Individuals with high-risk spontaneous coronary artery dissection (SCAD) were selected for sequencing and analyzed: (1) for genetic variation in 22 candidate genes and (2) all genes (genome wide). We identified 3 variants in the candidate genes that were annotated as pathogenic or likely pathogenic and therefore were clinically actionable. Variants in vascular connective tissue disease (CTD), genome-wide association study (GWAS)–prioritized genes, COL3A1, and Loeys-Dietz syndrome (LDS) genes were enriched, as compared with the Genome Aggregation Database (gnomAD). Aggregated rare variant association tests were performed genome wide for all genes, comparing patients with SCAD to healthy controls in an overall analysis of all cases and controls, and 3 subgroups. CanSCAD indicates the Canadian SCAD Registry; FHx, family history; P-SCAD, peripartum SCAD; R-SCAD, recurrent SCAD.
A, Distribution of samples with variants in genome-wide association study (GWAS)–prioritized genes and vascular connective tissue disease (CTD) genes. Variants were identified in 6 of 94 samples (6.4%) and 16 of 94 samples (17.0%) from GWAS-prioritized genes or vascular CTD genes, respectively. Individuals harboring multiple variants are labeled in Table 2 (footnotes d-g). The proportion of findings for all variants in the study is shown, and the light blue and dark blue indicate samples with variants from vascular CTD genes and GWAS-prioritized genes, respectively. B, Forest plots for gene set and gene findings. Case and Genome Aggregation Database (gnomAD): total number of variants in individuals with high-risk SCAD and gnomAD. The total sample sizes of individuals with high-risk SCAD and gnomAD were 188 and 251 496 respectively, by multiplying 2 chromosomes of sample size. Forest plots demonstrate the odds ratio (OR) and the 95% CIs inferred by the normal approximation method. P value was calculated implementing a Fisher exact test by comparing the number of variants identified in the case group and gnomAD.
aIndicates gene sets.
eTable 1. Genes Prioritized for Annotation
eTable 2. Enrichment Analysis in High-Risk SCAD Cases as Compared With gnomAD
eTable 3. Gene-Based Genome-Wide Aggregated Variant Association Testing
eTable 4. Variants in SEC31A and SCARF1 in P-SCAD Cases
eFigure 1. Quantile-Quantile Plot of VT Aggregated Variant Association Test of P-SCAD
eFigure 2. GTEx Human Tissue Expression of P-SCAD–Associated Genes
Customize your JAMA Network experience by selecting one or more topics from the list below.
Wang Y, Starovoytov A, Murad AM, et al. Burden of Rare Genetic Variants in Spontaneous Coronary Artery Dissection With High-risk Features. JAMA Cardiol. 2022;7(10):1045–1055. doi:10.1001/jamacardio.2022.2970
Do individuals presenting with high-risk spontaneous coronary artery dissection (SCAD) features have an increased burden of rare genetic variants?
In a whole-exome sequencing genetic study, 17% of individuals with high-risk SCAD had rare variants in vascular connective tissue disease genes, representing a significant enrichment compared with the Genome Aggregation Database (gnomAD), especially in COL3A1 and Loeys-Dietz syndrome genes. Rare variants in genes recently discovered by genome-wide association study, ADAMTSL4 and LRP1, identified in 6% of the cohort, were also enriched.
In this cohort, rare genetic variants were identified in approximately 1 in 5 individuals with SCAD with high-risk features; genetic testing may be considered in affected individuals.
The emerging genetic basis of spontaneous coronary artery dissection (SCAD) has been defined as both partially complex and monogenic in some patients, involving variants predominantly in genes known to underlie vascular connective tissue diseases (CTDs). The effect of these genetic influences has not been defined in high-risk SCAD phenotypes, and the identification of a high-risk subgroup of individuals may help to guide clinical genetic evaluations of SCAD.
To identify and quantify the burden of rare genetic variation in individuals with SCAD with high-risk clinical features.
Design, Setting, and Participants
Whole-exome sequencing (WES) was performed for subsequent case-control association analyses and individual variant annotation among individuals with high-risk SCAD. Genetic variants were annotated for pathogenicity by in-silico analysis of genes previously defined by sequencing for vascular CTDs and/or SCAD, as well as genes prioritized by genome-wide association study (GWAS) and colocalization of arterial expression quantitative trait loci. Unbiased genome-wide association analysis of the WES data was performed by comparing aggregated variants in individuals with SCAD to healthy matched controls or the Genome Aggregation Database (gnomAD). This study was conducted at a tertiary care center. Individuals in the Canadian SCAD Registry genetics study with a high-risk SCAD phenotype were selected and defined as peripartum SCAD, recurrent SCAD, or SCAD in an individual with family history of arteriopathy.
Main Outcomes and Measures
Burden of genetic variants defined by DNA sequencing in individuals with high-risk SCAD.
This study included a total of 336 participants (mean [SD] age, 53.0 [9.5] years; 301 female participants [90%]). Variants in vascular CTD genes were identified in 17.0% of individuals (16 of 94) with high-risk SCAD and were enriched (OR, 2.6; 95% CI, 1.6-4.2; P = 7.8 × 10−4) as compared with gnomAD, with leading significant signals in COL3A1 (OR, 13.4; 95% CI, 4.9-36.2; P = 2.8 × 10−4) and Loeys-Dietz syndrome genes (OR, 7.9; 95% CI, 2.9-21.2; P = 2.0 × 10−3). Variants in GWAS-prioritized genes, observed in 6.4% of individuals (6 of 94) with high-risk SCAD, were also enriched (OR, 3.6; 95% CI, 1.6-8.2; P = 7.4 × 10−3). Variants annotated as likely pathogenic or pathogenic occurred in 4 individuals, in the COL3A1, TGFBR2, and ADAMTSL4 genes. Genome-wide aggregated variant testing identified novel associations with peripartum SCAD.
Conclusions and Relevance
In this genetic study, approximately 1 in 5 individuals with a high-risk SCAD phenotype harbored a rare genetic variant in genes currently implicated for SCAD. Genetic screening in this subgroup of individuals presenting with SCAD may be considered.
Spontaneous coronary artery dissection (SCAD) is a nonatherosclerotic cause of myocardial infarction, typically in young women, for which both complex genetic and monogenic influences have been recently defined.1,2 Current estimates are that approximately 5% of all patients with SCAD have a monogenic etiology involving genes that have been previously implicated in vascular connective tissue diseases (CTDs),3,4 including genes underlying vascular Ehlers-Danlos syndrome (attributable to pathogenic variation in COL3A1),5 Marfan syndrome (FBN1),6 Loeys-Dietz syndrome (LDS) (TGFB2, TGFB3, TGFBR1, TGFBR2, SMAD2, SMAD3),7,8 and fibrillar collagens (COL3A1, COL5A1).9 Fibromuscular dysplasia (FMD) is the most common underlying arterial disease association, identified in 25% to 86% of individuals with SCAD.1 Multifocal FMD is defined angiographically by multiple arterial stenosis within an artery with intervening mural dilations10; histologic findings include medial fibroplasia with excess extracellular matrix accumulation and smooth muscle cell disorganization.11,12 Recently reported genetic associations of rare variants for multifocal FMD include PTGIR and COL5A1.13,14
Previous whole-exome sequencing (WES) studies of SCAD have defined variants in genes known to underlie monogenic arteriopathies13-19 and preliminary findings for TLN1, TSR1, PTGIR, and PKD1.13-18 Common variant associations with SCAD include several genetic loci with relatively high-effect estimates, with odds ratios (ORs) of approximately 1.5 to 2.0.20-22 By pairing genome-wide association study (GWAS) results with arterial transcriptome data from the Genotype-Tissue Expression (GTEx) containing several hundred human arterial samples, expression quantitative trait locus colocalization analyses demonstrated strong evidence for 1 gene in each of the top 3 GWAS-identified loci: ADAMTSL4, PHACTR1, and LRP1.21
We conducted a WES study to test the hypothesis that low frequency, deleterious coding genetic variants in known and recently proposed sequencing-determined genes for SCAD, as well as GWAS-prioritized genes, are enriched in high-risk SCAD.
The Canadian SCAD registry (CanSCAD) prospectively recruited 750 patients with nonatherosclerotic SCAD from May 15, 2014, to August 14, 2018, of which most were women (88.5%) and self-reported White race (87.7%).21 Patients with SCAD with high-risk features were selected from the CanSCAD genetic substudy for sequencing, including peripartum SCAD (P-SCAD),23 recurrent SCAD (R-SCAD), or SCAD in an individual with a family history of arteriopathy defined as arterial aneurysm, dissection, or FMD. Healthy controls were selected from the Michigan Genomics Initiative (MGI) biorepository. As this was an exploratory observational registry of a newly described condition, it was essential to examine prevalence in all populations, therefore, race and ethnicity data were included. The participants were asked how they identified themselves. The following races and ethnicities were included: African Canadian, East Asian, First Nation, South Asian, White, and other. Other race and ethnicity included mixed ethnicities, unknown (adopted), and those who declined to answer. All participants provided written informed consent, and study activities received institutional review board approval. A detailed included/excluded criterion was included in the eMethods in the Supplement. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
Genomic DNA was isolated from blood or saliva samples as previously described,21 and was sequenced (NovaSeq 6000 [Illumina]) after tagmentation, polymerase chain reaction amplification, gel purification, and exome capture (SureSelectV5 platform [Agilent]). After aligning to GRCh37(hg19) with bwa-0.7.17 mem,24 single-nucleotide variants were called with the GATK(22.214.171.124) best practices.25 Variants were annotated in 22 genes implicated in vascular CTD diseases (COL3A1, COL5A1, COL5A2, FBN1, FLNA, LMX1B, LOX, MYH11, NOTCH1, PKD1, PTGIR, SMAD2, SMAD3, TGFB2, TGFB3, TGFBR1, TGFBR2, TLN1, and TSR1) and SCAD-GWAS genes: (ADAMTSL4, LRP1, and PHACTR1)1,3,4,13,14,17-19,21 (eTable 1 in the Supplement), if meeting (1) nonsynonymous variant with CADD Phred score, version 1.3, greater than 20 and Genome Aggregation Database (gnomAD), version 2.1, allele frequency less than 0.1%; or (2) loss of function (LoF) variants with gnomAD allele frequency less than 0.1%; or (3) likely pathogenic (LP) or pathogenic (P) variants by any contributors in the ClinVar database (accessed April 21, 2022) where LoF was defined as start-lost, stop-loss, stop-gain, frameshift indels, or splice-site variants.26,27 Finally, pathogenicity of variants was estimated according to the 2020 Association for Clinical Genomic Science (ACGS) and American College of Medical Genetics and Genomics (ACMG) criteria.28,29
We tested for enrichment of variants in candidate genes and gene sets (eTable 1 in the Supplement) in individuals with SCAD by comparison with the gnomAD reference database (version 2.1) using a 2-sided Fisher exact test. The significance threshold for gene-wise analysis used a Bonferroni-corrected significance threshold for the 22 genes tested (0.05 / 22 = .002); the significance threshold for gene set analysis was .05 / 4 = .013 because 4 gene sets were tested using a Bonferroni correction accounting for the number of tests performed.
To discover new genes, we conducted genome-wide association tests of variants aggregated by genes with EPACTS software, version 3.2.6 (University of Michigan).30 High-quality variants (as defined in the Supplement) were prioritized to analyze the genetic burden in the 94 individuals with SCAD compared with age-, sex-, and ancestry-matched MGI controls, as well as 3 subgroups:(1) P-SCAD; (2) R-SCAD; (3) SCAD cases with family history. The significance threshold was P ≤ 4 × 10−6 after Bonferroni correction.
False-discovery rates (FDRs) were estimated randomizing the case and control phenotype designations and rerunning the burden test 10 times.31 FDR of .05 or less was the significance threshold. Detailed methods are included in the eMethods of the Supplement.
The CanSCAD registry collected DNA from 336 individuals (mean [SD] age, 53.0 [9.5] years; 301 female participants [90%]; 35 male participants [10%]) with nonatherosclerotic SCAD. Participants identified with the following race and ethnic groups: 1 African Canadian (0.3%), 33 Asian (9.8%), 294 White (87.5%), and 4 other (1.1%). Criteria for a high-risk SCAD phenotype were met by 94 individuals, including 8 P-SCAD (2%), 33 R-SCAD (10%), and 65 individuals (19%) with family history of arteriopathy as defined (Figure 1). The demographics and clinical characteristics of both sequenced (WES) and nonsequenced (non-WES) samples are summarized in Table 1. The control group consisted of 282 age-, gender-, and ancestry-matched healthy individuals selected from MGI (mean [SD] age, 59.3 [11.6] years; 88 female controls [31%]). A total of 170 MGI control individuals (60%) had hypertension.
Case (n = 94) and MGI control (n = 282) samples were sequenced in different batches using consistent reagents. The mean target region read depth was 79X (range, 59-321); 7 122 669 single-nucleotide variations were called. No related individuals were found by identity-by-descent analysis (PI_HAT <0.2).
In vascular CTD genes (COL3A1, COL5A1, COL5A2, FBN1, FLNA, LMX1B, LOX, MYH11, NOTCH1, PKD1, PTGIR, SMAD2, SMAD3, TGFB2, TGFB3, TGFBR1, TGFBR2, TLN1, and TSR1) and GWAS-prioritized genes (ADAMTSL4, LRP1, and PHACTR1), we identified 27 variants in the high-risk SCAD cohort, where 4 variants observed in the MGI healthy controls were removed from all following analysis. In total, 23 variants in 13 of the 22 prioritized genes were identified (Table 2, Figure 2A). Most of the vascular CTD genes are autosomal dominant (AD) or X-linked gene in the ClinGen database (https://clinicalgenome.org/), with 3 recently identified SCAD-associated genes with unknown mechanism: PTGIR, TLN1, and TSR1 (eTable 1 in the Supplement).
Variants across all genes of interest, including all vascular CTD and GWAS-prioritized genes, were significantly enriched in the high-risk SCAD cohort (n = 94) compared with the gnomAD exome reference database (n = 125 748), with an OR of 2.8 (95% CI, 1.8-4.3; 12.2% individual cases [23 of 188] vs 4.4% gnomAD [11 117 of 251 496]; P = 3.7 × 10−5) (Figure 2B; eTable 2 in the Supplement). In a secondary analysis separately testing for enrichment of variants in vascular CTD and GWAS-prioritized genes, variants in vascular CTD genes were identified in 17.0% of individuals (16 of 94) with high-risk SCAD. Variants in GWAS-prioritized genes were observed in 6.4% of individuals (6 of 94) with high-risk SCAD. There was an enrichment of variants in vascular CTD genes (OR, 2.6; 95% CI, 1.6-4.2; 9.0% individual cases [17 of 188] vs 3.5% gnomAD [8916 of 251 496]; P = 7.8 × 10−4) and GWAS-prioritized genes (OR, 3.6; 95% CI, 1.6-8.2; 3.2% individual cases [6 of 188] vs 0.9% gnomAD [2201 of 251 496]; P = 7.4 × 10−3). COL3A1 (OR, 13.4; 95% CI, 4.9-36.2; P = 2.8 × 10−4) showed a significant enrichment of variants in the high-risk SCAD cohort compared with gnomAD, whereas LRP1 (OR, 3.7; 95% CI, 1.4-10.1; P = 2.5 × 10−2) and TGFBR2 (OR, 16.0; 95% CI, 3.9-65.1; P = 7.4 × 10−3) were only nominally enriched after Bonferroni correction for multiple-hypothesis testing. As an additional subset analysis, we evaluated LDS genes (TGFBR1, TGFBR2, SMAD2, SMAD3, TGFB2, TGFB3) as an aggregated group which was previously reported to be enriched in individuals with SCAD.19 We validated this result in our analysis of high-risk SCAD as compared with the gnomAD reference (OR, 7.9; 95% CI, 2.9-21.2; 2.1% individual cases [4 of 188] vs 0.3% gnomAD [680 of 251 496]; P = 2 × 10−3). In summary, variants were significantly enriched in the aggregated group of vascular CTD and GWAS-prioritized genes together, and the subsets of vascular CTD, GWAS-prioritized, and LDS genes, after Bonferroni correction for multiple-hypothesis testing.
In order to identify specific genes driving the gene-set associations, we performed gene-based comparisons of variants in individuals with high-risk SCAD compared with gnomAD. COL3A1, LRP1, and TGFBR2 were the leading genes with the lowest P value in gene set of vascular CTD, GWAS-prioritized, and LDS genes, respectively.
In genes previously implicated in vascular CTDs and/or SCAD, we identified in our high-risk SCAD cohort 17 variants in COL3A1 (n = 4), COL5A1 (n = 1), COL5A2 (n = 2), LOX (n = 1), MYH11 (n = 2), NOTCH1 (n = 1), TGFBR1 (n = 1), TGFBR2 (n = 2), TGFB2 (n = 1), and TLN1 (n = 2). As described previously, variants in COL3A1 (n = 4) were significantly enriched compared with gnomAD (eTable 2 in the Supplement), including 2 LoF variants and 2 nonsynonymous variants. The first LoF variant, COL3A1 c.283-1G>A, is a splicing acceptor site, resulting in in-frame skipping of exon 3 (codon 95-111), whereas COL3A1 c.3898G>T(p.Glu1300Ter) is predicted to delete the last 2 exons and lead to nonsense-mediated decay (167 amino acids, 11% of COL3A1 protein).32 Both LoF variants were not reported in gnomAD or ClinVar (accessed April 21, 2022) and were identified in individuals with P-SCAD. Two additional COL3A1 nonsynonymous variants were c.923G>A(p.Arg308Gln) and c.3061C>A(p.Leu1021Ile). All COL3A1 variants except the last one were annotated as LP or P in Varsome (Table 2).33 Combined with other fibrillar collagens, 7 variants were identified from 3 fibrillar collagen genes (COL3A1 [n = 4]; COL5A1 [n = 1]; COL5A2 [n = 2]), consistent with a recent report of fibrillar collagen variants in SCAD.9
We validated the previously reported enrichment of deleterious variants across 6 genes underlying LDS (TGFBR1, TGFBR2, SMAD2, SMAD3, TGFB2, and TGFB3)19 in individuals with high-risk SCAD compared with gnomAD. Of all 4 variants in LDS genes (TGFBR2 [n = 2]; TGFB2 [n = 1]; TGFBR1 [n = 1]) (Table 2), TGFBR2 c.1591G>A(p.Ala531Thr) was a recurrent variant identified in 2 unrelated probands with family history of dissection (n = 1) and aneurysm (n = 1). Both individuals had normal aortic root z scores (−0.41), and one had a confirmed diagnosis of LDS. This variant was annotated as likely pathogenic in ClinVar (identification [ID] = 165399) with segregation evidence for thoracic aortic aneurysm and dissection and/or features of LDS. In Varsome, the variant was annotated as pathogenic (https://varsome.com/variant/hg19/rs727503477). Three additional variants from LDS genes included TGFBR2 c.1159G>T(p.Val387Leu), TGFB2 c.368A>G(p.Tyr123Cys), and TGFBR1 c.1285T>C(p.Tyr429His).
In TLN1, 2 variants were identified: c.4721C>T(p.Ala1574Val) and c.7070G>A(p.Ser2357Asn), with gnomAD frequencies of 0 and 4 × 10−6 and CADD scores of 24 and 26, respectively. One variant in COL5A1 c.850G>A(p.Glu284Lys) was identified in an individual with FMD and dissection, consistent with the previously reported association of COL5A1 variants with arterial dissections in FMD.13 The COL5A1 c.1540G>A(p.Gly514Ser) variant that has been reported for a phenotype characterized by arterial dissections and multifocal FMD was not observed in our study.13 Additional variants were identified in COL5A2, LOX, MYH11, and NOTCH1, whereas FBN1, FLNA, LMX1B, PKD1, PTGIR, SMAD2, SMAD3, TGFB3, and TSR1 had no findings (Table 2).
Recent GWAS efforts have identified novel associations with SCAD and prioritized ADAMTSL4, LRP1, and PHACTR1 from associated genetic loci, through integrative analyses of DNA sequence and GTEx arterial RNASeq-generated transcriptomes.21 We identified the enrichment of GWAS-prioritized genes in our high-risk SCAD cohort with an OR of 3.6 (95% CI, 1.6-8.2), as compared with gnomAD. In total, we identified 6 variants in ADAMTSL4 (n = 2) and LRP1 (n = 4) (Figure 2, Table 2). One insertion, ADAMTSL4 c.2270dupG(p.Gly758Trpfs*59), is predicted to cause a premature termination codon (PTC) and nonsense-mediated decay (deleted and loss of 259 amino acids, 24% of protein).32 Two pathogenic reports were found for this variant in ClinVar (ID = 39559), one with isolated ectopia lentis et pupillae (OMIM = 225200)34; and the other is unknown. Prompted by the genetic finding, the individual in our study underwent ophthalmologic evaluation that identified a narrowed eye outlet considered a preglaucoma phenotype requiring corrective surgery, a phenotype that has also been reported in association with other ADAMTS gene variants.35-37 ADAMTSL4 c.1249C>T(p.Arg417Cys) has no ClinVar record. Variants in LRP1 (n = 4) were nominally enriched in high-risk SCAD (eTable 2 in the Supplement). All were nonsynonymous variants with no ClinVar records.
In total, we identified 23 genetic variants in 20 individuals, including 17 from candidate SCAD genes and 6 from GWAS-prioritized genes. Individuals harboring identified variants tended to have a higher proportion of FMD, although not statistically significant compared with individuals with high-risk SCAD and no variants (OR, 2.1; 95% CI, 1.0-4.6; P = .07) (Table 2).
According to the ACMG/ACGS 2020 criteria, 3 identified variants were annotated as LP or P in 4 individuals with SCAD, including ADAMTSL4 c.2270dupG(p.Gly758Trpfs*59), recurrent TGFBR2 c.1591G>A(p.Ala531Thr), and COL3A1 c.3898G>T(p.Glu1300Ter) (Table 2), where the first 2 variants were annotated as LP/P in ClinVar. Both ADAMTSL4 c.2270dupG(p.Gly758Trpfs*59) and COL3A1 c.3898G>T(p.Glu1300Ter) were predicted to delete more than 10% of the gene’s protein product and cause nonsense-mediated decay and therefore haploinsufficiency.32 The splicing acceptor variant in COL3A1 (c.283-1G>A) was annotated as a variant of uncertain significance (VUS) owing to the unclear function of the skipped exon, although it was annotated as likely pathogenic in Varsome (https://varsome.com/variant/hg19/2%3A189849922%3AG%3AA). All variants identified in the current study were not reported in previous SCAD genetic studies.3,4,9,15,19
In a genome-wide case-control analysis of all individuals with SCAD (n = 94) compared with MGI controls (n = 282), no genes met the genome-wide significance threshold (P ≤ 4 × 10−6) (eTable 3 in the Supplement). In the subgroup analyses of P-SCAD (n = 8), R-SCAD (n = 33), and SCAD with a family history of arteriopathy findings (n = 65), each compared with MGI controls (n = 282), we identified 2 novel associations with P-SCAD meeting genome-wide significance (genomic inflation factor λ = 1.05) (eFigure 1 in the Supplement): SEC31A (z score, −10.2; P = 6.5 × 10−25), and SCARF1 (z score, −8.3; P = 4.6 × 10−17), which require further validation in larger studies (eTables 3 and 4 and eFigure 2 in the Supplement).
In this WES study of high-risk SCAD in a prospective SCAD registry, we identified several genetic associations with translational significance: (1) 17 variants were identified in 16 individuals from previously sequencing-defined genes, inclusive of genes considered when performing clinical genetic testing for SCAD, representing 17% of the high-risk SCAD cases in the current study, with 6 additional variants in 6 individuals from GWAS-prioritized genes, representing an additional 6% of the high-risk SCAD cases and with relevant gene sets demonstrating an enrichment of variants; (2) variants in COL3A1 were significantly enriched, and COL3A1 LoF variants were identified in 2 of 8 individuals with P-SCAD; (3) LDS gene variants were significantly enriched, consistent with a prior report,19 with a newly identified recurrent variant in TGFBR2; (4) among GWAS-prioritized genes, pathogenic variation was identified in ADAMTSL4, and variants in LRP1 were enriched; and (5) novel gene-based associations with P-SCAD were identified and will require follow-up analysis in new samples.
In the current study of high-risk SCAD, approximately 1 in 5 individuals harbored a variant in the studied genes, and approximately 1 in 6 individuals harbored a variant in a currently defined genes considered in clinical gene testing for SCAD. We observed a significant enrichment of variants across all genes studied and the subsets of vascular CTD genes and GWAS-prioritized genes, especially in COL3A1 and LDS genes (TGFB2, TGFB3, TGFBR1, TGFBR2, SMAD2, SMAD3)13,15-19 but not SMAD2 individually as reported previously.19 The enrichment of variants in LDS genes was higher than previously reported19 (OR of 7.9 in the current study vs 3.019). Notably, most individuals found to have a variant in COL3A1 or an LDS gene had no prior clinical diagnosis of vascular Ehlers-Danlos syndrome or LDS nor any evidence of associated features such as a dilated aortic root on echocardiogram, as was previously reported in a general SCAD cohort as well,19 suggesting the possibility of a different phenotype of the identified LDS gene variants. The finding of COL3A1 LoF variants in 25% of the small subset of P-SCAD is noteworthy.
The observation that approximately 1 in 6 (approximately 17%) individuals with high-risk SCAD harbored variants from previously reported genes for vascular CTD and SCAD suggests that expanded clinical screening may have utility in individuals presenting with SCAD with high-risk features. However, most variants were annotated as VUS, and whether these variants are pathogenic or modifiers of an underlying arterial predisposition to SCAD remains to be determined. The recent finding that COL5A1 variants13 are associated with increased risk of arterial dissections among those with FMD supports a potential mechanism of rare and low-frequency genetic variants acting as gene modifiers. The relatively high prevalence of FMD in high-risk SCAD in the current study supports further consideration of this possibility.
Recent GWAS have identified novel associated genes prioritized by colocalization of expression quantitative trait loci, chiefly ADAMTSL4, LRP1, and PHACTR1.21 The index risk-conferring single-nucleotide variation (1q21.2) regulating ADAMTSL4 expression had an allele frequency of 0.26 and a relatively high OR (1.8) for a common variant association with human disease. ADAMTSL4 coding variants may cause recessive ectopia lentis35 owing to deleterious effects on microfibril structure,38 fibrillin microfibril, and ocular zonule assembly,39,40 but whether coding variants are relevant for SCAD has not been previously considered. Our finding of a pathogenic ADAMTSL4 variant with an actionable eye disease raises the possibility of an eye phenotype association with SCAD that has not previously been considered. ADAMTSL4 is expressed in the artery by medial smooth muscle cells,21,39 and binds to fibrillin 139 to promote microfibril formation. Fibrillin 1 is encoded by FBN1, in which pathogenic variants underlie Marfan syndrome and ectopia lentis38 and have been reported for sporadic cases of SCAD.3 Paralogues of ADAMTSL4, ADAMTSL6 (alias of THSD4), and ADAMTSL2 have been described in either aortic aneurysm,41 or probands with features of autosomal dominant connective tissue disorders,42 with similar functional roles to bind with fibrillin 1 and promote microfibril assembly.43,44 Combining evidence from unbiased GWAS, proband phenotypic correlation, functional roles in extracellular matrix microfibril formation, localization to the arterial tunica media vascular smooth muscle cells,21,39 support the consideration of ADAMTSL4 as a candidate gene for SCAD.
LRP1 encodes a cell membrane-associated protein that mediates protein endocytosis45 or signaling pathways activation.46 Pleiotropic common variant associations have been reported at the SCAD risk locus for LRP1 (OMIM = 107770) with carotid artery dissection, abdominal aortic aneurysm,47 and FMD.48 In a murine model, the expression of LRP1 in arterial smooth muscle cells affected cell migration and the integrity of the vascular wall.49,50 Alleles associated with SCAD are expected to increase the expression of LRP1.51 Coding variants in LRP1 have been reported for keratosis pilaris atrophicans52 and aortic dissections.53 Among recently proposed novel genes for SCAD based upon sequencing studies, TLN1, TSR1, PTGIR, and PKD1, although not meeting statistical criteria for association, the occurrence of variants in TLN1 in 2 individuals supports the potential relevance of TLN1 in SCAD.
Finally, we identified novel associations of P-SCAD with SEC31A and SCARF1, both genes that have not previously been implicated in arterial diseases. SEC31A has no known vascular role but is highly expressed in human arteries.51 SCARF1 is the acetyl low-density lipoprotein receptor, a scavenger receptor expressed in endothelial cells and regulates the uptake of chemically modified low-density lipoproteins, with a largely understudied function.54 Statistically, both genes meet P value and FDR criteria but given the very small sample size of the analysis, will require replication.
Limitations of our study included modest sample size, homogenous population, and power for discovery case-control association approaches, particularly for a disease with locus and allelic heterogeneity. By focusing on individuals with high-risk SCAD, we have not defined the frequency of variants in unselected SCAD cases. Variants prioritized through in-silico analysis are based on genetic criteria, which is not equivalent to being clinically actionable. Resolving the VUSs identified in this study will require further familial segregation studies and/or experimental data. Further functional validation of coding variants in GWAS-prioritized genes ADAMTSL4 and LRP1, will be needed and is warranted, based upon the observed phenotypes in our study and mechanistic support of roles maintaining integrity of the arterial wall and extracellular matrix.
In summary, in this genetic sequencing study, we identified rare coding variants in known and new genes that are associated with high-risk SCAD phenotypes. The findings highlight genes underlying vascular CTDs, chiefly those related to vascular Ehlers-Danlos syndrome and LDS, and newly implicate rare variants in the GWAS-prioritized genes ADAMTSL4 and LRP1. We identified several additional VUSs through annotations of the queried genes. Although monogenic variants in vascular CTD genes have been previously understood to affect approximately 5% of individuals in general SCAD cohorts, the current report supports a much higher frequency (approximately 17%) among individuals with SCAD and high-risk features. These findings provide new biologic insights into the genetic basis of SCAD and support clinical genetic testing considerations for individuals with high-risk SCAD presentations.
Accepted for Publication: June 24, 2022.
Published Online: September 14, 2022. doi:10.1001/jamacardio.2022.2970
Corresponding Author: Santhi K. Ganesh, MD, Division of Cardiovascular Medicine, Department of Internal Medicine, University of Michigan Medical School, 1150 W Medical Center Dr MSRB 3 Room 7220A, Ann Arbor, MI 48109-0644 (email@example.com); Jacqueline Saw, MD, Division of Cardiology, Vancouver General Hospital, University of British Columbia, 2775 Laurel St, Level 9, Vancouver, BC, Canada (firstname.lastname@example.org).
Author Contributions: Drs Ganesh and Saw had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Wang, Brunham, Li, Ganesh.
Acquisition, analysis, or interpretation of data: Wang, Starovoytov, Murad, Hunker, Li, Saw, Ganesh.
Drafting of the manuscript: Wang, Hunker, Li, Saw, Ganesh.
Critical revision of the manuscript for important intellectual content: Wang, Starovoytov, Murad, Brunham, Saw, Ganesh.
Statistical analysis: Wang, Li, Ganesh.
Obtained funding: Saw, Ganesh.
Administrative, technical, or material support: Wang, Starovoytov, Hunker, Brunham, Saw, Ganesh.
Supervision: Saw, Ganesh.
Conflict of Interest Disclosures: Dr Wang reported receiving a National Heart, Lung, and Blood Institute training grant outside the submitted work. Ms Murad reported receiving personal fees from Concert Genetics as a paid consultant outside the submitted work. Dr Brunham reported receiving personal fees from Amgen, Sanofi, HLS Therapeutics, and Novartis for serving on their advisory boards and having a patent pending for a genetic risk predictor of SCAD. Dr Saw reported receiving personal fees from Abbott, Boston Scientific, and Baylis for working as a consultant outside the submitted work, being a noncompensated member of the scientific advisory board of the spontaneous coronary artery dissection (SCAD) Alliance, receiving consultant and advisory board honoraria from Abbott Vascular, Boston Scientific, Baylis, and Gore, and serving as a proctor for Boston Scientific and Abbott Vascular. Dr Ganesh reported receiving grants from the National Heart, Lung, and Blood Institute, Heart and Stroke Foundation of Canada, Canadian Institutes of Health Research, Frankel Cardiovascular Center, and University of Michigan A. Taubman Institute during the conduct of the study; having a patent pending (The University of Michigan and University of British Columbia have filed for a patent) on a genetic risk predictor for SCAD; being a noncompensated member of the medical advisory board of the FMD Society of America; and being a noncompensated member of the scientific advisory board of the SCAD Alliance. No other disclosures were reported.
Funding/Support: This work was supported in part by grant T32-HL007853 from the US National Institutes of Health (Dr Wang); grants R01HL139672 and R35HL161016, Bethesda, Maryland, from the National Heart, Lung, and Blood Institute; grant G-17-0016340, Ottawa, Ontario, Canada, from the Heart and Stroke Foundation of Canada; grant 136799, Ottawa, Ontario, Canada, from the Canadian Institutes of Health Research; the Frankel Cardiovascular Center M-BRISC program; and the University of Michigan A. Taubman Institute, Ann Arbor, Michigan.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We thank the University of Michigan Precision Health Initiative and Medical School Central Biorepository for providing biospecimen storage, management, processing, and distribution services; the Center for Statistical Genetics in the Department of Biostatistics at the School of Public Health for the Michigan Genomics Initiative genotype data curation and management in support of this research; the Vancouver SCAD Conference organizers for enabling study enrollments at patient meetings; and all of the study participants. No participants received financial compensation. The investigators received research and salary support to conduct the research.