Congenital heart disease (CHD) is the most common birth defect and affects nearly 1% of newborns.1 2 Despite considerable therapeutic advances over the last decades, CHD still claims a considerable proportion of paediatric morbidity and mortality.3 The aetiology of CHD is largely unknown. Only a small percentage can be assigned to environmental risk factors such as maternal diseases (eg, rubella infection, diabetes, phenylketonuria) or exposure to teratogenic agents during pregnancy (eg, retinoic acid, lithium, antiepileptic drugs, alcohol).3 Chromosome aberrations detectable by conventional cytogenetics, including trisomy 13, 18 and 21, cri-du-chat syndrome or Turner syndrome, are found in about 10% of children with CHD.4

Submicroscopic chromosomal rearrangements resulting in so-called genomic disorders, such as the microdeletion 22q11.2 syndrome (OMIM 188400), Williams–Beuren syndrome (microdeletion 7q11.23; WBS; OMIM 194050) and Smith–Magenis syndrome (microdeletion 17p11.2; SMS; OMIM 182290), comprise another group of well-studied causes of CHD.

Single gene defects have also been shown to induce CHD in humans (eg, NKX2-5, GATA4, CRELD1, TBX5, JAG1, PTPN11 and TFAP2B),5 and mutations in some of these genes can result in clinically recognisable malformation syndromes such as Holt–Oram syndrome (TBX5; OMIM 142900), Noonan syndrome (PTPN11; OMIM 163950) and Alagille syndrome (JAG1; OMIM 118450).

The majority of patients with CHD are sporadic cases (ie, no other family members are affected), and the empirical recurrence risk for siblings is about 3%.6 This low rate of recurrence argues for polygenic inheritance, but might also be due to a high rate of de novo mutations.

Array comparative genome hybridisation (CGH)7 8 is a molecular cytogenetic technique dedicated to the identification of DNA copy number changes in a test genome relative to a reference genome. Recently, this method has been successfully used to verify the presence of chromosomal aberrations in patients with syndromic CHD.9^–18 We report the first comprehensive sub-megabase resolution array CGH screening of 105 patients with CHD as the sole phenotypic abnormality at the time of diagnosis and with an apparently normal karyotype.

METHODS

The study was approved by the institutional review board, and informed consent was obtained from the children or their guardians.

Patients

Children with CHD but without other major malformations or neurological problems were examined in the Department of Pediatrics, Section for Pediatric Cardiology, Rigshospitalet, Copenhagen. The patients were from a Caucasian population with a low degree of consanguinity. Except for one case with aortic stenosis (patient 1559-00), there was no family history of CHD. Patients with features suggestive of microdeletion syndromes such as WBS or microdeletion 22q11.2 syndrome were not included in the cohort. After informed consent conventional karyotyping (G-banding) and DNA preparation were carried out using peripheral blood lymphocytes. One patient with a cytogenetically visible (450-band resolution) inversion inv(8)(p23;q13) was excluded from the study group. Finally, 105 patients with various heart defects, but no other malformations at the time of diagnosis were included in the array CGH study (table 1).

Array CGH

Array CGH7 8 was performed as described previously.19 In brief, sonicated patient and reference DNA was labelled by random priming (Bioprime Array CGH, Invitrogen, Carlsbad, CA) with Cy3 and Cy5 (Amersham Biosciences, Piscataway, NJ), respectively, and hybridised onto a tiling path bacterial artificial chromosome (BAC) array, consisting of a human BAC set (32K BAC Re-Array Set; BACPAC Resources Center) (DNA kindly provided by Pieter de Jong,20^–22), a 1 Mb resolution BAC set (clones kindly provided by Nigel Carter, Wellcome Trust Sanger Centre23) and a set of subtelomeric clones (assembled by members of the COSTB19 Action: molecular cytogenetics of solid tumours). All protocols are provided in detail on our website, and details concerning this platform have been submitted to the Gene Expression Omnibus (GEO;24 GSE 7525). For the analysis and visualisation of BAC array data, our software package CGHPRO25 was used. No background subtraction was applied. Raw data were normalised by the algorithm Subgrid LOWESS. For the assessment of copy number gains and losses, we used conservative log₂ ratio thresholds of 0.3 and −0.3, respectively. Deviant signal intensity ratios involving ⩾3 neighbouring BAC clones were considered as genomic aberrations unless they were fully covered by a known DNA copy number variant, as listed in the Database of Genomic Variants and/or repeatedly identified in a reference set of >700 normal individuals and patients with other disorders analysed in our laboratory. Circular binary segmentation26 in combination with log₂ ratio threshold of 0.2 and −0.2, respectively, was used as an independent algorithm to control the automatic read-out of aberrations and to assist in the estimation of the aberration size. Verification experiments using a 244K oligonucleotide array (Agilent Technologies, Santa Clara, California, USA) were performed according the manufacturer’s recommendations and analysed by means of software provided by the same company using the default settings (Feature Extraction and CGH Analytics; Agilent Technologies).

Quantitative PCR, fluorescent in situ hybridisation and parentage testing

Primer pairs for quantitative PCR (qPCR) were selected using Primer3 software.27 Mfold28 was used to rule out primers forming secondary structures. PCR reactions were carried out in a real-time PCR system (7900HT Fast Real Time PCR System; Applied Biosystems, Foster City, California, USA) using SYBR Green (SYBR Green PCR Master Mix; Applied Biosystems) and the following conditions: 95°C for10 min, followed by 40 cycles at 95°C for 10 seconds and 58°C for 1 min, then 95°C for 10 seconds, 58°C for 15 seconds and 95°C for 15 seconds. A list of primer pairs is given in supplementary table S1 online. In addition, a change in size of the aberration between index patient and parents was excluded by array CGH analysis of the parental DNA from de novo and familial cases with sufficient amounts of DNA (1610, 1644, 1745, 1789, 2930, 3110, and 4770). Fluorescent in situ hybridisation (FISH) was carried out using 250 ng BAC DNA labelled with biotin-14-dATP by nick translation and hybridised to metaphase chromosomes using standard procedures. Parentage was tested for all cases with de novo aberrations using polymorphic short tandem repeats (STR) as previously described.29 The list of primer pairs selected from the same publication is given in supplementary table S2 online.

RESULTS

In 105 patients with CHD investigated, we detected 18 genomic aberrations that neither coincided with DNA copy number changes as listed in the Database of Genomic Variants nor could be found in our database comprising 700 array CGH analysis performed in our laboratory. These genomic aberrations included one de novo deletion, two de novo duplications and eight familial copy number variations (CNVs; one deletion and seven duplications). Inheritance could not be verified for two deletions (patients 4876 and 3551) and five duplications (patients 1688, 1939, 1952, 3488 and 4875) due to lack of parental DNA (table 2). Paternity was verified in all de novo cases. No aberration was detected in patient 1559, the only patient with a familial heart defect.

Four aberrations overlapped with already reported genomic imbalances: del(22)(q11.2) in patient 4876, dup(22)(q11.2) in patient 1610, del(17)(p11.2) in patient 1771, and del(1)(q21.1) in patient 2332. All aberrations as listed in table 3 were confirmed by FISH or quantitative PCR. De novo aberrations were further verified using a commercial 244K oligoarray (Agilent Technologies; fig 1). Array CGH data discussed in this publication have been deposited in the National Centre for Biotechnology Information Gene Expression Omnibus24 (GEO series accession no GSE 7527).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Array comparative genome hybridisation (CGH) results of cases with de novo DNA copy number changes. Top, patient 1920, chromosome 7. Middle, patient 1771 chromosome 17. Bottom, patient 2289, chromosome 13. Data analysis and visualisation was performed by CGHPRO and CGHanalytics, respectively. For each BAC clone or oligo, the Cy3/Cy5 signal intensity ratios are plotted alongside the relevant chromosomes. For the BAC arrays, red and green lines correspond to log₂ ratios of −0.3 (loss) and 0.3 (gain), respectively. Red to blue colour codes indicate BACs with increasing percentage of LCRs, as described previously.25 Besides the whole chromosome view, close-ups of the relevant regions, as well as the results of the corresponding verification experiment by means of 244K oligo array CGH are depicted. Owing to different labelling schemes, the orientation of the ratio shift observed in the graphical representation of the oligo array results does not always correlate to that for the BAC array. All chromosome coordinates are given according to HG18.

DISCUSSION

Chromosomal aberrations are well- known causes of syndromes that involve malformations of the heart. Many of the chromosomal deletions/duplications associated with such syndromes are large and encompass numerous genes. This fact not only complicates the identification of genes associated with a heart phenotype, but also conveys the impression that chromosomal aberrations mainly entail complex phenotypes, but not isolated forms of CHD. In contrast to this assumption, our data provide evidence that submicroscopic deletions and duplications play an important role in the aetiology of this condition, either as direct causes or as possible genetic risk factors for CHD.

The identified changes can roughly be grouped in three categories: de novo aberrations, inherited changes that are not known as frequent CNVs and commonly encountered CNVs. In terms of genotype–phenotype correlation and genetic counselling, the first group is the least complex. It represents 3% of the cohort and comprises one deletion and two duplications (table 2). In one patient in this group (patient 1771), and in another one with no parents available (patient 4876), array CGH was instrumental in the retrospective identification of the syndromic nature of the phenotype. In patient 1771, carrier of a del(17)(p11.2), the behavioural and neurological features of SMS only became obvious at 3 years of age. At the age of 4 months, when blood was sampled for this study, even experienced clinical geneticists would have been unable to recognise this syndrome on clinical grounds alone. For patients 1610 and 4876, both the age at the time of diagnosis and the wide clinical variability of the 22q11 microdeletion/duplication syndrome33 34 could have obscured diagnosis. The retrospective identification of genomic disorders in “non-syndromic” patients by array CGH has been reported previously,9 11 and it is not only of diagnostic relevance, but also has therapeutic implications. In SMS, diagnosis at a very young age is important for the timely onset of special education,35 and early diagnosis of 22q11.2 microdeletion/duplications is crucial for optimal management of the patients, especially in view of possible immunological problems and disturbance of calcium homeostasis.36

The second group of DNA copy number changes summarises imbalances that have been inherited from healthy parents, but are not known as frequent CNVs. These chromosomal imbalances might represent genetic risk factors that predispose to disease, although it is difficult to firmly establish the clinical relevance of such aberrations. For example, interstitial deletions of 1q21.1, as detected in patient 2332 and his phenotypically normal father, were previously reported in 3 of 481 patients with CHD.37 Also in their study, the penetrance of the phenotype was shown to be incomplete. The clinical picture of these three patients showed some variability, as one also had epilepsy, developmental delay and a dysmorphic face.37 In another study, a 1q21.1 deletion was detected in a patient with mental retardation, but no heart defects or other problems were reported.38 Interestingly, the reciprocal duplication of 1q21.1 was detected as a de novo rearrangement in a patient with mental retardation39 and as a paternally inherited duplication in a patient with tetralogy of Fallot.40 It is noteworthy that this 1q21.1 deletion/duplication is immediately adjacent to, but does not overlap a recently identified deletion predisposing to thrombocytopenia–absent radius (TAR) syndrome (OMIM 274000).10

In contrast to gross chromosomal abnormalities, the incomplete penetrance that seems to be characteristic of submicroscopic changes may depend on the genetic background—that is, on the presence or absence of modifier genes that either compensate for the defect or aggravate it. Alternatively, gene–environment interaction may play a role. The striking preponderance of duplications in our dataset suggests a greater tolerance towards gain of function. Loss of function may be more deleterious and give rise to more severe disorders, including syndromic phenotypes that are under-represented in our cohort.

In conclusion, we show here for the first time that de novo and familial sub-microscopic deletions and duplications are common, not only in syndromic but also in non-syndromic CHD. We further show that in some cases array CGH can be instrumental to reveal the syndromic nature of disease long before additional symptoms manifest. Thus our data indicate that screening patients with CHD for such genomic imbalances by high-resolution array CGH should become an integral aspect of clinical genetic diagnosis. The resulting accumulation of data will also increase our ability to discriminate disease-associated CNVs from those that are probably not relevant for the phenotype. The various microdeletions and duplications detected in the course of this study should be useful for identifying the underlying disease-causing or predisposing genes and shed more light on the molecular mechanisms that give rise to congenital abnormalities of the heart.