De novo copy number variants are associated with congenital diaphragmatic hernia

Abstract

Background

Congenital diaphragmatic hernia (CDH) is a common birth defect with significant morbidity and mortality. Although the etiology of CDH remains poorly understood, studies from animal models and patients with CDH suggest that genetic factors play an important role in the development of CDH. Chromosomal anomalies have been reported in CDH.

Methods

In this study, the authors investigate the frequency of chromosomal anomalies and copy number variants in 256 parent-child trios of CDH using clinical conventional cytogenetic and microarray analysis. The authors also selected a set of CDH related training genes to prioritize the genes in those segmental aneuploidies and identified the genes and gene sets that may contribute to the etiology of CDH.

Results

The authors identified chromosomal anomalies in 16 patients (6.3 %) of the series including 3 aneuploidies, 2 unbalanced translocation, and 11 patients with de novo CNVs ranging in size from 95 kb to 104.6 Mb. The authors prioritized the genes in the CNV segments and identified KCNA2, LMNA, CACNA1S, MYOG, HLX, LBR, AGT, GATA4, SOX7, HYLS1, FOXC1, FOXF2, PDGFA, FGF6, COL4A1, COL4A2, HOMER2, BNC1, BID, and TBX1 as genes that may be involved in diaphragm development. Gene enrichment analysis identified the most relevant gene ontology (GO) categories as those involved in tissue development (p=4.4×10⁻¹¹) or regulation of multicellular organismal processes (p=2.8×10⁻¹⁰) and “receptor binding” (p = 8.7×10⁻¹⁴) and “DNA binding transcription factor activity” (p= 4.4×10⁻¹⁰).

Conclusions

Our findings support the role of chromosomal anomalies in CDH and provide a set of candidate genes including FOXC1, FOXF2, PDGFA, FGF6, COL4A1, COL4A2, SOX7,BNC1, BID, and TBX1 for further analysis in CDH.

Introduction

Congenital diaphragmatic hernia (CDH) is a failure of complete diaphragm formation which allows abdominal organs to herniate into the thoracic cavity and compromises lung development. It is a common birth defect with an incidence of approximately one in 3000 live births.[1] Newborns with CDH often have severe life-threatening pulmonary hypoplasia and pulmonary hypertension. CDH mortality varies between centers (20% to 40%), and long term morbidity remains relatively high despite advances in therapy for treatment of the pulmonary disease.[2, 3] Approximately 40% of cases with CDH are associated with an additional anomaly, most commonly congenital heart, brain, renal, and genitourinary malformations.[4] The etiology of CDH is complex and poorly understood and likely involves both environmental and genetic factors. Studies showed that vitamin A homeostasis and retinoic acid contribute to the development of CDH.[5] Identical twin studies and incomplete penetrance of CDH suggest the possibility of epigenetic modifications in CDH development.[6] In addition to environmental factors, single gene knock out mouse models, rare single gene disorders, familial aggregation, and chromosome disorders associated with CDH support the contribution of genetics in disease etiology.[7–9]

Few disease-causing genes have been identified in CDH because of the small number of large CDH pedigrees due to the high mortality rate which limits the access to affected individuals and makes classical genetic approaches of linkage analysis more challenging. Chromosomal anomalies have been reported in as many as 30% of cases of CDH.[4] Karyotypes and/or fluorescence in situ hybridization (FISH) have identified many recurrent complete or partial aneuploidies in CDH such as trisomy 21, trisomy 18, 15q26 (OMIM 142340, DIH1), 8p23.1 (OMIM 222400, DIH2) and 1q41-q42 (OMIM 612530).[8, 10–13] Recently, new genomic technologies including array-based comparative genomic hybridization (aCGH) and single nucleotide polymorphism (SNP) microarrays have been widely used to identify copy number variants (CNVs) associated with CDH at 15q26 and 1q41-q42.[14, 15] De novo events have been found to be significantly enriched in CDH cases compared to controls or the unaffected siblings[16] and are often large and may alter one or more genes in dosage[17] and/or expression.[18]

In this study, we investigated the frequency and distribution of cytogenetic anomalies and de novo CNVs in a large series of CDH parent-proband trios. We also conducted a gene enrichment analysis for all the genes within the de novo CNVs and prioritized genes for future investigation.

Methods

Subjects

Patients were recruited as part of the DHREAMS (Diaphragmatic Hernia Research & Exploration; Advancing Molecular Science) study (http://www.cdhgenetics.com/). The DHREAMS study is a birth cohort of infants with a diaphragm defect identified at one of seven collaborating medical centers. Columbia University began enrollment in January, 2005, and recruitment at six other sites (Washington University Medical Center/St. Louis Children’s Hospital, University of Pittsburgh, Cincinnati Children’s Hospital and Medical Center/University of Cincinnati, Omaha Children’s Hospital/University of Nebraska, University of Michigan/CS Mott Children’s Hospital and Vanderbilt University) began from 2009 to 2010. All neonates with a diagnosis of a diaphragm defect were eligible for the study. Infants born at the study sites and those transferred into the study sites were eligible. Blood samples were collected from the affected neonate and both parents. Skin and diaphragm biopsies were collected at the time of the diaphragm repair. The birth cohort consists of 170 neonates born during the enrollment period.

In addition, we included a retrospective cohort of 71 patients opportunistically identified through CDH follow up clinics at the DHREAMS study sites, CDH patient support group networks and our research website. Eligible participants had to have a documented personal history of a diaphragm defect confirmed by medical record review. Deceased individuals were included if a post mortem tissue sample was available. A blood, saliva and/or tissue sample was collected from the affected patient and both parents.

Our series also includes 15 fetal cases recruited from Columbia University with a diaphragm defect confirmed by ultrasound and/or fetal MRI and electively terminated. Skin, diaphragm or other tissues or placenta were collected from fetal cases and blood or saliva samples were collected from parents.

The diaphragm defect was classified as an eventration, posteriolateral (Bochdalek) hernia, anterior hernia (Morgagni-Larrey) or agenesis of the hemidiaphragm and classified as left, right or central. Patients with no structural anomalies other than the diaphragm defect were classified as isolated and patients with at least one additional of the following birth defects as non-isolated: congenital heart disease, central nervous system defect, pyloric stenosis, omphalocele, polysplenia, asplenia, gastric dysmotility, Hirschprung disease, intestinal malrotation, situs inversus, genital urinary defect, skeletal anomalies, cleft lip/palate, abnormal new born hearing screen, dysmorphic features, club foot, congenital cystic adenomatoid malformation (CCAM), or bronchopulmonary sequestration (BPS). Pulmonary hypoplasia, cardiac displacement and intestinal herniation were considered to be part of the diaphragm defect sequence and were not considered to be an additional malformation. Clinical data were abstracted from the medical chart by study personnel at each site and included data on prenatal history, neonatal outcome, and longitudinal follow up data including Bayley III and Vineland II developmental assessments and a parent interview about the patient’s health since discharge at 2 years of age. A complete family history including history of diaphragm defects and major malformations was collected on all patients by a single genetic counselor. All the individuals with CDH and their parents provided informed consent/assent for participation in this study, which was performed in accordance with institutional review board approved protocols at the study site of enrollment.

Cytogenetic and clinical chromosome microarray analyses

Test reports were collected and reviewed for clinical conventional cytogenetic and clinical chromosome microarray analysis (karyotype, FISH; Array-CGH: NimbleGen CGX-3 array, SignatureChipOS 135K oligonucleotide array; Affymetrix SNP6.0 at LabCorp). DNA was available from diaphragm, blood, skin, amniocentesis, amniocytes, chorionic villus sampling (CVS) or other available tissues.

CNVs characterization

All probands who did not have a clinical chromosome microarray were analyzed with their parents using Affymetrix Human Genome-Wide SNP Array 6.0 (Santa Clara, CA). DNA was extracted from blood, diaphragm, saliva or other available tissues using Gentra PUREGENE DNA Blood and Tissue Isolation Kit (Qiagen, Valencia, CA). DNA sources available for genotyping are summarized in Table S1. Samples were prepared according to the manufacturer’s protocol (Affymetrix, Santa Clara, CA). The arrays were scanned by the GeneChip Scanner 3000 7G, following by the genotyping quality control (QC) using Affymetrix Genotyping Console (GTC) version 4.0.

We used both PennCNV [19]and Nexus Copy Number software (V6.0) (BioDiscovery Inc., El Segundo, CA) to call CNVs. In PennCNV copy number states with 1 and 2 were assigned as loss, 3 as normal and 4 or 5 as gain. In Nexus all the parameters thresholds were set to default values (segment mean log2 ratio of >0.23 or −0.37 were called gain or loss event, respectively). CNVs that were called in both algorithms with at least one base pair overlap were counted as the same segment. The Human NCBI Build 36.1 was used as reference genome. CNVs identified in the proband alone and absent from both parents were considered putative de novo CNVs. Putative de novo CNVs were then compared to the public databases such as Toronto Database of Genomic Variation (DGV) and the Copy Number Variation project at the Children’s Hospital Philadelphia (CHOP). Those CNVs that were present in normal controls with similar length were excluded from further analysis. To identify the potentially pathogenic variations, the remaining de novo CNVs were compared to the Database of Genomic Structural Variation (dbVar), DECIPHER database and the International Standards for Cytogenomic Arrays Consortium (ISCAC) databases. Gene functions were considered when there is no similar length of pathogenic CNVs in the databases.

Quantitative real-time PCR Confirmations

To independently confirm CNVs, real-time quantitative PCR (qPCR) analysis was performed for all the putative rare, de novo CNVs. Primer assays were either designed by primer3 or directly purchased from Applied Biosystems. Quantitative PCRs were carried out in a reaction volume of 10 μl with 20 ng genomic DNA in LightCycler 480 (Roche Diagnostics), using Lightcycler® Faststart DNA Master^PLUS SYBR Green I or USB® VeriQuest^™ Probe qPCR master mix. The conditions used for amplification were: one cycle of 95 °C for 10 min, followed by 50 cycles of 95 °C for 15 s and 59 °C for 1 min. All samples were run in triplicate, and data were normalized against the reference gene RNase P. Relative copy number was determined by the relative standard curve method. The relative copy number values for each interested region between 0.8–1.2 were considered normal while values ≥1.3 and ≤0.7 were considered evidence of duplication and deletion, respectively.

Gene prioritization and enrichment analysis

To determine the genes that were more likely to be involved in the pathogenesis of CDH, we performed gene prioritization analysis with the genes in the confirmed, de novo pathogenic CNVs. We composed a list of genes related to the development of the diaphragm through a comprehensive PubMed literature search using the keywords “diaphragm” and “gene” and put into ToppGene Suite[20] to make a comprehensive list of genes associated with human or mouse diaphragmatic hernia or diaphragm abnormality (Table S2). We use it as a training set to prioritize genes that may lead to an increased risk for CDH in our CNV intervals. We then tested the functional enrichment of those prioritized genes to identify the potential biological pathways involved in CDH. Gene prioritization was analyzed using ToppGene Suite[20] and Endeavour [21]. The p value in the ToppGene Suite was set as a default threshold 0.05 with false discovery rate correction. For better visualization, enriched gene sets were organized into a network. The network was generated by the plugin “Enrichment Map”[22] using Cytoscape[23] network software v.2.8.3 with moderately conservative parameters.

Statistics

The statistical analyses were performed using the R program. Chi-square test was used to assess the gene ontology enrichment and the association between de novo CNVs and survive rate and clinical syndromes.

Results

We studied 256 CDH trios composed of birth, retrospective and terminated fetal cohorts (Table 1). There are 143 males (55.8%) and 146 (57.0%) isolated CDH cases. Two hundred seventeen (84.8%) cases occurred on the left side. Among the 256 cases, 165 cases had either karyotypes (106) or clinical chromosome microarray (6) or both (53) and 91 cases had no genetic evaluation as part of routine clinical care. Except for two probands with Trisomy 21, all the other 254 cases had either clinical chromosome microarray (59) or Affymetrix 6.0 arrays (195) (Table 1).

Table 1.

Description of the CDH cohort

	Prospective Birth cohort	Retrospective cohort	Terminated fetal cases	Total
Number (256 in total)	170	71	15	256
Male/Female	97/73	41/30	5/10	143/113
Isolated/Non-Isolated	93/77	47/24	6/9	146/110
Left/Right/Other	141/23/6	63/4/4	13/2/0	217/29/10
White NH*/White H*/Black/Asian/Other	100/32/6/12/20	59/8/0/2/2	3/6/1/2/3	162/46/7/16/25
Karyotypes	109	35	15	159
Clinical Chromosome Microarrays	43	9	7	59
Research Chromosome Microarray	125	62	8	195

Note:

^*White NH refers to White non-Hispanic; White H refers to White Hispanic

Cytogenetic calls

A karyotype was completed on 159 of the 256 total patients (109/170 birth patients, 35/71 of the retrospective patients and 15/15 fetal patients). Eight of the 159 cases (5.0%) had chromosomal anomalies (Table 2). Two patients (04-0030 and 05-0007) had trisomy 21, one patient (05-0004) had mosaic trisomy 18 (50% cells had T18). Two patients (01-0118 and 01-437) had unbalanced translocations inherited from one parent with a balanced translocation. Two patients (01-0162 and 07-0010) had deletions at 8p23.1, and one patient (01-0024) had a deletion at 13q34.

Table 2.

Abnormal chrosomomal results in individuals with CDH

Patient ID	Gender	Cohort	Side of Lesion	Isolated/Non-isolated	Discharge status	Source	Result	detect method	Inheritance	Additional anomalies
04-0030	F	prospective	Left	Non-isolated	Died	Amniocentesis	47, XX +21	Karyotype	de novo	ASD, VSD
05-0007	M	prospective	morgagni hernia	Non-isolated	alive	Blood	47, XY +21	Karyotype	de novo	dysmorphic facial features consistent with Down syndrome
05-0004	M	prospective	Left	Non-isolated	alive	Blood	47, XY, +18[20]/46, XY[20]; 18p11.32-q23 (138,763-76,111,164)x3	Karyotype/clinical chromosomal microarray	de novo	VSD
01-0118	M	prospective	Left	Non-isolated	Died	Blood	46 XY, +der(22)t(11;22)(q23.3;q11.2)mat; 11q23.1-q25(116187461-134452384)x3; 22q11.1-q11.21(14432516-18709056)x3	Karyotype/research chromosomal microarray	Maternal	Dandy Walker malformation, Hirschprung’s disease, Preauricular pits, bilateral malformed thumbs, Pulm HTN
01-0437	F	prospective	Left	Non-isolated	alive	Blood	46, XX, add(7)(p22).ish der(7)t(7;12)(p22;p11.2); 7p22.3(52899-2004194)x1; 12p13.33-p11.22(20691-28622103)x3	Karyotype/clinical chromosomal microarray	Paternal	bilateral postaxial polydactyly, dysmorphic facial features
01-0162	F	prospective	Left	Non-isolated	alive	Diaphragm	46 XX, del(8)(p23.1p23.1).ish del(8)(pter+); 8p23.1(8043620-11883409)x1	Karyotype/research chromosomal microarray	de novo	ASD, VSD, Incomplete myelination in anterior limb of right and left
07-0010	M	prospective	Left	Isolated	Died	Amniocytes	46, XY, del(8)(p23.1); 8p23.2-p22 (192262-15227167)x1	Karyotype/clinical chromosomal microarray	de novo	None
01-0024	M	prospective	Left	Non-isolated	Died	Blood	46, XY, .ish del(13)(q34)(subtel13-, ZIC2+); 13q33.3-q34(107348518-114142980)x1	Karyotype/research chromosomal microarray	de novo	ASD, VSD, AV Canal, Absent right kidney with low-lying left kidney, left thumb aplasia, right talipes equinovarus, vertebral anomalies - gracile ribs and fusion of multiple vertebral bodies, scoliosis, contractures of right hand, Pulm HTN
01-0581	F	fetal	Left	Non-isolated	pregnancy termination	liver	1q12-qter(142693887-247249716)x3	clinical chromosomal microarray	de novo	CHD, agenesis of the corpus callosum, bilatera; ventriculomegaly, micrognathia, retrognathi posterior displacement of the hard and soft palate, posterior displacement of the tounge, bilateral low set ears, left sided microtia (small vestigial structure and absence of external ear canal), right sided underdevelopment of the helix, bilateral finger contractures
01-0375	M	fetal	Left	Non-isolated	pregnancy termination	CVS	1p13.3-p12 (107379866-119388162)X1	clinical chromosomal microarray	de novo	VSD, Abnormal posterior fossa with inferior splaying of cerebellar vermis, thickened nuchal fold (0.7 cm), no measurable cisterna magna, left absence of lens and orbit, left club foot, shorten long bones, flatten face, umbilical cord cyst
01-0536	F	retrospective	Left	Isolated	alive	Blood	15q25.2(80999115-82497270)x1	research chromosomal microarray	de novo	None
01-0454	F	retrospective	Left	Isolated	alive	Blood	17q12(32058213-33668463)x1	research chromosomal microarray	de novo	None
02-0008	F	retrospective	Left	Non-isolated	alive	Blood	17q12(31897638-33362422)x1	research chromosomal microarray	de novo	abnormal newborn hearing screen
05-0005	F	retrospective	Left	Non-isolated	alive	Diaphragm	6p25.3-p25.2(1190603-2762165)x1	research chromosomal microarray	de novo	VSD, Dandy walker malformation, cleft palate, optic nerve hypoplasia
07-0005	F	prospective	Right	Isolated	Died	Diaphragm	11p15.4(6733461-6850699)x1	research chromosomal microarray	de novo	None
01-0015	F	retrospective	Left	Isolated	alive	Blood	14q24.1(68911762-69007402)x3	research chromosomal microarray	de novo	None

Abbreviation: F, female; M, male; NA, not available; CVS, chorionic villus sampling; ASD, atrial septal defect; VSD, ventricular septal defect; AV Canal, Atrioventricular canal; Pulm HTN, pulmonary hypertension; CHD, congenital heart disease.

A clinical chromosome microarray was completed on 59 of the 256 total patients (43/170 birth patients, 9/71 retrospective patients, 7/15 fetal patients) (Table 1). Of the 59 cases, 53 also had karyotypes. Five patients had an abnormal chromosome microarray which included three patients who also had an abnormal karyotype (05-0004, 01-0437, and 07-0010). One patient (01-0581) had a mosaic duplication of 1q12-qter in 20% cells not observed by karyotype. The other patient (01-0375) had a 12 Mb deletion at 1p13.3-1p12 that was not detected on CVS karyotype. 9/10 patients with chromosomal anomalies detected by karyotype and/or clinical chromosome microarrays were non-isolated CDHs (Table 2).

Research chromosome microarray results

A total of 195 patients with CDH and their available parents were analyzed with Affymetrix SNP 6.0 arrays. Three probands had been initially reported with chromosomal anomalies by karyotypes and were analyzed to improve resolution of the breakpoints. We evaluated 54 putative de novo CNVs not found in normal controls using qPCR to independently confirm the results. Eight CNVs were inherited from one of the parents, and 36 did not confirm by qPCR. Forty-four (81.5%) of the 54 CNVs were false positives. Six CNVs were independently confirmed using qPCR and were de novo (Table 2), and ranged in size from 95 kb to 1.6 Mb. One CNV was duplication (01-0015); all others were deletions. The number of genes in these CNVs ranged from 2 to 22. Two of 6 patients with de novo CNVs were non-isolated CDHs. In addition, the breakpoints were refined for the four chromosomal anomalies which were detected by karyotype alone (Table 2). Patient 01-0118 had an 18.3 Mb duplication at 11q23.1-q25 and a 4.3 Mb length of duplication at 22q11.1-q11.21. Patient 01-0024 had a 6.8 Mb deletion at 13q33.3-q34. Patient 01-0162 had a 3.8 Mb deletion at 8p23.1.

When the chromosomal anomalies were identified in diaphragm, we tested all DNA from the other available tissues. No difference was found between DNA from diaphragm or other tissues. In summary, there were 16 CDH patients with cytogenetic anomalies including 3 patients with aneuploidies, two patients with unbalanced translocations inherited from a parent with a balanced translocation, and 11 patients with de novo CNVs ranging in size from 96 kb to 104.6Mb at 1p13.3-p12, 1q12-qter, 6p25.3-p25.2, 8p23.1, 8p23.2-p22, 11p15.4, 13q33.3-q34, 14q24.1, 15q25.2, and 17q12. There were 9/170 (5.3%) patients from the birth cohort, 5/71 (7.0%) patients from retrospective cohort and 2/15 (13%) patients from fetal cohort with cytogenetic anomalies. There is an association of chromosomal anomalies and survival rate (4/9 (44.4%) vs. 141/161(87.6%)) in the birth cohort (p=0.004). The participants with abnormal cytogenetic or CNV results were more likely to be non-isolated CDHs (7/9 (77.8%) vs. 70/161 (43.5%)) (p=0.05) in the birth cohort.

Prioritized genes

To further characterize the relevant genes in the CNVs, we performed gene prioritization analysis. Trisomy18 and 21 were excluded for this analysis. A training set of 112 genes was used to test the genes in the pathogenic CNV regions (Table S2). There are 1769 protein-coding genes within the CNVs. Seven hundred and ten genes were predicted to be significant based on the all parameter sets in Toppgene. KCNA2 at 1p13.3-1p12; LMNA, CACNA1S, MYOG, HLX, and AGT at 1q12-qter; GATA4 at 8p23.1 or 8p23.2-p22; HYLS1 at 11q23.1-q25 were already included in our training gene set, and were designated as priorities. Studies in a CDH cohort suggested DISP1 at 1q41q42.12 as a candidate CDH gene.[24] We also include this as our priority gene. FOXC1, FOXF2 at 6p25.3-p25.2; PDGFA at 7p22.3; FGF6 at 12p13.33-p11.22; COL4A1, COL4A2 at 13q33.3-q34; HOMER2, BNC1 at 15q25.2; LHX1 at 17q12; BID, TBX1 at 22q11.1-q11.21 were the top novel genes predicted in each CNV interval by ToppGene and Endeavour (Table 4). No significant genes were found in regions of 11p15.4 (07-0005) and 14q24.1 (01-0015); these two regions were then excluded for following analysis.

Gene ontology enrichment analysis

Besides the top predicted priority genes, there were 1,769 genes in the CNV regions and 710 genes were predicted to be significant. To systematically investigate the connections among those 710 genes in CNVs, we performed gene ontology (GO) enrichment for all the 710 significance genes (p<0.05) and 8 priority training genes in CNV regions. The 718 genes were enriched in 238 biological process and 46 molecular function GO categories with significance (p<0.05). For better visualization, we generated a functional map using enrichment map plugin in cytoscape (Figure 1). We found the most significant biological process are tissue development (p=4.4×10⁻¹¹) and regulation of multicellular organismal processes (p=2.8×10⁻¹⁰), which are the most likely relevant biological processes to CDH (Table S3). The 162 genes in these GO categories are distributed in all CDH CNV regions, including diaphragm hernia related or predicted genes LMNA, CACNA1S, MYOG, HLX, AGT, GATA4, FOXC1, FOXF2, PDGFA, FGF6, LHX1 and TBX1. The most significant molecular functions for these 162 genes are “receptor binding” (p = 8.7×10⁻¹⁴) and “DNA binding transcription factor activity” (p = 4.4×10⁻¹⁰). Figure S1 shows the top 20 molecular function gene enrichment set for the 162 genes.

Figure 1.

Gene enrichment sets were mapped as a network. Gene enrichment analysis results in 238 biological process and 46 molecular function gene ontology (GO) categories. Each node represents a GO. The GO categories are related by mutual overlap (edges). The major functional enrichment sets are highlighted by text. The two GO categories with bold fonts are the most significant ones.

Discussion

CDH is a multifactorial birth defect with a significant genetic contribution. In our study of 256 parent child trios with CDH, we identified a total of 16 of 256 (6.3%) patients with aneuploidies, unbalanced translocations, or de novo copy number variations which is consistent with the previous studies with a prevalence of cytogenetic anomalies ranging from 2%–31%.[4] No tissue specific cytogenetic anomalies were found, suggesting that genetic anomalies isolated to the diaphragm are not common. Since pleuroperitoneal folds are thought to be the origin of posterolateral diaphragmatic hernia forms,[25] our method to screen for mosaic CNVs in diaphragm muscle could still have missed CNVs isolated to the pleuroperitoneal folds.

We identified aneuplodies in three patients (trisomy 18 and 21). We identified two patients with unbalanced translocations inherited from parents with balanced translocations which significantly increase the risk of recurrence for these two families. In addition, we identified 11 patients with de novo CNVs on chromosome microarray ranging in size from 96 kb to 104.56 Mb. The chromosome microarrays were useful in defining the breakpoints of the cytogenetically visible anomalies to precisely map the genes contained within these regions of gain or loss. We observed losses more frequently than gains (9 losses and 2 gains in the segmental aneuploidies). In general, deletions are generally more deleterious than their corresponding duplications. In our birth cohort, we found that the CDH patients with chromosomal anomalies were more likely to be non-isolated and to have other additional birth defects and had an increased mortality suggesting a worse prognosis for those CDH cases with cytogenetic anomalies.

Our study is the largest CDH series evaluating cytogenetic abnormalities. Our prospectively collected CDH birth cohort is also the least biased series to analyze the association between chromosomal anomalies and mortality associated anomalies in CDH patients compared with the previous reported series which were retrospectively collected. Our rate of 6.3% visible cytogenetic anomalies and microdeletions/microduplications is lower than many of the previously reported series (Table S3), but many of the previous series were based on small numbers of cases with as few as 12 patients and oversampled cases with known cytogenetic anomalies. Our series may have some bias of ascertainment as CDH is increasingly diagnosed prenatally and in some cases chromosome prenatal microarrays are performed with elective termination of some fetuses with cytogenetic anomalies. We observed a larger fraction of chromosomal anomalies in patients with non-isolated compared with isolated patients in our birth cohort (9.1% vs. 2.1%), consistent with prior studies.[26, 27]

We identified recurrent deletions of 8p23.1 in two patients (01-0162 and 07-0010) a deletion previously reported in CDH patients.[12, 28–30] Overlapping 3.8 Mb and 15.1 Mb de novo deletions were identified in a non-isolated (01-0162) and isolated (07-0010) CDH patient, respectively (Figure 2A). Our results narrowed down the 8p23.1 critical region to 3.8 Mb, and further strengthen the association of haploinsufficiency of one or more genes in 8p23.1 with CDH. Twenty-three genes map to the 3.8 Mb critical region including GATA4, a zinc finger transcription factor. The majority of patients with CDH with deletions of 8p23.1 also have cardiac anomalies, and deletions or loss-of-function mutations in GATA4 have been identified in individuals with structural congenital heart disease. Heterozygous Gata4^+/Δex2 mice display cardiac, lung and diaphragm defects.[31] However, no point mutations in GATA4 have been identified in any humans with CDH to date. One explanation is that CDH may require altered expression of more than one gene in the deletion interval to affect development of the diaphragm. Alternatively, the number of CDH patient screened for GATA4 mutations has been quite modest [30]. Gene prioritization and network results demonstrated that SOX7 is an additional candidate gene in this minimal deletion region. SOX7 is a transcription factor involved in the regulation of embryonic development. The gene plays crucial roles in parietal endoderm differentiation through regulating GATA4.[32] Wat MJ et al suggested that haploinsufficiency of SOX7 may increase the severe cardiac phenotype in individuals with a GATA4 deletion.[30]

Figure 2.

The recurrent CDH regions. A. The deletions of 8p23.1 associated with CDH. The red bars with solid or gradient shape are the maximally deleted regions for each patient. There are two deletions identified in our patients and 01-0162 deletion narrowed to the minimum critical 8p23.1 CDH CNV region. Genes in the critical CNV are shown with blue bars. Priority genes are shown as red font and bar. B. The deletions of 15q25.2 associated with CDH. The minimum deleted regions for each patient are shown with solid bars. The light blue bar is the 1.5 Mb deletion in 01-0536. Thirteen genes in the narrowed critical 15q25.2 CDH region are indicated with blue bars. Priority genes are shown as red font and bar. C. The recurrent pathogenic 17q12 deletions among ISCAC database and CDH patients. Red squares indicate the data from ISCAC; green square represents a 1.4 Mb 17q12 deletion reported recently; light blue squares indicate the deletions of our CDH patients (02-0008 and 01-0454). Genes in the overlapped deletion regions are indicated with blue bar. LHX1 is the predicted priority gene in this region shown as red font and bar.

Another recurrent deletion is the de novo 1.5 Mb microdeletion at 15q25.2 in patient (01-0536). Three other patients with CDH have been reported with 15q25.2 deletions [33, 34] (Figure 2B). A fetus with CDH and mild hydrocephalus with a ~2.5 Mb deletion [34] and two 1.6 Mb and 2.5 Mb non-isolated CDH patients were reported.[33] Other phenotypes associated with 15q25.2 deletions include intellectual disabilities, congenital heart disease, cryptorchidism and short stature.[33–35] The deletion in our patient is within the reported CDH 15q25.2 deletion region, and narrows the critical CDH regions to 13 protein coding genes in this region. HOMER2 and BNC1 were predicted to be one of the priority genes mapping to this region. HOMER2 is a scaffolding protein that links carious synaptic signaling proteins. Homer proteins and their mRNAs expression in mouse heart, skeletal muscle and diaphragm.[36] Basonuclin (BNC1) as a transcription factor that performs a regulatory function in the cell growth and differentiation may also plays a role in the development of diaphragm since studies showed that impaired cell proliferation might contribute to the abnormal diaphragm development.[25]

Microdeletion of 17q12 has been frequently reported to be associated with psychiatric abnormality, macrocephaly, developmental delay or renal disease (data from ISCAC, Figure 2C). Recently, Hendrix et al found a 1.4 Mb deletion at 17q12 in a CDH fetus.[37] As shown in Figure 2C, our two patients and this CDH fetus had overlapping 17q12 deletions and narrowed the critical region to 1.39 Mb containing 9 protein-coding genes. Gene enrichment analysis predicts Lim Homeobox1 (LHX1) as a candidate gene for CDH. LHX1 has a unique cysteine-rich zinc-binding domain, and chromatin immunoprecipitation analysis found that Lhx1 promoter region is directly bound by Dlx5, a candidate gene for split-hand/split-foot type 1 malformation associated with sensorineural hearing loss.[38] One of our patients (02-0008) had a mild hearing loss. Dlx5 also plays a key role in development of many organs. The balance of antagonistic interaction between Lhx1 and Dlx5[38] may be disrupted by haploinsufficiency of LHX1leading to CDH.

There are several previously reported CDH patients with duplication of 1q.[39, 40] Most of these duplications are the consequence of a balanced parental translocation. No patients including our patient with a large 1q duplication have survived for long periods except one case with CDH and multiple anomalies.[41]

Three de novo deletions at 13q33.3-q34, 6p25.3-p25.2 and 1p13.3-p12 are novel CDH deletions. There are two genes (ERH, SLC39A9) in 95 kb duplication region at 14q24.1 and no similar pathogenic CNVs found in other databases. In analyzing the 117 kb deletion of 11p15.4, we identified there are two similar size CNVs (nsv467700 and nsv467701) in dbVar and one (chr11:6710956-6762848) in DECIPHER. CNVs nsv467700 and nsv467701 are from a control set study. The CNV in DECIPHER is inherited. There are no similarly sized CNVs in ISCAC. There are five protein coding genes in the 11p15.4 interval; all of them are olfactory receptors. The gene priority analysis did not identify significant genes that related to CDH at 11p15.4 and 14q24.1 intervals. All the evidence suggests these two variants may not be pathogenic although they are de novo.

Gene prioritization and enrichment analysis identified the most significant biological processes: tissue development/organ regulation categories. Genes within these two categories were significantly enriched in transcription factor molecular functions, consistent with findings that transcription factor genes such as GATA4 (OMIM: 600576)[31] and MYOG (OMIM: 159980)[42] are implicated in diaphragm development. Immune response related gene sets were frequently included in the aneuploid segments, consistent with previous studies that genes involved in immune and environment response are frequently included in pathogenic CNVs.[43]

A limitation of our study is that the training CDH gene list is based on the limited knowledge of the implicated relationship of these genes with diaphragm development. However, there is little direct genetic evidence showing that mutations in these genes are implicated in CDH. Our enrichment and prioritization studies are based on these genes and an assumption of association with CDH; thus there may be a bias toward detecting similar genes based upon our training list genes. Although the genes we selected based on the above analyses were most likely involved in CDH, we did not establish a direct relationship between these genes and CDH. Ultimately, it will be important to identify the mutations in individual genes to definitely demonstrate association with CDH. Russell et al.[44] prioritized CDH candidate genes based upon primordial/mature diaphragm transcriptome data and identified Crabp2 and Pbx1 which map to our 1q12-qter deletion region as highly expressed in the pluriperitoneal folds and potentially relevant to human CDH. Our study may have underestimated the overall frequency of CNVs associated with CDH. We could have eliminated some CNVs by filtering with DGV since not all the CNVs in DGV have been verified or are necessarily benign. We could have missed some CNVs called by only one of the two algorithms but believes this is unlikely given the high percentage of CNVs that were not confirmed by qPCR. Most of the chromosomal anomalies were identified in non-isolated CDH patients, and there may be other genetic or non-genetic causes of isolated CDH. Since the CNVs alter the copy number of several genes in these intervals, we cannot determine if the same or different genes in these intervals cause all the structural anomalies. Although the number of patients included in our series was the largest CDH series reported for CNV analysis, the number of patients was modest and replication of these results and identification of other patients with overlapping CNVs will confirm and further define the relevant intervals.

In summary, using high resolution copy number variation analysis and network analysis, we have identified 16 CNVs in 256 CDH parents trios, of which 14 are likely to be pathogenic. The genes we have identified provide a set of high priority genes for further analysis in CDH studies.

Table 3.

Candidate genes in pathogenic CNVs

Chromosome Location	Coordinates (hg18; bp)		Size (bp)	Proband	Event	Genes
	Start	End
1p13.3-p12	107379866	119388162	12008296	01-0375	Loss	KCNA2
1q12-qter	142693887	247249716	104555829	01-0581	gain	LMNA, CACNA1S, MYOG, HLX, AGT, DISP1
6p25.3-p25.2	1190603	2762165	1571562	05-0005	Loss	FOXC1, FOXF2
7p22.3	52899	2004194	1951295	01-0437	Loss	PDGFA
8p23.1	8043620	11883409	3839789	01-0162	Loss	GATA4, SOX7
8p23.2-8p22	192262	15227167	15034905	07-0010	Loss	GATA4, SOX7
11q23.1-q25	116187461	134452384	18264924	01-0118	Gain	HYLS1
12p13.33-p11.22	20691	28622103	28601412	01-0437	Gain	FGF6
13q33.3-q34	107348518	114142980	6794462	01-0024	Loss	COL4A1, COL4A2
15q25.2	80999115	82497270	1498156	01-0536	Loss	HOMER2, BNC1
17q12	31897638	33362422	1464784	02-0008	Loss	LHX1
17q12	32058213	33668463	1610250	01-0454	Loss	LHX1
22q11.1-q11.21	14432516	18709056	4276541	01-0118	Gain	BID, TBX1