Genome-wide association studies: providers of candidate genes for identification of rare variants?

This editorial refers to ‘Screening of KCNN3 in patients with early-onset lone atrial fibrillation’ by M. S. Olesen et al., on page 963. Classical linkage analysis is a powerful approach to associate new genes with disease without any a priori pathophysiological hypothesis. Linkage analysis evaluates markers widely spaced across the genome to determine whether they are inherited along with disease in families with numerous affected individuals. Using this strategy, several ion channel-encoding genes have been associated with monogenic electrophysiological disorders including the identification of KCNQ1 as a gene underlying familial atrial fibrillation (AF).1 This approach may not, however, be generally applicable since in many cases the size of the affected family is too small. Consequently, some researchers have resorted to candidate gene analysis for identification of new genes underlying AF.2 Success of the candidate gene approach relies strongly on a relevant prior hypothesis. However, although some pathways are highly suspected to play a role in pre-disposition to arrhythmia, unknown mechanisms are also very likely involved. Over the last years we have seen a flurry of genome-wide association studies (GWAS) in the literature addressing a number of electrophysiological and arrhythmia phenotypes. In these studies, single-nucleotide polymorphisms (SNPs) that are relatively common in the general population are tested for association with the phenotype of interest. This approach has also been applied to AF and has led to the identification of three distinct genetic loci on chromosomes 1q21, 4q25, and 16q22 that are associated with this arrhythmia.3 The observation that genetic variants detected by most GWAS are associated with small effects and do not markedly affect the phenotype (e.g. arrhythmia) by themselves and the fact that the variants identified usually explain only a small amount of the disease heritability, tempered, to some extent, the enthusiasm surrounding these studies.4 Furthermore, much of the heritability remains unexplained in these studies. This could be due to the involvement of environmental factors in the manifestation of these disorders, alone or in association with genetic variants (also called gene–environment interaction). Another reason could be that variants which explain a larger part of the variation of the phenotype in the population are contained in the GC-rich region (i.e. a region containing an above average number of the bases guanine and cytosine) of the genome that could be insufficiently interrogated by the commercial genotyping arrays, but may correspond to relevant enhancer or promoter regions. In addition, complex diseases can follow a polygenic model in which the disease only manifests when a whole combination/series of frequent variants, each carrying a small effect, are co-inherited. Although SNPs uncovered by GWAS are associated with a small effect and consequently only modulate risk in a modest way, such studies are important as they may uncover genetic variants in genes previously unsuspected to be related to the phenotype under study. This applies for instance for KCNN3 and AF. This triggers basic scientific research on these genes which in turn furthers our understanding of the disorder. A by-product of the GWAS approach is that they also point to genes that are prime candidates for involvement in the familial form of the particular disease. Furthermore, these genes constitute prime targets for testing the Common Disease-Rare Variant hypothesis which argues that rare DNA sequence variants, each with relatively moderate-to-high penetrance, are important contributors to genetic susceptibility for common disease.5 This latter strategy has already been used successfully in the identification of rare variants associated with common diseases in genes located in genomic regions initially identified by GWAS.4 For instance, Nejentsev et al.6 identified four rare variants in the IFIH1 gene, protecting against type 1 diabetes. This study has demonstrated that resequencing studies can identify disease-causing genes in genomic regions initially identified by GWAS. However, although several GWAS have uncovered common genetic variation impacting on several cardiac electrical traits such as heart rate and electrocardiographic indices of conduction and repolarization, to our knowledge, no rare variants have been identified in genes identified through these GWAS, neither in common, nor in rare arrhythmia disorders. A case in point is the highly significant association of genetic variation in the NOS1AP gene (previously unlinked to cardiac electrophysiology) and the QT interval duration, a measure of cardiac repolarization.7 Since its implication as a modulator of the QT-interval in 2006, no rare variants have been identified. Similarly, mutation screening of PITX2 and NKX2.5, associated with AF in GWAS studies and both involved in cardiac development, did not reveal rare variants when screened in patients with idiopathic AF in one study.8 Olesen et al.9 screened 209 young lone AF patients for identification of rare variants in the coding region of KCNN3 on chromosome 1q21, recently associated with lone AF in a large meta-analysis.10KCNN3 encodes a potassium channel protein involved in atrial repolarization. Again, no rare variants were identified. Although the study by Olesen et al. did not report rare variants in this gene, the strategy of searching for rare variants in genes identified by GWAS remains highly instrumental in facilitating the identification of the underlying gene within the locus and in providing tools for functional studies into the underlying pathophysiological pathways. On the other hand, Olesen et al. found an association between the synonymous SNP rs1131820 in this gene and lone AF. Carriers of the G-allele, which is the most common allele in the general population at SNP rs1131820, had an increased risk of AF. The intronic SNP rs13376333, identified in the previous GWAS as correlated with lone AF, does not seem to correlate (low linkage disequilibrium) with this SNP, which suggests that AF could be impacted by multiple alleles within KCNN3. If we add the independent susceptibility SNPs described on locus 4q25, this finding would be in line with a polygenic model for AF.3 The mechanism by which variations at this locus increase AF risk remains unknown. The authors are supported by research grants from the Netherlands Heart Foundation (2008B044), the Inter-University Cardiology Institute of the Netherlands (ICIN, 06102) and the Academic Medical Center. Conflict of interest: none declared.