The application of array CGH is rapidly identifying new recurrent microdeletion and microduplication syndromes 1 and a previously unsuspected level of copy number variation which needs to be distinguished from pathogenic change 2 . Among these new syndromes is the 8p23.1 duplication between the 8p23.1 olfactory receptor/defensin repeats (ORDRs) at REPD in distal 8p23.1 (REPeat Distal) and REPP (REPeat Proximal) in proximal 8p23.1 (Table 1). This genomic disorder is the reciprocal of the 8p23.1 deletion syndrome 3 and has, to our knowledge, only been confirmed at the molecular level in four families to date 4 5 . Duplications of 8p23.1 have been associated with a variable phenotype that may include one or more of developmental delay, mild dysmorphism and heart defects. The single prenatal case had only mild dysmorphism and normal development at the age of 15 months with no evidence of a heart defect (Case 1 5 ).

The REPD and REPP repeats mediate a remarkable variety of simple and complex chromosome rearrangements 6 and are themselves copy number variable with 2 to 7 copies of the beta defensin components in the normal population 7 . Numbers as high as 9 to 12 become cytogenetically visible as "euchromatic variants" 7 that are only associated with a predisposition to psoriasis 8 . These high level copy number variations are cytogenetically indistinguishable from the 8p23.1 duplications 4 and both the copy number variants and genuine duplications can be transmitted from parents to children. Here we report on our experience of using follow-up MLPA and FISH testing of apparent cytogenetic duplications of 8p23.1 detected during prenatal diagnosis. The four cases include two de novo duplications, a paternally transmitted duplication of 8p23.1 and a benign maternally transmitted defensin copy number variation.

Amniotic fluid cells were cultured, G-banded and analysed using established techniques. Quantitative Fluorescent Polymerase Chain Reaction (QF-PCR) analysis was performed using an autosomal multiplex according to a method adapted from Mann et al 9 (Case 1). DNA was extracted using a Qiagen EZ1 machine and MLPA 10 was carried out according to the manufacturer's instructions with the P139 defensin kit which contains 29 probes mapping across the distal short arm of chromosome 8 (Cases 1 and 4) (Table 1) (please see the MRC-Holland web site for further information). MLPA PCR products were separated on an ABI 3100 Sequencer, analysed using Applied Biosystems Inc Genotyper version 2.0 (Table 1) and the results collated in an in-house Excel spreadsheet as previously described 11 . Array CGH was carried out with the BlueGnome Cytochip Focus BAC array and BlueFuse software according to the manufacturer's instructions with minor modifications 12 (Case 1). FISH was carried out using standard methods with Ensembl 37 k cloneset bacterial artificial chromosomes (BACs) (Table 1) chosen from the Ensembl web browser http://www.ensembl.org/Homo_sapiens/Location/Genome (Cases 2 and 3). The BACs were grown, validated and prepared for FISH by the National Genetics Reference Laboratory (Wessex). Additional BAC FISH was performed in Case 2 as previously reported 13 .

We have presented four prenatal cases in which an 8p23.1 duplication was suspected on cytogenetic grounds. MLPA or FISH confirmed 8p23.1 duplication syndrome in Cases 1 to 3 and only copy number variation of the defensin cluster in Case 4. Cases 1 and 2 were de novo, the duplication in Case 3 was directly transmitted from the father and the copy number variation in Case 4 was maternally transmitted. It is reasonable to conclude that Cases 1 to 3 all had a core duplication of ~3.75 Mb between the proximal and distal ORDRs (REPD and REPP) as shown using array CGH in Case 1 (Figure 2A, Figure 3). When these cases are added to those in the literature, the 8p23.1 duplication has now been confirmed, using molecular cytogenetic methods, in eleven individuals of whom four cases were de novo and another four had duplications transmitted from a father and two mothers (Table 2). An estimate of the prevalence of this condition can be derived from a recent series of 2,419 diagnostic patients analysed using oligonucleotide array CGH 15 ; one dup(8)(p23.1p23.1) was found compared with sixteen 22q11.2 DiGeorge/Velocardiofacial syndrome (DG/VCFS) deletions. As DG/VCFS has a population frequency of ~1 in 4,000 16 , the 8p23.1 duplication syndrome has an estimated prevalence of 1 in 64,000. No examples of the full 8p23.1 duplication syndrome region have been reported among the 29,133 CNVs currently in the Database of Genome Variants (DGV) http://projects.tcag.ca/variation/ (Figure 3).

A summary of the phenotypic data on the eleven patients with 8p23.1 duplication syndrome is provided in Table 2. Ascertainment has been as a result of congenital heart disease (CHD) in only one of four prenatal cases and one of three postnatal probands but is now the most common single feature having been found in 6/11 individuals. Developmental delay and/or learning difficulties have been found in 5/11 but, of the remaining 6/11, one prenatal case was developmentally normal at 15 months of age and the three prenatal cases reported here have not yet reached an age at which this can assessed. A variable degree of facial dysmorphism was also present in 5/11 individuals. These results are broadly in line with those of Tsai et al 17 but, unfortunately, their results rely on cytogenetics alone and do not differentiate between the duplications and copy number variants. By contrast, partial toe syndactyly has been found in only one mother and son and adrenal anomalies in two probands but not in the mother, of one of these two, who had the same duplication.

Excluding the copy number variable regions, REPP and REPD, the duplicated interval contains 57 genes of which 34 are known and 23 are novel. These include the two transcription factors GATA4 and SOX7 and three micro-RNA loci. Deletions and heterozygous loss of function mutations of the GATA binding protein 4 gene (GATA4, OMIM *600576) are already strongly associated with conotruncal and septal heart defects 3 18 19 20 21 22 23 and it has been proposed that duplication of GATA4 is responsible for the pulmonary atresia and Tetralogy of Fallot found in two of the four published probands with 8p23.1 duplication syndrome 4 5 . The idea that GATA4 is responsible for the heart defect component of the 8p23.1 duplication syndrome is strengthened by the hypoplastic left heart in Case 1, the complex VSD in Case 2, the meso-septal interventricular heart defect in Case 3, at autopsy, and the mild heart defect in the eldest sister of Case 3.

The existence of a second heart disease gene in a 5-cM region of 8p23.1 between WI-8327 and D8S1825 (6,469,539 to 8,962,119 base pairs according to UCSC, March 2006) was proposed by Giglio et al 24 (Figure 3). However, the overlap between this ~2.5 Mb region and the REPD to REPP interval contains only four single copy genes (Figure 3), of which neither PRAGMIN, CLDN23 (*OMIM 609203), MFHAS1 (OMIM *605352) nor ERI1 (OMIM *608739) are currently good candidates for heart disease. Thus, it seems more likely that the absence of heart disease in some 8p23.1 duplication syndrome probands 5 , four members of a family with a 133 kb microduplication of the GATA4 gene 25 and seven individuals with a 4.37 Mb duplication of 8p23.1 to 8p22 that included GATA4 26 , is more likely to reflect non-penetrance rather than the existence of a further heart disease gene between REPP and REPD. Normal heart development is thought to require interaction between GATA4 and the T-Box 5 (TBX5) gene 19 27 which suggests that variation in TBX5, or other genes involved in the development of the heart, might modify the consequences of altered GATA4 dosage. We conclude that both duplication and deletion of the GATA4 gene can give rise to a variety of conotruncal and septal heart defects but with variable penetrance and expressivity.

Other candidate genes derived from atypical microdeletions of proximal 8p23.1 may be considered candidate genes for features of the 8p23.1 duplication syndrome 3 28 (Figure 3). These include the TRF1-interacting, ankyrin related ADP-ribose polymerase gene (TNKS, OMIM*603303) for behavioural difficulties, and the SRY-Box 7 transcription factor (SOX7, OMIM *612202) for the developmental delay, as mutations of the related SOX3 gene have been associated with X-linked mental retardation 28 . By contrast, the diaphragmatic hernia found in a number of patients with the reciprocal deletion syndrome 29 30 31 has not, to our knowledge, been recorded in any 8p23.1 duplication syndrome patient to date.

There was evidence, using FISH, for both REPD and REPP being included in the duplication in Case 2 (Figure 2E) and Case 3 (Figure 2F and 2H) but not for the inclusion of either repeat in Case 1 using MLPA. REPP was also implicated in a previously published case (Family 1 5 ). Altered copy number might be expected at either or both repeats if the reciprocal deletions and duplications are generated by ectopic recombination (or NAHR) between the repeats 6 . Alternatively, altered copy number may be due to independent copy number variation of the ORDRs themselves.

Benign copy number variation of the defensin cluster was found in over 14% of a recent series of 1,275 patients analysed using array CGH 32 , but cytogenetically visible amplifications, of the kind found in Case 4, are uncommon. These segregate in families with no significant clinical or reproductive effects other than a predisposition to Crohn's disease at low copy number 33 and psoriasis at high copy number 8 . Most chromosomes 8 have two copies of the defensin cluster and most individuals a total of four 34 . Thus, the triplication of the defensin cluster, relative to control DNA, implies a total of 12 copies in the Case 4 fetus and her mother. If the normal chromosome 8 had 2 copies, the variant chromosome would have 10 copies of the ORDR repeat and, as the repeat is a minimum of 240 kb in size 7 , the ORDR array would extend to at least 2.4 Mb and thus become cytogenetically visible in the light microscope (Figure 1F) 4 7 .

Both FISH and MLPA have been reliably used to confirm or exclude an 8p23.1 duplication between REPD and REPP. However, even normal chromosomes 8 can look duplicated with BACs that map to these repeats and thus differential signal strength between FISH probes does not constitute proof that copy number variation of REPD or REPP is the cause of an increase in the size of the 8p23.1 band. Recent evidence also suggests that the defensin clusters are switched between REPD to REPP by the polymorphic inversion between them 35 , and this may be expected to change the appearance of the FISH signals seen on homologous pairs of chromosome 8 even if their copy number is the same or similar. Array CGH will also discriminate the duplication from the variant, and exclude additional imbalances, but careful choice of control samples may be required to accurately confirm the extent of the defensin copy number variation in all cases.

In conclusion, our results underline the need to distinguish the 8p23.1 duplication from benign defensin copy number variation at prenatal diagnosis. Direct transmission of duplications and copy number variants from a parent to a child has been found on multiple occasions and transmission does not therefore discriminate between copy number variations of the defensin cluster and the 8p23.1 duplication syndrome. Cardiac defects were ascertained by ultrasound in only one of the three duplication 8p23.1 pregnancies but were visible in two of the three at 21 to 22 weeks gestation. Phenotypic data also indicate a relatively mild but variable syndrome and support the idea that duplication of the GATA4 transcription factor can give rise to a variety of conotruncal or septal heart defects with variable penetrance and expressivity.

AMA: Advanced maternal age; BAC: Bacterial Artificial Chromosome; array CGH: array Comparative Genomic Hybridisation; CHD: Congenital heart disease; DGV: Database of Genome Variants; DNA: De-oxyo Ribose Nucleic Acid; FISH: Fluorescence In Situ Hybridisation HLH: Hypoplastic Left Heart; MCA: Multiple Congenital Anomalies; Multiple Ligation-dependent Probe Amplification (MLPA); OFC: Occipito-Frontal Circumference; PDA: Patent Ductus Arteriosus; ORDR: Olfactory Receptor and Defensin Repeat; Patent Ductus Arteriosus; PNAI: Primary Neonatal Adrenal Insufficiency; QF-PCR: Quantitative Fluorescent Polymerase Chain Reaction; REPD: REPeat Distal; REPP: REPeat Proximal; VSD: Ventricular Septal Defect.

The authors declare that they have no competing interests.

JCKB assembled the results and drafted the manuscript; DB carried out the MLPA analyses on Cases 1 and 4; MC and DR provided the laboratory and clinical information on Case 1; SM and AD provided the laboratory and clinical information on Case 2 and TL contributed the additional FISH results; CA, JT and AIP provided the laboratory and clinical information on Case 3 and her family; CC and KO provided the laboratory and clinical information on Case 4; EJT drafted Table 2; VKM prepared and validated the FISH probes and carried out the FISH analyses on Cases 2 and 3. All authors read and approved the final manuscript.