CGH has successfully identified
the origin of extra or missing chromosomal material CGH, specifically array CGH, represents one of the most
involved in rearrangements and marker chromosomes that significant technological advancements in the diagnosis
was unidentifiable by chromosome analysis as well as and discovery of human disease. It has also revealed that
determined that an apparently balanced chromosome CNV is common in the human genome, which leads to

8 eLS & 2011, John Wiley & Sons, Ltd. www.els.net

 Comparative Genomic Hybridization in the Study of Human Disease

interpretation challenges when encountered in the clinical Kim YS, Leventhal BL, Koh YJ et al. (2011) Prevalence of autism
setting. Different aCGH platforms have been rapidly spectrum disorders in a total population sample. American
developed with increasing resolution to diagnosis conditions Journal of Psychiatry 9 [Epub ahead of print].
ranging from single gene diseases to polyploidy and are Ledbetter DH and Engel E (1995) Uniparental disomy in humans:
being utilized in both the prenatal and postnatal settings. development of an imprinting map and its implications for
prenatal diagnosis. Human Molecular Genetics 4: 1757–1764.
Lee C, Iafrate AJ, Brothman AR et al. (2007) Copy number
References variations and clinical cytogenetic diagnosis of constitutional
disorders. Nature Genetics 39: S48–S54.
ACOG Committee Opinion No. 446 (2009) Array comparative Levy D, Ronemus M, Yamrom B et al. (2011) Rare de novo and
genomic hybridization in prenatal diagnosis. Obstetrics & transmitted copy-number variation in autistic spectrum dis-
Gynecology 114: 1161–1163. orders. Neuron 70(5): 886–897.
Alkan C, Coe BP and Eichler EE (2011) Genome structural Lu X, Shaw CA, Patel A et al. (2007) Clinical implementation of
variation discovery and genotyping. Nature Reviews Genetics chromosomal microarray analysis: summary of 2513 postnatal
12(5): 363–376. cases. PLoS One 2(3): e327.
Aradhya S and Cherry AM (2007) Array-based comparative Lu XY, Phung MT, Shaw CA et al. (2008) Genomic imbalances in
genomic hybridization: clinical contexts for targeted and whole- neonates with birth defects: high detection rates by using
genome designs. Genetics in Medicine 9(9): 553–559. chromosomal microarray analysis. Pediatrics 122(6): 1310–
Ballif BC, Rorem EA, Sundin K et al. (2006) Detection of low- 1318.
level mosaicism by array CGH in routine diagnostic specimens. Lupski JR (1998) Genomic disorders: structural features of the
American Journal of Medical Genetics Part A 140A(24): genome can lead to DNA rearrangements and human disease
2757–2767. traits. Trends in Genetics 14(10): 417–422.
Berg JS, Brunetti-Pierri N, Peters SU et al. (2007) Speech delay Mefford HC and Eichler EE (2009) Duplication hotspots, rare
and autism spectrum behaviors are frequently associated with genomic disorders and common disease. Current Opinion in
duplication of the 7q11.23 Williams–Beuren syndrome region. Genetics & Development 19(3): 196–204.
Genetics in Medicine 9(7): 427–441. Miles JH (2011) Autism spectrum disorders – a genetics review.
Boone PM, Bacino CA, Shaw CA et al. (2010) Detection of clin- Genetics in Medicine 13(4): 278–294.
ically relevant exonic copy-number changes by array CGH. Miller DT, Adam MP, Aradhya S et al. (2010) Consensus state-
Human Mutation 31(12): 1326–1342. ment: chromosomal microarray is a first-tier clinical diagnostic
Cai WW, Mao JH, Chow CW et al. (2002) Genome-wide detec- test for individuals with developmental disabilities or con-
tion of chromosomal imbalances in tumors using BAC micro- genital anomalies. American Journal of Human Genetics 86(5):
arrays. Nature Biotechnology 20(4): 393–396. 749–764.
Cheung SW, Shaw CA, Scott DA et al. (2007) Microarray-based Muhle R, Trentacoste SV and Rapin I (2004) The genetics of
CGH detects chromosomal mosaicism not revealed by con- autism. Pediatrics 113(5): e472–486.
ventional cytogenetics. American Journal of Medical Genetics Ou Z, Berg JS, Yonath H et al. (2008) Microduplications of
Part A 143A(15): 1679–1686. 22q11.2 are frequently inherited and are associated with vari-
Cheung SW, Shaw CA, Yu W et al. (2005) Development and able phenotypes. Genetics in Medicine 10(4): 267–277.
validation of a CGH microarray for clinical cytogenetic diag- Ou Z, Kang SH, Shaw CA et al. (2008b) Bacterial artificial
nosis. Genetics in Medicine 7(6): 422–432. chromosome-emulation oligonucleotide arrays for targeted
Conrad DF, Pinto D, Redon R et al. (2010) Origins and functional clinical array-comparative genomic hybridization analyses.
impact of copy number variation in the human genome. Nature Genetics in Medicine 10(4): 278–289.
464: 704–712. Papenhausen P, Schwartz S, Risheg H et al. (2011) UPD detection
Darilek S, Ward P, Pursley A et al. (2008) Pre- and postnatal using homozygosity profiling with a SNP Genotyping Micro-
genetic testing by array-comparative genomic hybridization: array. American Journal of Medical Genetics Part A 155: 757–768.
genetic counseling perspectives. Genetics in Medicine 10(1): Pinkel D, Seagraves R, Sudar D et al. (1998) High resolution
13–18 [PMID: 18197052]. analysis of DNA copy number variation using comparative
Gilman SR, Iossifov I, Levy D et al. (2011) Rare de novo variants genomic hybridization to microarrays. Nature Genetics 20(2):
associated with autism implicate a large functional network of 207–211.
genes involved in formation and function of synapses. Neuron Pinto D, Pagnamenta AT, Klei L et al. (2010) Functional impact
70(5): 898–907. of global rare copy number variation in autism spectrum dis-
Jacobs PA (1977) Epidemiology of chromosomal abnormalities in orders. Nature 466: 368–372.
man. American Journal of Epidemiology 105: 180–191. Ravnan JB, Tepperberg JH, Papenhausen P et al. (2006) Sub-
Jacquemont ML, Sanlaville D, Redon R et al. (2006) Array-based telomere FISH analysis of 11 688 cases: an evaluation of the
comparative genomic Hybridization identifies high frequency frequency and pattern of subtelomere rearrangements in indi-
of cryptic chromosomal rearrangements in patients with syn- viduals with developmental disabilities. Journal of Medical
dromic autism spectrum disorders. Journal of Medical Genetics Genetics 43(6): 478–489.
43(11): 843–849. Redon R, Ishikawa S, Fitch KR et al. (2006) Global variation in
Kallioniemi A, Kallioniemi OP, Sudar D et al. (1992) Compara- copy number in the human genome. Nature 444: 444–454.
tive genomic hybridization for molecular cytogenetic analysis Reiter LT, Hastings PJ, Nelis E et al. (1998) Human meiotic
of solid tumors. Science 258: 818–821. recombination products revealed by sequencing a hotspot for

eLS & 2011, John Wiley & Sons, Ltd. www.els.net 9

 Comparative Genomic Hybridization in the Study of Human Disease

homologous strand exchange in multiple HNPP deletion Vorstman JA, Staal WG, van Daalen E et al. (2006) Identification
patients. American Journal of Human Genetics 62(5): 1023– of novel autism candidate regions through analysis of reported
1033. cytogenetic abnormalities associated with autism. Molecular
Sakai Y, Shaw CA, Dawson BC et al. (2011) Protein interactome Psychiatry 11(1): 18–28.
reveals converging molecular pathways among autism dis- Weiss LA, Shen Y, Korn JM et al. (2008) Association between
orders. Science Translational Medicine 3(86): 86ra49. microdeletion and microduplication at 16p11.2 and autism.
Sanders SJ, Ercan-Sencicek AG, Hus V et al. (2011) Multiple New England Journal of Medicine 358(7): 667–675.
recurrent de novo CNVs, including duplication of the 7q11.23 Zhang S, Li FY, Bass HN et al. (2010) Application of oligonu-
Williams syndrome region, are strongly associated with autism. cleotide array CGH to the simultaneous detection of a deletion
Neuron 70(5): 863–885. in the nuclear TK2 gene and mtDNA depletion. Molecular
Schaaf CP, Scott DA, Wiszniewska J and Beaudet AL (2011) Genetics and Metabolism 99(1): 53–57.
Identification of incestuous parental relationships by SNP- Zoghbi HY (2003) Postnatal neurodevelopmental disorders:
based DNA microarrays. Lancet 377: 555–556. meeting at the synapse? Science 302: 826–830.
Sharp AJ, Hansen S, Selzer RR et al. (2006) Discovery of previ-
ously unidentified genomic disorders from the duplication
architecture of the human genome. Nature Genetics 38(9):
1038–1042.
Shinawi M, Shao L, Jeng LJ et al. (2008) Low-level mosaicism of
trisomy 14: phenotypic and molecular characterization. Further Reading
American Journal of Medical Genetics Part A 146A(11): 1395–
1405. Betancur C (2011) Etiological heterogeneity in autism spectrum
Solinas-Toldo S, Lampel S, Stilgenbauer S et al. (1997) Matrix- disorders: more than 100 genetic and genomic disorders and still
based comparative genomic hybridization: Biochips to screen counting. Brain Research 1380: 42–77.
for genomic imbalances. Genes, Chromosomes and Cancer 20(4): Girirajan S and Eichler EE (2010) Phenotypic variability and
399–407. genetic susceptibility to genomic disorders. Human Molecular
Stankiewicz P and Lupski JR (2002) Genomic architecture, Genetics 19(R2): R176–187.
rearrangements and genomic disorders. Trends in Genetics Kearney HM, Thorland EC, Brown KK et al. (2011) American
18(2): 74–82. College of Medical Genetics standards and guidelines for
Stankiewicz P and Lupski JR (2010) Structural variation in the interpretation and reporting of postnatal constitutional copy
human genome and its role in disease. Annual Review of Medi- number variants. Genetics in Medicine 7: 680–685.
cine 61: 437–455. Lee C and Scherer SW (2010) The clinical context of copy number
Stankiewicz P, Pursley AN and Cheung SW (2009) Challenges in variation in the human genome. Expert Reviews in Molecular
clinical interpretation of microduplications detected by array Medicine 12: e8.
CGH analysis. American Journal of Medical Genetics Part A Mills RE, Walter K, Stewart C et al. (2011) Mapping copy number
152A(5): 1089–1100. variation by population-scale genome sequencing. Nature
Van den Veyver IB, Patel A, Shaw CA et al. (2009) Clinical use of 470(7332): 59–65.
array comparative genomic hybridization (aCGH) for prenatal Morrow EM (2010) Genomic copy number variation in disorders
diagnosis in 300 cases. Prenatal Diagnosis 29(1): 29–39. of cognitive development. Journal of the American Academy of
Vissers LE, van Ravenswaaij CM, Admiraal R et al. (2004) Child and Adolescent Psychiatry 49(11): 1091–1104.
Mutations in a new member of the chromodomain gene family Savage MS, Mourad MJ and Wapner RJ (2011) Evolving appli-
cause CHARGE syndrome. Nature Genetics 36(9): 955–957. cations of microarray analysis in prenatal diagnosis. Current
Vissers LE, Veltman JA, van Kessel AG and Brunner HG (2005) Opinion in Obstetrics and Gynecology 23(2): 103–108.
Identification of disease genes by whole genome CGH arrays. Slavotinek AM (2008) Novel microdeletion syndromes detected
Human Molecular Genetics 14(suppl. 2): R215–R223. by chromosome microarrays. Human Genetics 124(1): 1–17.

10 eLS & 2011, John Wiley & Sons, Ltd. www.els.net