Bicuspid aortic valve (BAV) is the most frequent congenital heart defect and has a male predominance of 3 to 1. A large proportion of patients develop valvular and aortic complications. Despite the high prevalence of BAV, its cause and genetic origins remain elusive. The goal of this study was to identify genetic variants associated with BAV. Nine genes previously associated with BAV ( NOTCH1 , AXIN1 , EGFR , ENG , GATA5 , NKX2-5 , NOS3 , PDIA2 , and TGFBR2 ) were sequenced in 48 patients with BAV using the Ion Torrent Personal Genome Machine. Pathogenicity of genetic variants was evaluated with the Combined Annotation Dependent Depletion framework. A selection of 89 variants identified by sequencing or in previous BAV genetic studies was genotyped, and allele frequencies were compared in 323 patients with BAV confirmed at surgery and 584 controls. Analyses were also performed by gender. Nine novel and 19 potentially pathogenic variants were identified by next-generation sequencing and confirmed by Sanger sequencing, but they were not associated with BAV in the case-control population. A significant association was observed between an in silico–predicted benign EGFR intronic variant (rs17290301) and BAV. Analyses performed by gender revealed different variants associated with BAV in men ( EGFR rs533525993 and TEX26 rs12857479) and women ( NOTCH1 rs61751489, TGFBR2 rs1155705, and NKX2-5 rs2277923). In conclusion, these results constitute the first association between EGFR genetic variants and BAV in humans and support a possible role of gender-specific polymorphisms in the development of BAV.
Bicuspid aortic valve (BAV) is the most common congenital heart defect with a prevalence estimated at 1%. Valvular and aortic complications, such as calcific aortic valve stenosis (AVS), are frequent in patients with BAV and usually occur earlier in life compared with those with tricuspid aortic valves (TAV) ; 30% to 50% of all patients who underwent surgical aortic valve replacement for AVS are affected by BAV. BAV is, therefore, a significant medical and economic burden that must be addressed. Despite the high prevalence of BAV, our understanding of its cause and genetic origins is still fairly limited. NOTCH1 is known as the most well-proved candidate gene for BAV and has been associated with both familial and sporadic forms of the defect in humans. However, deleterious genetic variants in NOTCH1 are only detected in a minority of patients with BAV. The contribution of additional genes, including AXIN1 , EGFR , ENG , GATA5 , NKX2-5 , NOS3, PDIA2 , and TGFBR2 , has been suggested. The purpose of the present study was to identify potentially pathogenic genetic variants in these 9 genes and to assess the frequencies of these variants and other previously identified BAV-related variants in a case-control series of patients with and without BAV. This approach will complement other types of genetic studies to elucidate the inherited component of BAV.
Methods
A flowchart diagram outlining the main stages of this study is provided in Supplementary Figure 1 . Nine hundred forty-five unrelated French Canadians with either BAV or TAV were recruited at the Institut universitaire de cardiologie et de pneumologie de Québec (Quebec City, Canada) from 2000 to 2014. The BAV subset consisted of 339 patients who underwent their first surgical aortic valve replacement. BAV phenotype was confirmed by the cardiac surgeon during the procedure and confirmed by pathological analyses of explanted tissues. BAV patients with previous cardiac surgery were excluded from the study. The TAV subset consisted of 606 patients, including 480, who underwent isolated coronary artery bypass grafting with no signs of AVS at preoperative echocardiography and 126 with severe AVS requiring aortic valve replacement. Patients with TAV and a history of previous cardiac surgery, disease involving other cardiac valves, or presenting with moderate or severe aortic insufficiency were excluded. The study was approved by the local Research Ethics Board, and all participants gave their written informed consent (REB #20771).
Genomic DNA was extracted from buffy coat using QIAamp DNA Blood Midi Kit (QIAGEN, Hilden, Germany). Total DNA concentration was measured by UV with Nanovue Plus spectrophotometer (GE Healthcare, Little Chalfont, UK), and DNA quality was evaluated by the 260/280 nm absorbance ratio (≥1.8). Double-stranded DNA quantification was assessed by Quant-iT PicoGreen fluorescence (Invitrogen, Carlsbad, CA) with the Synergy HT fluorometer (Biotek, Winooski, VT).
Variants were identified by next-generation sequencing (NGS) in a homogeneous and representative sample of 48 participants from the BAV subset. These patients were not affected by coarctation of the aorta, ascending aortic aneurysm, and moderate-to-severe aortic insufficiency. DNAs were sequenced at the McGill University and Génome Québec Innovation Center (Montreal, Canada) using the Ion Torrent Personal Genome Machine (PGM) (Life Technologies, Carlsbad, CA). An Ion AmpliSeq custom panel of primers covering the entire coding DNA sequence, 1,000 bp of the promoter region, and both 3′ and 5′ untranslated regions (UTR) was designed for the 9 genes of interest (Ion AmpliSeq Designer, version 3.0, reference genome hg19). The final panel covered 90% (53 kb) of the regions included in the initial design for a total of 333 amplicons ranging from 125 to 275 bp in length in 2 pools of 168 and 165 primer pairs, respectively. The Ion Torrent DNA libraries were prepared according to the manufacturer’s instructions. Quantification of the libraries was performed using a Bioanalyzer High Sensitivity DNA kit (Agilent, Santa Clara, CA) and Quant-iT PicoGreen. Purification was achieved with Agencourt AMPure XP (Beckman Coulter, Brea, CA). Libraries were loaded on three 316 chips, each supporting 16 of the 48 BAV samples. The data generated were mapped to the hg19 genome. Variants were considered previously characterized if reported in dbSNP (build 138), 1,000 Genomes Project (phase 3), Exome Aggregation Consortium (v.0.3), or Exome Variant Server (release ESP6500 SI-V2). Variants were considered novel if not listed in any of these databases.
Scaled scores from Combined Annotation Dependent Depletion (CADD) framework (C-scores) were obtained for all variants identified by NGS to assess their pathogenicity. CADD ranks single-nucleotide variants, insertions, and deletions (indels), coding or not, according to their deleteriousness. A C-score ≥10 indicates that a particular variant is part of the top 10% most deleterious changes in the human genome and a C-score ≥20 indicates the top 1%. The cut-off value to separate potentially pathogenic from harmless variants is arbitrary, but a C-score threshold value between 10 and 15 is recommended. In this study, variants with C-scores ≥10 were considered potentially pathogenic. Missense variants were further evaluated with PolyPhen-2, a bioinformatics tool estimating the functional impact of missense variants according to their sequence homology throughout evolution. For any given variant, a score ranging from 0 (benign) to 1 (probably damaging) is generated by PolyPhen-2.
Sixty-six variants were selected for validation using Sanger sequencing. The selection consisted of every novel genetic variant identified by the NGS experiment (n = 14 single-nucleotide variants, 15 indels), previously characterized indels (n = 9), and missense variants (n = 20) and synonymous or noncoding variants with C-scores ≥10 (n = 8). For each selected variant, at least 1 and up to 3 patients with BAV were sequenced if available: 1 homozygous for the common allele, 1 heterozygous, and 1 homozygous for the rare allele. Sequencing was performed using a 3730xl DNA Analyzer (ABI, Foster City, CA) at the Centre Hospitalier de l’Université Laval sequencing and genotyping platform (Quebec City, Canada). Primer sequences designed to validate the selected DNA variants are provided in Supplementary Table 1 . Nonvalidated variants were discarded from subsequent analyses.
The frequencies of 44 variants identified by NGS and 45 additional variants identified in previous BAV genetic studies were genotyped in 339 cases and 606 controls using the Illumina GoldenGate genotyping assay on the Illumina VeraCode and BeadXpress platform. All variants and the rationale for their selection are listed in Supplementary Table 2 . Clusters of genotyping data were generated by Illumina Genome Studio v2011.1 and quality controls were carried out with PLINK v1.07. Clustering accuracy was assessed for the 37 variants for which Sanger sequencing data were available in the 48 patients with BAV and clusters were manually adjusted as needed. Thirteen variants were excluded from further analyses following the quality controls; 7 had a genotype call rate <90%, 4 deviated significantly from Hardy-Weinberg equilibrium (p value <0.0001), and 2 had poor genotyping clusters. Thirty-eight patients were excluded because of low completion rate (<90%, n = 34) or sex mismatch (n = 4). A total of 323 cases and 584 controls were considered after all quality control filters. A rare NOTCH1 missense variant resulting in an amino acid change from glycine to serine at position 152 (G152S, g.139417590 C > T) failed quality controls but was genotyped in 339 cases and 606 controls using an allele-specific PCR assay. Primer sequences created for this assay are listed in Supplementary Table 3 .
Minor allele frequencies of genotyped variants were compared between patients with BAV and TAV using chi-square tests. Given the 3 to 1 male predominance in BAV, additional analyses were performed by gender. p Values ≤0.05 were considered statistically significant.
Results
Demographic and clinical characteristics of the study populations following genotyping quality controls are presented in Table 1 (see Supplementary Table 4 for characteristics stratified by gender). Patients with BAV were younger, had a higher proportion of men, and a lower proportion of diabetes and hypertension than their TAV counterparts. Patients with BAV showed a larger aortic valve area compared with those with calcific AVS TAV. AVS was the most predominant valvulopathy in patients with BAV.
| Variable | BAV (n = 323) | TAV (n = 584) | p-Value |
|---|---|---|---|
| Age (years) | 62.8 ± 10.2 | 71.3 ± 8.6 | 2.20×10 -16 |
| Male gender | 235 (72.8%) | 333 (57.0%) | 3.86×10 -6 |
| Body mass index (kg/m 2 ) | 27.6 ± 4.8 [1] | 27.7 ± 4.5 | 0.69 |
| Diabetes mellitus | 53 (16.4%) | 171 (29.3%) | 2.40×10 -5 |
| Hypertension | 178 (55.1%) | 417 (71.4%) | 1.09×10 -6 |
| Peak aortic gradient (mm Hg) | 73.1 ± 28.6 [35] | 73.9 ± 26.4 [9] ∗ | 0.80 |
| Mean aortic gradient (mm Hg) | 44.8 ± 18.2 [30] | 44.9 ± 17.3 [7] ∗ | 0.96 |
| Aortic valve area (cm 2 ) | 0.81 ± 0.43 [28] | 0.74 ± 0.21 [1] ∗ | 0.02 |
| Ascending aorta replacement | 96 (29.7%) | N/A | N/A |
| Predominant aortic valve disease | |||
| Aortic stenosis | 235 (72.8%) | N/A | N/A |
| Aortic insufficiency | 22 (6.8%) | N/A | N/A |
| Mixed | 37 (11.4%) | N/A | N/A |
| None | 27 (8.4%) | N/A | N/A |
| Missing | 2 (0.6%) | N/A | N/A |
∗ In the TAV subset, peak and mean aortic gradient and aortic valve area values were only available for the AVS patients (n = 122). Aortic stenosis was defined by a mean aortic gradient >25.0 mm Hg and/or an aortic valve area <1.0 cm 2 . Aortic insufficiency was defined by an insufficiency score ≥3 (moderate or severe insufficiency).
A total of 217 genetic variants were identified across the 9 candidate genes using the Ion Torrent PGM, including 29 novel variants (mean coverage of 340× across all amplicons) ( Table 2 ; Supplementary Table 5 and Supplementary Figure 2 ). Eighty-two variants were located in introns, 68 in exons (47 synonymous and 21 missense), 25 in 3′UTR, 14 in promoter regions, 3 in 5′UTR, and 1 in downstream gene region. Seven indels were found in introns, 6 in 3′UTR, 9 in coding regions, and 2 in promoter regions.
| Gene | Total Variants (novel) | Variants Considered for Validation | ||
|---|---|---|---|---|
| Total | Validated (novel) | Validated With C-Score ≥ 10 | ||
| AXIN1 | 26 (0) | 6 | 6 (0) | 1 |
| EGFR | 34 (5) | 8 | 6 (3) | 2 |
| ENG | 15 (2) | 4 | 2 (0) | 1 |
| GATA5 | 20 (5) | 10 | 6 (1) | 2 |
| NKX2-5 | 10 (3) | 4 | 2 (1) | 0 |
| NOS3 | 22 (3) | 5 | 4 (2) | 0 |
| NOTCH1 | 56 (6) | 15 | 10 (1) | 7 |
| PDIA2 | 22 (2) | 10 | 8 (0) | 6 |
| TGFBR2 | 12 (3) | 4 | 2 (1) | 0 |
| Total | 217 (29) | 66 | 46 (9) | 19 |
Forty-six of the 66 genetic variants (70%) selected for re-sequencing were validated ( Table 2 ). All previously known variants (n = 37) sent for re-sequencing were validated. Of the 29 novel variants originally identified by NGS, 9 were confirmed. Novel variants were identified in EGFR , GATA5 , NKX2-5 , NOS3 , NOTCH1 , and TGFBR2 , but they were not considered deleterious ( Table 3 ).
| Gene | Genomic Position | Function | Exon | cDNA Position | Protein Position | Minor Allele | MAF BAV (n = 48) | Number of Patients | C-Score |
|---|---|---|---|---|---|---|---|---|---|
| EGFR | g.55211066 C > T | synonymous | 3 | c.555 C>T | p.N103N | T | 0.010 | 1 (2.1%) | 9.07 |
| g.55238316 G > C | intronic | C | 0.010 | 1 (2.1%) | 3.29 | ||||
| g.55241789 T > G | intronic | G | 0.010 | 1 (2.1%) | 5.51 | ||||
| GATA5 | g.61050015 C > T | intronic | T | 0.010 | 1 (2.1%) | 7.11 | |||
| NKX2-5 | g.172659219 C > T | 3’UTR | T | 0.010 | 1 (2.1%) | 6.61 | |||
| NOS3 | g.150700993 T > A | intronic | A | 0.010 | 1 (2.1%) | 8.36 | |||
| g.150704101 C > T | intronic | T | 0.010 | 1 (2.1%) | 0.48 | ||||
| NOTCH1 | g.139405287 AC > A | deletion (intronic) | A | 0.010 | 1 (2.1%) | 4.09 | |||
| TGFBR2 | g.30733726 G > A | 3’UTR | A | 0.010 | 1 (2.1%) | 9.38 |
Nineteen previously characterized variants validated by Sanger sequencing were evaluated as potentially pathogenic by CADD ( Table 4 ). The highest burden of deleterious variants was found in NOTCH1 , where 7 variants identified by NGS had a C-score ≥10. Among these was rare missense variant G152S, identified in 1 heterozygous BAV patient by NGS. This variant was evaluated as probably damaging by PolyPhen-2 (score of 0.969) and had a C-score of 19.74. AXIN1 , EGFR , ENG , GATA5, and PDIA2 also carried previously known variants evaluated as potentially pathogenic ( Table 4 ).
| Gene | Genomic Position | rs Number | Function | Exon | cDNA Position | Protein Position | Minor Allele | MAF BAV (n = 48) | Number of Patients | PP2 Score | C-Score |
|---|---|---|---|---|---|---|---|---|---|---|---|
| AXIN1 | g.348222 C > T | rs214250 | synonymous | 6 | c.1673 C>T | p.S428S | T | 0.292 | 23 (47.9%) | 12.33 | |
| EGFR | g.55240803 C > T | rs55669340 | synonymous | 17 | c.2293 C>T | p.L683L | T | 0.021 | 2 (4.2%) | 10.54 | |
| g.55249063 G > A | rs1050171 | synonymous | 20 | c.2607 G>A | p.Q787Q | G | 0.437 | 39 (81.2%) | 17.35 | ||
| ENG | g.130588091 C > T | rs41322046 | missense | 5 | c.990 C>T | p.G191D | A | 0.021 | 2 (4.2%) | 0.999 | 16.95 |
| GATA5 | g.61041653 G > C | rs6061244 | intronic | C | 0.365 | 28 (58.3%) | 10.05 | ||||
| g.61039662 T > C | rs73149261 | 3’UTR | C | 0.010 | 1 (2.1%) | 10.28 | |||||
| NOTCH1 | g.139440852 A > G | rs3013307 | promoter | A | 0.312 | 44 (91.7%) | 10.08 | ||||
| g.139440845 G > A | rs3013306 | promoter | G | 0.333 | 42 (87.5%) | 10.66 | |||||
| g.139417590 C > T | N/A | missense | 4 | c.454 C>T | p.G152S | T | 0.010 | 1 (2.1%) | 0.969 | 19.74 | |
| g.139413908 C > T | rs2229975 | synonymous | 5 | c.852 C>T | p.P284P | A | 0.146 | 14 (29.2%) | 10.25 | ||
| g.139401233 C > T | rs61751543 | missense | 23 | c.3836 C>T | p.R1279H | T | 0.021 | 2 (4.2%) | 0.019 | 16.23 | |
| g.139400320 G > A | rs183156491 | missense | 25 | c.4028 G>A | p.A1343V | A | 0.010 | 1 (2.1%) | 0.944 | 17.59 | |
| g.139399876 G > A | rs369915496 | missense | 25 | c.4472 G>A | p.T1491M | A | 0.010 | 1 (2.1%) | 0.870 | 17.68 | |
| PDIA2 | g.334543 C > G | rs45614840 | missense | 7 | c.1464 C>G | p.T119R | G | 0.094 | 8 (16.7%) | 1.000 | 13.79 |
| g.336377 C > G | rs45529833 | missense | 13 | c.2252 C>G | p.P382A | G | 0.010 | 1 (2.1%) | 0.999 | 13.08 | |
| g.336381 CCT > C | rs375604412 | nonsense | 13 | c.1148_1149delCT | p.P383R (p.V390X) ∗ | C | 0.010 | 1 (2.1%) | 14.35 | ||
| g.336701 ACT > A | rs201624048 | nonsense | 14 | c.1388_1389delCT | p.L464Q (p.I476X) ∗ | A | 0.010 | 1 (2.1%) | 14.94 | ||
| g.336731 G > A | rs116969376 | missense | 14 | c.2526 G>A | p.R473A | A | 0.010 | 1 (2.1%) | 0.839 | 12.48 | |
| g.336916 C > T | rs1048786 | missense | 15 | c.2612 C>T | p.P502S | T | 0.135 | 12 (25.0%) | 0.002 | 10.80 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
