Whole Exome Sequencing Reveals a Monogenic Cause of Disease in ≈43% of 35 Families With Midaortic SyndromeNovelty and Significance
Midaortic syndrome (MAS) is a rare cause of severe childhood hypertension characterized by narrowing of the abdominal aorta in children and is associated with extensive vascular disease. It may occur as part of a genetic syndrome, such as neurofibromatosis, or as consequence of a pathological inflammatory disease. However, most cases are considered idiopathic. We hypothesized that in a high percentage of these patients, a monogenic cause of disease may be detected by evaluating whole exome sequencing data for mutations in 1 of 38 candidate genes previously described to cause vasculopathy. We studied a cohort of 36 individuals from 35 different families with MAS by exome sequencing. In 15 of 35 families (42.9%), we detected likely causal dominant mutations. In 15 of 35 (42.9%) families with MAS, whole exome sequencing revealed a mutation in one of the genes previously associated with vascular disease (NF1, JAG1, ELN, GATA6, and RNF213). Ten of the 15 mutations have not previously been reported. This is the first report of ELN, RNF213, or GATA6 mutations in individuals with MAS. Mutations were detected in NF1 (6/15 families), JAG1 (4/15 families), ELN (3/15 families), and one family each for GATA6 and RNF213. Eight individuals had syndromic disease and 7 individuals had isolated MAS. Whole exome sequencing can provide conclusive molecular genetic diagnosis in a high fraction of individuals with syndromic or isolated MAS. Establishing an etiologic diagnosis may reveal genotype/phenotype correlations for MAS in the future and should, therefore, be performed routinely in MAS.
Midaortic syndrome (MAS) is a rare form of aortic coarctation. It represents 0.5% to 2% of all cases with aortic coarctation1,2 but is an important cause of severe hypertension in children. Other vessels, including the renal (>80%) and splanchnic (50%–70%) branches of the aorta can be involved. Extracranial and intracranial cerebrovascular disease has been noted in ≈44% and 31% of affected individuals, respectively.3 Treatment often requires a combination of medical therapies as well as surgical and endovascular interventions.2 Morbidity is high in severe cases, with patients at considerable risk for acute events, such as hypertensive encephalopathy, stroke, progressive cardiac, and renal dysfunction.2
MAS may occur as part of a genetic syndrome such as neurofibromatosis type 1,4 Williams Syndrome,5 Alagille syndrome,6 or as a consequence of a pathological inflammatory disease such as Takayasu’s arteritis.7 However, most cases are idiopathic.3 We here hypothesized that a substantial fraction of cases with MAS may in fact be caused by a single gene (monogenic) mutation. In a previous cohort of patients with chronic kidney disease and the phenotype of increased renal echogenicity, we identified a causative mutation in 63% using whole exome sequencing (WES).8 Additionally, we have shown previously, in the case of congenital anomalies of the kidney and urinary tract (CAKUT), that syndromic CAKUT is caused by truncating mutations, whereas isolated (nonsyndromic) CAKUT is caused by missense mutations in the same monogenic genes.9 We therefore presumed that in a fraction of patients, MAS may be caused by mutations in genes already known to be associated with syndromic MAS and others by mutations in monogenic candidate genes that have not been implicated in MAS, but have been shown to cause vasculopathy.
We generated a list of 38 vasculopathy candidate genes, and we examined 35 families with MAS by WES for disease causing mutations in these candidate genes. We detected likely causal mutations in >40% of individuals with MAS. We then evaluated the mutations detected for genotype–phenotype correlations. Because of the high fraction of patients with MAS in whom a likely causal mutation can be detected and because establishing an etiologic diagnosis may reveal additional genotype/phenotype correlations in the future, we recommend to routinely perform WES in patients with MAS.
Materials and Methods
Study Design and Participants
The data that support the findings of this study are available from the corresponding author on reasonable request. The study was approved by the institutional review board (IRB) of Boston Children’s Hospital. From January 2014 to December 2016, patients with MAS were enrolled after obtaining informed consent. The inclusion criteria were a diagnosis of MAS before 25 years based on evidence of narrowing of the abdominal aorta at the suprarenal level (above the celiac or superior mesenteric arteries), intrarenal level (begins at or above the renal arteries but below the superior mesenteric artery), or infrarenal level (below the renal arteries) by computed tomographic angiography, ultrasound, or other imaging. A total of 36 individuals from 35 different families were enrolled and underwent WES.
WES and Variant Calling
WES and variant burden analysis were performed as previously described.8,10–12 Genomic DNA was isolated from blood lymphocyte or saliva samples and subjected to exome capture using Agilent SureSelect human exome capture arrays (Life technologies), followed by next-generation sequencing on the Illumina HighSeq sequencing platform. Sequence reads were mapped to the human reference genome assembly (NCBI build 37/hg19) using CLC Genomics Workbench (version 6.5.2) software (CLC bio, Aarhus, Denmark). After alignment to the human reference genome (GRCh37/hg19), variants were filtered for likely nondeleterious variants as previously described.10 In the first step, variants with minor allele frequencies >1% in the dbSNP (version 142) or in the 1000 Genomes Project (1094 subjects of various ethnicities; May 2011 data release) databases were excluded, as they were unlikely to be deleterious. Synonymous variants and intronic variants that were not located within splice site regions were excluded. Remaining variants included nonsynonymous variants and splice site variants.
Mutation Calling in Known Vasculopathy Genes
We evaluated WES data for causative mutations in 38 monogenic candidate genes that were known to cause a disease phenotype that included a vasculopathy if mutated (Table S1 in the online-only Data Supplement). The list of 38 candidate genes was generated by querying Online Mendelian Inheritance in Man for the terms aortic stenosis or aortic syndrome. Mutation calling was performed as stated above by filtering remaining variants for changes in the exon and splice regions of the 38 vasculopathy genes. Remaining variants were ranked based on their probable impact on the function of the encoded protein. We considered evolutionary conservation among orthologous genes across phylogeny using ENSEMBL Genome Browser and Clustal Omega. We also used the web-based prediction programs PolyPhen-2,13 Sorting Intolerant From Tolerant,14 and MutationTaster.15 Mutation calling for recessive or dominant disease-causing mutations was performed by clinician scientists/geneticists, who had knowledge of the clinical phenotypes and pedigree structure, as well as experience with homozygosity mapping and WES evaluation, using criteria given in Table S2. Remaining variants were confirmed in patient DNA by Sanger sequencing as previously described.16 A limited number of parental samples were available for segregation. Where available, parental segregation is given in Figures S1 through S4.
As a control, we evaluated WES data of 35 families for 38 known ciliopathy genes, in which only one variant was identified in a known ciliopathy gene. We identified one homozygous call in the gene MKKS. The variant has never been described previously, also the patient phenotypically does not fit Bardet–Biedl syndrome 6 or McKusick–Kaufman syndrome. Also, genetically, the variant is present 2× and 3× in control databases such as EVS and gnomAD, respectively, which makes it unlikely to be the causative variant (Table S3).
Homozygosity Mapping and Linkage Analysis
Prior to 2014, for genome-wide homozygosity mapping, the GeneChip Human Mapping 250k d Array from Affymetrix was used. Alternatively, homozygosity mapping data were generated from WES data. Nonparametric LOD (logarithm of the odds) scores were calculated using a modified version of the program GENEHUNTER2.117,18 through stepwise use of a sliding window with sets of 110 single nucleotide polymorphisms and the program ALLEGRO19 to identify regions of homozygosity as described20,21 using a disease allele frequency of 0.0001 and Caucasian marker allele frequencies. After 2014, to generate homozygosity mapping from WES data, downstream processing of aligned binary alignment map files was done using Picard and SAMtools4.22 Single nucleotide variants calling was performed using Genome Analysis Tool Kit,23 and the generated variant call format file was subsequently used in Homozygosity Mapper.24
If WES evaluation for single nucleotide variants and small insertions/deletions was negative, we performed copy number variant (CNV) analysis using CoNIFER software.25 In particular, we focused on the classic Williams deletion, as MAS may be part of Williams syndrome in the 7q11.23 region.5
UCSC Genome Browser, http://genome.ucsc.edu/cgi-bin/hgGateway
1000 Genomes Browser, http://browser.1000genomes.org
Clustal Omega, http://www.ebi.ac.uk/Tools/msa/clustal
Ensembl Genome Browser, http://www.ensembl.org
Exome Variant Server, http://evs.gs.washington.edu/EVS
Exome Aggregation Consortium, http://exac.broadinstitute.org
HGMD Professional 2016.3, https://portal.biobase-international.com/hgmd
Online Mendelian Inheritance in Man, http://www.omim.org
Sorting Intolerant From Tolerant, http://sift.jcvi.org
Identification of MAS Causing Mutations
We recruited 43 individuals from 42 families with severe hypertension and a suspected diagnosis of MAS from the Center for Midaortic Syndrome and Renovascular Hypertension at Boston Children’s Hospital between January 2014 and December 2016. Seven individuals from 7 families were initially thought to have MAS but ultimately did not have malformations of the aorta and did not meet inclusion criteria. We performed WES in the remaining 36 individuals from 35 families. One sibling, B869_22 did not have narrowing of the aorta but because of the other sibling’s MAS was included in the cohort for genotype and phenotype comparison. We evaluated WES data for mutations in the 38 candidate genes (See Methods and Table S1).
In 15 of the 35 families (42.9%) with MAS, we detected a likely causal mutation in one of the 38 monogenic candidate genes (Figure 1 and Table). The most common gene in which a likely causal mutation was detected was NF1, with 6 families (Figure 1 and Table). Four families were found to have a likely causal mutation in JAG1, 3 families in ELN, and 1 family each in RNF213 and GATA6 (Figure 1 and Table). Each family in whom we detected a likely causal mutation had a unique mutation, for a total of 15 different mutations (Table and Figures S1 through S4). Ten of the mutations were novel and 5 had been described before to cause clinical syndromes (Table S4).
Clinical Description of Individuals in Whom a Likely Causal Mutation Was Identified
The median age at presentation in the individuals in whom a likely causal mutation was detected was 5 years (range 0–12 years). In individuals in whom no causative mutation was detected, median age of onset was 2.25 years (range 0.1–14 years; Tables S5 and S6). Twenty-three of 36 (64%) individuals with MAS were male. Nine of 16 (56%) individuals in whom a likely causal mutation was detected were male. This difference was not statistically significant by χ2 test (P=0.39).
The degree of vessel involvement sorted by monogenic cause is shown in Figures 2 and 4A through 4D. Suprarenal aortic stenosis was the most common location in 5 of 7 individuals with NF1 mutation (Figure 2). In individuals in whom JAG1 mutations were detected, all had suprarenal aortic stenosis (Figure 2). In individuals with ELN mutations, aortic stenosis occurred at the intrarenal level in 2 of 3 individuals (Figure 2). Additional phenotypic information can be found in Table and Tables S5 and S6. Thus, with this size of a cohort, no clear-cut genotype/phenotype correlation was revealed between gene mutation and vessel involvement.
The medical and surgical interventions performed are listed by monogenic cause in Figure 3, Table, and Table S5. All individuals in whom a likely causal mutation was detected required antihypertensive therapy (Table). Given the low number of individuals in any given group, no conclusions on intervention and genotype correlation can be drawn.
WES Detects Mild Mutations in Syndromic Genes in Nonsyndromic Individuals
We have shown previously, in the case of CAKUT, that syndromic CAKUT can be caused by truncating mutations, whereas isolated (nonsyndromic) CAKUT is caused by missense mutations in the same monogenic genes.9,26 We hypothesized that individuals with syndromic or nonsyndromic MAS may have causative mutations in genes associated with genetic syndromes that include a vasculopathy phenotype.9,26 Although the number of cases are small, we did detect a genotype/phenotype correlation where patients with missense mutations had isolated MAS (B632_21, B1195_21, and B725_21), whereas patients with truncating, frameshift, or splice site mutations had syndromic MAS (B1388_21, B869_21, B1550_21, B1087_21, B1220_21, and B586_21), such as neurofibromatosis or Alagille syndrome (Table).
In 5 of 7 individuals (4/6 families) with NF1 mutations, other syndromic features of neurofibromatosis were also present at the time of diagnosis of MAS (Table and Figure S1). However, in 2 individuals in whom we detected NF1 mutations, no other clinical features of neurofibromatosis were evident at the time of presentation. B632_21 developed café au lait spots and neurofibromas after diagnosis with MAS was made and was diagnosed with neurofibromatosis. However, B1195_21 has not had any other symptoms of neurofibromatosis.
Four families (B869, B467, B1388, and B1550) had a clinical diagnosis of neurofibromatosis at the time of presentation with MAS. Three families (B869, B1388, and B1550) had mutations that resulted in frameshift and splice site changes (Table, Figure S1, and Table S5). This evidence is supportive of our hypothesis that protein truncating mutations (stop codon, splice site, and frameshift mutations) may result in MAS associated with additional syndromic features.
Further evidence for this hypothesis is found in 3 of the 4 families with JAG1 mutations (Table, Figure S2, and Table S5). In families B586, B1087, and B1220, the clinical diagnosis of Alagille syndrome had been made prior to the diagnosis of MAS. In all 3 families, protein truncating mutations were detected (Table, Figure S2, and Table S5). One individual (B725) had a missense mutation. This individual has never had and continues to not have any symptoms of Alagille syndrome (Table, Figure S2, and Table S5). Thus, similar to our findings in individuals with mutations in NF1, there is a correlation between mutation type and whether or not an individual has isolated versus syndromic MAS.
In the 3 families with ELN mutation, none had syndromic features of Williams syndrome. In individual B1213_21, more extensive vascular involvement was noted compared with B653_21 and B1462_21, including the pulmonary and iliac arteries (Table, Figure 4C, Figure S3, and Table S5). Interestingly, this individual had a splice site change detected, whereas B653_21 and B1462_21 had missense mutations (Table, Figure 4C, Figure S3, and Table S5). B653_21 had been previously diagnosed with Takayasu’s arteritis prior to WES. This individual had little response to immunosuppression and progression of her vascular abnormalities. In retrospect, this nonresponsiveness may be explained by the newly established diagnosis of ELN mutation (Table, Figure 4C, Figure S3, and Table S5).
For both individuals with RNF213 and GATA6, neither had syndromic features and both had missense mutations (Table, Figure 4D, Figure S4, and Table S5).
CNV and Williams Syndrome
Williams syndrome is characterized by a 1.8 Mb heterozygous deletion that involves several genes, including ELN.5 A selection of genes from the Williams deletion are given in Table S1. To identify heterozygous deletions, we performed CNV analysis. Prior to WES, B385_21, B779_21, and B994_21 had been clinically diagnosed with Williams syndrome (Table S5). A heterozygous Williams deletion was detected on CNV analysis in all 3 cases (data not shown). In additional one case (B1157_21), a Williams deletion was detected, but these individuals have no additional symptoms of Williams syndrome (Table S5).
Summary and Impact of This Work
We here present the first evaluation of the hypothesis that in a cohort of children with MAS, the disease could be caused by mutations in monogenic vasculopathy candidate genes. Previous publications had only reported on syndromic vasculopathies as case reports or small cohorts.6,27 Given the known pleiotropy of developmental genes, we hypothesized that rather than the full syndrome, the specific phenotype of MAS could be caused by allele-specific mutations in the same genes. This phenomenon has previously been observed in genes determined to cause CAKUT, many of which are linked to the development of the urinary system. We detected likely causal mutations in the high fraction of ≈43% of families with MAS in one of 38 vasculopathy genes in both syndromic and nonsyndromic individuals.28 As such, this pleiotropy may be the case in MAS, as many of the genes given here are developmental genes. Additionally, variable expressivity has been seen in the case of monogenic diseases caused by developmental genes, which may account for the degree of phenotypic heterogeneity among individuals with MAS.29 Further investigation into the molecular pathogenesis of MAS is required to elucidate the pathways by which these genes dictate vascular development.
We were able to detect heterozygous Williams deletions in 3 of 20 (15%) families without a monogenic diagnosis. Our data also demonstrate that WES is a valuable tool to provide conclusive molecular genetic diagnosis in families with MAS.
Additionally, we detect a total of 15 different mutations, 10 of which are novel, that is they have not been detected in any patient before. Assignment of causality to a novel variant requires strict adherence to criteria for deleteriousness, including scores from predictive models of pathogenicity and from population databases. We used the criteria given in Table S2 when deciding if a gene was likely causative. Further investigation using molecular biology techniques are needed to determine true pathogenicity. This is the first report of RNF213 and GATA6 mutations in individuals with MAS. Previously, ELN has classically been described to cause pulmonary artery stenosis,30 supravalvular aortic stenosis,31,32 and thoracic aortic and renal artery stenosis.33 RNF213 has been implicated in the development of moyamoya disease but has not previously been described to cause other variants of arterial stenosis.34 GATA6 has been implicated in cardiac outflow tract obstruction, but we detected for the first time pathology in vessels other than the aortic and pulmonary arteries.35
With regards to genotype/phenotype correlation, we note that in individuals with a molecular diagnosis, a protein truncating mutation, MAS, occurs in the setting of syndromic phenotype. This is best demonstrated in JAG1 mutations, in which the 3 families (B1087, B586, and B1220) with Alagille syndrome have mutations that would lead to a truncated protein. In contrast, in the one individual (B725_21) who had no additional symptoms of Alagille syndrome, a mild mutation (missense mutation) was detected of c.1313G>A, p.Cys438Tyr (Table and Figure S2).
Additionally, for the one individual (B1195_21) in whom we detected an NF1 missense mutation with no other features of neurofibromatosis. Thereby, diagnosis of NF1 (neurofibromatosis type 1) for the first time by WES in this individual having a molecular diagnosis will help guide further observation for other symptoms of NF1, including moyamoya disease.36
With regards to predictions of phenotype of the MAS vasculopathy based on genotype, there seems not to be a clear correlation between vessel involvement and need of invasive intervention (Figures 2 and 3 and Table), likely secondary to the small numbers of individuals examined with any given genetic diagnosis. However, we have generated initial evidence of a genotype/phenotype correlation, in which truncating mutations may cause syndromic MAS, whereas missense mutations can cause isolated MAS.
The future direction of medicine is toward personalized care. We show that using WES provides an unequivocal molecular diagnosis in a large proportion of patients with MAS. One limitation of this study is that while it represents the first genetic study exploring WES in children with MAS, the number of individuals with any one genetic diagnosis remains small. Therefore, conclusive genotype/phenotype relations must be viewed as preliminary at this time. While this study was small and no genotype–phenotype correlations were determined regarding gene mutation and best intervention, in the future, such correlations may be able to be determined. Additionally, in identifying such mutations, further understanding of the mechanism by which children develop MAS may be elucidated. With increasing prevalence of WES, genotype/phenotype correlations could be made in the future and impact choices of medical and surgical care in these complex patients.
We thank Leslie Spaneas and Brittany Fisher for their recruitment of patients for this study. We also acknowledge the following physicians who referred participating patients to the Boston Children’s Hospital Center for Midaortic Syndrome and Renovascular Hypertension: A. Chua (Durham, NC), C. Cramer (Rochester, MN), S. El-Dahr (New Orleans, LA), A. Guillot (Burlington VT), R. Holleman (Columbia, SC), N. Jain (New York, NY), M. Joseph (Phoenix, AZ), J. Lohr (Minneapolis, MN), T. Longoria (Tuscon, AZ), D. Magen (Haifa, Israel), R. Raafat (Norfolk, VA), I. Restaino (Norfolk, VA), K. Sanderson (Chapel Hill, NC), E. Simon (Albany, NY), J. Springate (Buffalo, NY).
Sources of Funding
F. Hildebrandt is the William E. Harmon Professor of Pediatrics. This research was supported by grants from the National Institutes of Health to F. Hildebrandt (DK088767) and J.K. Warejko (DK007726-31A1). The Yale Center for Mendelian Genomics is funded by U54 HG006504 granted to R.P. Lifton and by the National Institute of Diabetes and Digestive and Kidney Diseases Intramural Research Program to J.K. Warejko. Funding to W. Tan through the American Society of Nephrology Benjamin J. Lipps Research Fellowship Award (FP01014311) and DK007726-31A1. A. Vivante is supported by the Manton Center for Orphan Diseases Research grant.
F. Hildebrandt has intellectual property licensed to Claritas and is a cofounder of Goldfinch Bio. The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.117.10296/-/DC1.
- Received September 7, 2017.
- Revision received October 6, 2017.
- Accepted January 18, 2018.
- © 2018 American Heart Association, Inc.
- Sen PK,
- Kinare SG,
- Engineer SD,
- Parulkar GB
- Kohl S,
- Hwang DY,
- Dworschak GC,
- Hilger AC,
- Saisawat P,
- Vivante A,
- Stajic N,
- Bogdanovic R,
- Reutter HM,
- Kehinde EO,
- Tasic V,
- Hildebrandt F
- Sadowski CE,
- Lovric S,
- Ashraf S,
- et al
- Van der Auwera GA,
- Carneiro MO,
- Hartl C,
- Poplin R,
- Del Angel G,
- Levy-Moonshine A,
- Jordan T,
- Shakir K,
- Roazen D,
- Thibault J,
- Banks E,
- Garimella KV,
- Altshuler D,
- Gabriel S,
- DePristo MA
- Krumm N,
- Sudmant PH,
- Ko A,
- O’Roak BJ,
- Malig M,
- Coe BP,
- Quinlan AR,
- Nickerson DA,
- Eichler EE
- Vivante A,
- Hwang DY,
- Kohl S,
- Chen J,
- Shril S,
- Schulz J,
- van der Ven A,
- Daouk G,
- Soliman NA,
- Kumar AS,
- Senguttuvan P,
- Kehinde EO,
- Tasic V,
- Hildebrandt F
- Veean S,
- Thakkar N,
- Gupta S,
- Keshavamurthy J
- van der Ven AT,
- Vivante A,
- Hildebrandt F
- Kamath BM,
- Bauer RC,
- Loomes KM,
- et al
- Kodo K,
- Nishizawa T,
- Furutani M,
- Arai S,
- Yamamura E,
- Joo K,
- Takahashi T,
- Matsuoka R,
- Yamagishi H
- Santoro C,
- Di Rocco F,
- Kossorotoff M,
- et al
Novelty and Significance
What Is New?
This is the first study to examine the use of whole exome sequencing in individuals with both syndromic and nonsyndromic midaortic syndrome and renovascular hypertension.
What Is Relevant?
Midaortic syndrome is a rare cause of severe hypertension in children who often requires multidisciplinary care. Whole exome sequencing provides an unequivocal molecular diagnosis in a large proportion of patients with syndromic and nonsyndromic midaortic syndrome, which may help guide therapy in the future.
Diagnostic whole exome sequencing should be offered to patients with midaortic syndrome, as therapy may be tailored to the genotype of the individual with further research.