Introduction

Coronary artery disease and its ultimate consequence, myocardial infarction (MI), are believed to underlie 75% of the deaths of patients who experience sudden cardiac death (SCD).[1–3] It is also estimated that SCD accounts for 30 to 50% of all coronary deaths.[4] SCD is a major challenge for the clinician because most episodes occur in individuals without previously identified cardiac disease.[5,6] The pathophysiology of SCD caused by MI is complex. It is believed to involve an interaction between underlying disease and a transient event, such as acute myocardial ischemia, which results in electrical instability and ventricular fibrillation (VF).[2] VF is a life-threatening complication and occurs in nearly 12% of patients with first ST-segment elevation myocardial infarction (STEMI).[7,8] Therefore, efforts to improve risk stratification for SCD should in part be based on an understanding of the molecular mechanisms and pathways underlying the occurrence of VF.[9] In the GEnetic causes of Ventricular Arrhythmias in patients with first ST-elevation Myocardial Infarction (GEVAMI) study on ethnical Danes[7] and the Dutch AGNES (Arrhythmia Genetics in the NEtherlandS Study) study[10] both suggested a possible genetic component to VF since patients with a family history of sudden death among first-degree relatives had significantly higher odds of experiencing VF before primary percutaneous coronary intervention (PPCI). In the AGNES study, a common genetic variant at 21q21 (rs2824292, odds ratio (OR) = 1.78, 95% CI 1.47–2.13, P = 3.3x10^-10) was found to be associated with VF before PPCI in STEMI patients in a genome-wide association study (GWAS).[11] The rs2824292 is located upstream of CXADR gene which encodes the coxsackie and adenovirus receptor (CAR), a transmembrane cell adhesion molecule predominantly located at the intercalated disc between cardiomyocytes.[12] However, this association has yet to be replicated in a larger cohort. There have also been several other variants reported to be associated with SCD, primarily from candidate gene studies with limited replication.[9] Therefore, we aimed to determine whether the 21q21variant, as well as other variants previously associated with SCD, would be associated with underlying susceptibility to VF due to first STEMI in the GEVAMI case-control study.

Methods

Statistical analysis

Medians or proportions of baseline and presenting characteristics in Table 1 were computed for cases and controls, and significance of associations were tested using the Wilcoxon rank-sum test for continuous variables and the χ2 test or Fisher exact test (where appropriate) for categorical variables. A two-tailed p value ≤0.05 was considered statistically significant. An age and sex-adjusted logistic regression model were constructed to estimate OR for the association between each SNP and VF using an additive model of inheritance. Any SNPs which were significantly associated with VF were also adjusted for clinical differences between cases and controls such as family history of sudden death, alcohol consumption, previous atrial fibrillation, statin use, angina, culprit artery, and thrombolysis in myocardial infarction (TIMI) flow.[7] We considered each SNP as a single hypothesis however a Bonferroni correction for multiple testing was used to assess the association of single SNP with VF. Homozygote risk allele and adjusted odds ratios are presented in Table 2. Per copy risk allele frequency is provided in Table 3. All analyses were performed using the Stata software package version 12.0 (StataCorp).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics of the cohort.

https://doi.org/10.1371/journal.pone.0170193.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Homozygote genotype of inheritance for association of 27 SNPs previously associated with sudden cardiac death and in this study investigated for association with ventricular fibrillation before ST-segment elevation myocardial infarction.

https://doi.org/10.1371/journal.pone.0170193.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Additive genetic model of inheritance (per copy allele frequency) for association of 27 SNPs previously associated with sudden cardiac death, and in this study investigated for association with ventricular fibrillation before ST-segment elevation myocardial infarction.

https://doi.org/10.1371/journal.pone.0170193.t003

Results

Discussion

In this prospective case-control study in Denmark we identified a common variant (rs11720524) located in intron 1 of the SCN5A gene near the promoter region which was associated with VF in patients with STEMI. To our knowledge, this is the first study linking a common variant in SCN5A, the major sodium ion channel in the heart, to VF caused by first STEMI. Genetic studies have been successful in uncovering rare mutations in ion channel genes and these have been associated with rare Mendelian arrhythmic disorders, most notably the long-QT syndrome and to a lesser degree Brugada syndrome,[9,26] and for some of these disorders genetic testing has become a pivotal part of clinical care.[27,28] In contrast, studies aimed at identifying genetic risk factors for VF in the setting of MI affecting the majority and older segment of the population have been sparse[11] and therefore genetic architecture of this complex and multi-factorial disease largely remains unknown.

The major C allele of rs11720524 has previously been shown associated to SCD in a combined nested case-control analysis among 516 cases and 1522 matched control subjects of European ancestry enrolled in 6 prospective cohort studies.[14] Although the associations between the SNP (rs11720524) and sudden/arrhythmic death were relatively consistent across the 6 studies, nevertheless this finding required an independent replication study as it was done in our case-control setup which has a comparable size and is well-phenotyped (VF), which is necessary to establish certainty regarding the observed association. To support the association of rs11720524 with VF in our study a recent study also showed an association of the C/C genotype of the rs11720524 with SCD (OR = 1.351; p = 0.019).[29] Although this association was shown in a highly heterogeneous cardiac death cohort further subgroup analysis also showed an association of the C/C genotype of the rs11720524 with SCD in chronic ischemic heart disease patients (OR = 1.455; p = 0.012). Two other SNPs in SCN5A (rs41312391 and rs11708996) were not associated to VF in the GEVAMI cohort. The SNP rs41312391risk allele (A) which is also located in the intron near SCN5A has been associated with QT-interval prolongation in normal subjects in one study (n = 282, P = 0.04).[30] and QT-interval shortening in another study (n = 396, P = 0.02).[31] Furthermore, in the FinSCDgen Study, Lahtinen et al. and coworkers showed an association of the SNP rs41312391 risk allele (A) with SCD with relative risk 1.27 per minor T allele, 95% CI 1.11–1.45, p = 3.4x10^-4).[25] The expression analysis indicated that rs41312391 may change the expression level of a nearby gene, WDR48, which encodes a WD repeat-containing protein, a regulator of histone deubiquitinating complexes.[25] However, they concluded that SCN5A remains the more likely candidate gene for SCD due to its known function in cardiac conduction and repolarization. Furthermore, the SNP rs11708996 has been associated with PR interval (which is an intermediate phenotype of atrial fibrillation),[32] QT interval[33] and Brugada syndrome.[34] This finding suggests that the SNP rs11708996 modifying effects are distinct and in the GEVAMI cohort did not predict risk of VF in STEMI patients.

Identifying the exact causal SNP within the haplotype tagged by (rs11720524) to understand the underlying genetic mechanism is important and requires functional studies to further understand the possible role of this gene in the setting of VF. Since SNPs are highly correlated (in linkage) to the neighboring SNPs and thereby forming haplotypes and presumably these intronic variants in this halpotypes may exert their influence through the level of expression of SCN5A. A number of SNPs at the SCN5A and SCN10A locus (Table 2) have been associated with ECG changes such as QT, RR and QRS interval[9], however our lead SNP (rs11720524) was not in linkage disequilibrium with any one of these.

The rs9388451 SNP is a non-coding SNP and near the HEY2 gene and encodes a basic helix-loop-helix transcriptional repressor that is expressed in the cardiovascular system.[35] Recently, the rs9388451 SNP was shown to be associated with Brugada syndrome and data indicates that HEY2 regulates cardiac electrical activity in Brugada syndrome by altering the transcriptional programming during cardiac development.[13] Even though the association of the (C/C) rs9388451 was not significant (OR = 1.50; P = 0.070), it showed the same trend and it could be due to lack of power in our analysis.

Several studies[7,10,36] such as our Danish GEVAMI study and the Dutch AGNES study, showed an association of family history of sudden death with VF caused by first STEMI suggesting a genetic burden of the disease. The AGNES study reported the first GWAS for STEMI patients with VF before PPCI. In this GWAS study the most significant association with VF was found at 21q21 (rs2824292, OR = 1.78, 95% CI:1.47–2.13; P = 3.3x10^-10).[11] Despite this novel and important finding, the association of SNP rs2824292 was not detected in the GEVAMI study nor in a small German case-control study.[7,37] However, there is no obvious explanation for this disparity. It could be explained by lack of power or maybe the effect of the risk allele may differ between populations because of gene-gene or gene-environment interactions.[38] Larger sample sizes and homogenous enriched subgroups will be needed to identify and replicate additional genetic variants associated with VF during STEMI. For instance the location of the culprit coronary lesion may modify QRS duration and QTc interval and thereby risk of VF.[9,39] Stratification of VF patients based on culprit lesion may give us a more discrete phenotype, but this requires a very large sample size. An even more discrete phenotype based on the mechanism of VF (whether it is an immediate VF (phase Ia) or delayed (after the first 15 minutes) VF (phase Ib)) may also be warranted.[40] This hypothesis is supported by experimental data comparing wild-type (WT) and heterozygous CAR deficient (CAR^+/−) mice.[39] These data suggested that the burden of VF is much higher in CAR^+/− mice with acute left anterior descending (LAD) occlusion compared to the WT, and this higher VF burden was only observed within the first 15 minutes of acute occlusion of LAD. During later stages of the ischemic episode, arrhythmia inducibility was not different between WT and CAR^+/− mice.[39]

Lastly, in a combined meta-analysis of GWAS studies of individuals with SCD and control individuals of European ancestry, a locus at chromosome 2q24.2 (rs4665058) was found to be associated with SCD[17]; however, this SNP was not associated to VF in the AGNES study or in the GEVAMI study. These inconsistencies could be caused by heterogeneous underlying cardiac pathologies of SCD and differing phenotypes (i.e., VF vs. SCD).[9]

The common SNPs associated with VF in the studies mentioned above all by nature had small effect sizes on risk and are not currently used for clinical management. With the rapid development of next-generation sequencing technologies, large-scale sequencing projects are becoming possible allowing the examination of rare genetic variants with larger effects which could be useful for risk stratification.

We considered each SNP as a single hypothesis but a Bonferroni correction for multiple testing is often applied in genetic association studies but might be too conservative for highly selective candidate SNP-based approaches. However, we need to acknowledge that none of the SNPs would be significant if Bonferroni corrected for multiple testing (P = 0.0019[0.05/27]). Furthermore, the sample size of 257 VF cases is still fairly low and limits the ability of this study to find association of common variants with modest effect. It is also important to mention that collecting cases is very time consuming and limited due to relatively low incidence of VF and difficulties to enroll and collect blood samples in STEMI-patients with VF and cardiac arrest. This is mostly due to that patients who died outside of the hospital or died in-hospital prior to enrollment cannot be included. Most of the 27 SNPs reported to be associated with SCD/VF were identified in patients without acute coronary syndrome, hence may reflect the underlying pathophysiology of SCD/VF in a more general population rather patients with VF due to first STEMI.

Finally, our results in this population with white, European/Danish ancestry may not be generalizable to other populations.

The present study nicely shows that it is often difficult to replicate the genetic associations observed in previous studies in some other populations. The genetic factors involved in this complex and multi-factorial disease are still largely unknown; therefore the current study on all the previous SNPs associated with SCD/VF is important, especially for future direction of functional studies for more clarification of arrhythmia mechanism and targeted intervention strategies.