Fifty generations of Icelanders help to explain a cause of aortic coarctation

This editorial refers to ‘A rare missense mutation in MYH6 associates with non-syndromic coarctation of the aorta’†, by T. Bjornsson et al., on page 3243. Recent breakthroughs in genomic research have clarified that genetic and environmental factors jointly contribute to the aetiology of many cardiovascular diseases. The identification of genes or, even better, genetically driven disease mechanisms involved in this process are crucial steps for improving risk prediction and developing better treatments. In monogenic disorders, the identification of genetic factors was boosted by next-generation sequencing (NGS), specifically if large affected families were available for co-segregation analyses. For common diseases, which are believed to result from multiple variants with small but additive effects, genome-wide association studies (GWAS) led to numerous discoveries. As an example, more than 250 susceptibility loci and causal mutations have been identified for coronary artery disease within the last 10 years, yielding astonishing insights into the disease mechanisms.1–3 The blind spot in all these efforts are less common or even rare diseases without classical Mendelian inheritance. On the one hand, their sporadic occurrence precludes co-segregation analyses; on the other hand, the number of cases is often too small for conventional GWAS analyses, which need large sample sizes in order to account for statistical punishment for multiple testing. Congenital heart diseases (CHDs) are a good example of such challenges. Although their genetic underpinnings are well documented,4 the low prevalence, phenotypic diversity, and usually mild penetrance has hampered the successful identification of underlying mutations for CHD.5 , 6 Bjornsson and co-workers from deCODE genetics have now merged the advantages of NGS and GWAS into a single big experiment and have successfully discovered a variant that is responsible for one out of five cases of coarctation of the aorta (CoA).7 They studied the entire Icelandic population, which has a very special standing in the field of genetics. First, this is because it is a ‘founder population’, which means that is goes back in history to a rather small number of ancestors. On the one hand, the smaller gene pool reduces genetic variation, while on the other hand, otherwise rare variants may be found more frequently. The founders of the Icelandic population were predominantly settlers of Norwegian origin.8 Second, the Icelandic healthcare system allows for an extensive collection of medical records and the merging of these in the deCODE genetics database. The third important aspect is the Íslendingabók, the book of Icelanders, which offers comprehensive genealogical information on the Icelandic population. This treasure, dating back to the founders of the population in the 9th century, has made the tremendous efforts on sequencing, genotyping and analysing the genetic codes of almost the entire Icelandic population worthwhile. Initially, Bjornsson and co-workers sequenced 15 220 individuals with NGS in order to detect and localize the vast spectrum of genetic variants, rare and common, in the population. Next, they genotyped 151 677 individuals with a GWAS array in order to link all mutations in the genome to a single nucleotide polymorphism (SNP) or haplotype. Thereby, the scientists submerged the entire information of the costly sequencing experiment (1500€ per individual plus enormous storage costs) into the more affordable information obtained with a GWAS array (25€ per individual). Long-range phasing of the chip-genotyped individuals and imputation from variants identified by NGS, including SNPs as well as insertions and deletions, led to the identification and annotation of 32.5 million sequence variants. By using a subtle approach to impute these variants into first- and second-degree relatives using genealogic information from the Book of Icelanders,9 the authors further increased the cohort by more than 200% for cases (to 120) and by over 150% for controls (to 355 166) without the need to obtain DNA from these individuals. Thus, by genealogical workup and genotyping, the authors extrapolated their sequencing data into the entire Icelandic population (Take home figure). Their sophisticated study design allows the analysis of variants with a frequency of 2 in 10 000 for association with disease. One of these rare variants ultimately appeared to cause one out of five non-syndromic cases of CoA. Study design of the genome-wide association study that Bjornsson and co-workers performed on coarctation of the aorta. The study was performed on 32.5 million variants obtained by whole-genome sequencing and imputed into chip-typed and long-range phased individuals, along with their first- and second-degree relatives. With a cohort of 120 coarctation of the aorta cases and 355 166 controls, Bjornsson and co-workers successfully discovered a variant that is responsible for one out of five cases of coarctation of the aorta (the Manhattan plot is taken from the original publication by Bjornsson and co-workers). CoA, coarctation of the aorta; GWAS, genome-wide association study; OR, odds ratio; WGS, whole-genome sequencing; WT, wild-type. Among the different CHD subtypes, CoA represents 1 in 2500 live births,10 or about 4% of all cases. Together with aortic valve stenosis and hypoplastic left heart, CoA constitutes a mechanistically defined subgroup of congenital heart defects, the left-ventricular outflow tract obstruction malformations, for which there is substantial evidence of a genetic component.11 However, despite some reports on familial clustering,12 Mendelian inheritance is rarely observed such that our understanding of the precise causes of non-syndromic CoA remains rather patchy. Using the unique wealth of the genetic data acquired for the Icelandic population, Bjornsson and co-workers have now found a genome-wide significant association signal with CoA on chromosome 14q11. The top SNP of the association finding was located in exon 18 of the coding region of the MYH6 gene, leading to an amino acid change on position 721 from arginine to tryptophan. The variant occurs with an allele frequency of 0.34% in the Icelandic population and associates with CoA with an odds ratio of 44 (!). Indeed, the MYH6 p.Arg721Trp variant was prevalent in 20% of all Icelandic CoA cases. Whether it plays a similar role outside of the island remains unclear, a disadvantage when a founder population is being analysed. In fact, it is barely present in other genetic variation databases compiling publicly available sequencing data. The MYH6 gene codes for the alpha heavy chain subunit of cardiac myosin and the p.Arg721Trp variant is located in the highly conserved myosin motor domain.13 Cardiac myosin is the major component of the sarcomere, which facilitates cardiac muscle contraction. Based on computational predictions, the amino acid change might alter the protein’s function. However, it appears that the gene is not expressed in the aorta. Thus, one can only speculate on the mechanism by which this variant could induce CoA. Indeed, if the prediction holds true and if there is a gain or a loss of function of the protein, the mechanism that links this variant to CoA needs solid experimental analysis. In vitro experiments thus far have shown that overexpression of the p.Arg721Trp variant in neonatal rat ventricular cardiomyocytes leads to the disruption of sarcomere integrity.14 However, since Bjornsson and colleagues identified the p.Arg721Trp variant not only in CoA cases but also in healthy subjects, it seems that the p.Arg721Trp variant is not sufficient to induce CoA by itself and that other factors may contribute as well. This suggests that other variants or the conjunction of exogenous factors facilitate the actual disease risk in an affected individual.15 Besides the novel association with CoA, the p.Arg721Trp MYH6 variant has been previously associated with sick sinus syndrome in the Icelandic population.16 In their current work, Bjornsson and colleagues further enlarged the search for association with other phenotypes (phenome screen). Indeed, they found associations of the MYH6 p.Arg721Trp variant with other CHDs as well. Perhaps more importantly, the mutation was also associated with heart failure and ischaemic stroke, i.e. common diseases, pointing to high variability in the phenotypic outcome of the p.Arg721Trp mutation. A final word of caution comes back to the issue of studying a founder population. In such analysis, the subjects sharing the MYH6 variant may also share larger chunks of DNA at this specific locus. Here, other mutations may be hidden that could indeed be the causative ones. In this respect, it is reassuring that other MYH6 mutations were found in a few familial cases as well as sporadic cases of CHDs, also presenting with CoA, e.g. 4 out of 10 family members suffering from atrial septal defect and carrying a p.Ala290Pro MYH6 mutation also had CoA.17 P.Arg443Pro was also found in a family presenting with hypoplastic left heart syndrome, in which one of the four affected also showed CoA.18 A non-syndromic patient with a double outlet right ventricle, mitral atresia, and CoA carried a p.Glu98Lys variant of this gene.19 However, (further) replication and, even more importantly, the elucidation of the precise mechanism needs to be awaited before the books on MYH6 p.Arg721Trp and CoA can be closed. Despite this caveat, the paper of Bjornsson and colleagues displays an elegant approach to associating a rare variant in an isolated population to a complex disease, opening up new opportunities to study the underlying mechanism, improve risk prediction, and facilitate the development of novel therapies and interventions. We would like to thank Dr Cornelia Pankalla for providing the volume rendering technique reconstruction of the computed tomography scan of the aortic CoA. Conflict of interest: none declared.