Genetic risk factors and causal associations with sudden cardiac arrest

Abstract

Sudden cardiac arrest (SCA) is a major cause of mortality, with complex etiology involving multiple genetic and environmental factors. In this study, we aimed to identify genetic risk factors for SCA and evaluate their causal associations using Mendelian randomization (MR) analysis. We retrieved summarized genome-wide association study data for various exposures and the SCA outcome from public databases. Four MR methods, including inverse variance weighted (IVW), Egger, weighted median, and weighted mode, were applied to estimate causal effects. All MR methods revealed significant positive causal associations with SCA from diastolic blood pressure (IVW: b = 0.09, P = 5 × 10⁻¹⁸), systolic blood pressure (IVW: b = 0.06, P = 5 × 10⁻²⁵), and total cholesterol levels (IVW: b = 0.28, P = 4 × 10⁻³). No consistent causal link was found for heart rate, body mass index, type 2 diabetes, and certain cardiac electrophysiological factors (P values for all 4 MR methods were not consistently below the significance level of .05). These findings provide insights into the genetic architecture of SCA and highlight potential targets for risk stratification and preventive strategies.

1. Introduction

Sudden cardiac arrest (SCA), also called sudden cardiac death (SCD), is a leading cause of death, accounting for approximately 15% to 20% of adult mortality in Western population.^[1] The etiology of SCA is complex, involving multiple factors such as coronary artery disease (CAD), cardiac electrophysiological abnormalities, and lifestyle factors.^[2] Despite advances in resuscitation techniques and preventive measures, the survival rate of SCA remains low.^[3] Therefore, identifying genetic risk factors and understanding their causal associations with SCA is crucial for risk stratification and prevention strategies.^[4]

Over the past decade, genome-wide association studies (GWASs) have identified hundreds of genetic variants associated with cardiovascular traits, including blood pressure, lipid metabolism, and cardiac electrophysiology.^[5–7] These discoveries suggest that genetic susceptibility plays a critical role in SCA pathogenesis. However, translating these findings into actionable clinical tools has been challenging. Single genetic variants typically confer modest risk increments, limiting their utility in isolation.^[8]

Mendelian randomization (MR) employs genetic variation as a natural quasi-experiment to improve causal inferences from observational data.^[3] Under certain assumptions, MR avoids confounding biases and reverse causation inherent to observational studies, and has been instrumental in validating causal roles for traits such as hypertension and hypercholesterolemia in cardiovascular diseases.^[8] Therefore, MR on SCA could clarify which modifiable risk factors (such as blood pressure, lipid levels, or electrocardiographic parameters) are causally linked to SCA, providing mechanistic insights and guiding preventive interventions.^[7,9]

In this study, we used two-sample MR analysis to investigate 2 key questions: Can genetic risk loci refine SCA prediction beyond conventional clinical risk stratification? Which modifiable traits demonstrate causal associations with SCA, and how might these inform prevention strategies? Using summary-level GWAS data from large consortia, we applied MR to assess causal relationships between 12 candidate risk factors (blood pressure, lipid profiles, arrhythmias, etc) and SCA. Our findings provide supporting evidence on the potential of genetic risk stratification in SCA prevention and identify actionable targets for reducing SCA burden. By bridging genetic epidemiology and clinical cardiology, this study advances our understanding of SCA’s genetic architecture and provides a framework for personalized risk assessment.

2. Methods

2.1. Ethics statement

This study used publicly available GWAS data for analysis and did not involve any laboratory experiments or individual-level personal data. This study was conducted in accordance with ethical guidelines. This study was approved by the Ethics Committees of Chengdu Women’s and Children’s Central Hospital (2023(18)-2) and Chengdu Medical College (2024-MW-2).

2.2. Summarized GWAS data source

To ensure data quality and result reliability, summarized association data for the exposures of interest (e.g., blood pressure, body mass index [BMI], and cholesterol levels) and the outcome of SCA were downloaded from the Integrative Epidemiology Unit Open GWAS database (https://gwas.mrcieu.ac.uk/; see details in Table S1, Supplemental Digital Content 1). To minimize stratification bias, all studies included participants of European ancestry. For most phenotypes, GWAS with the largest sample size and most recent publication date was selected; for certain phenotypes, multiple sets of GWAS data were used in parallel analysis.

2.3. Mendelian randomization

Associations in the exposure and the outcome were from separate studies or populations to avoid sample overlap. For GWAS datasets, a threshold of P value under 5 × 10⁻⁸ for the exposure and a clumping threshold of 0.001 for linkage disequilibrium in a 10,000-kb clumping distance window are used to select instrument variables (IVs). IV strength was evaluated by the F statistic (>10 used as a general indicator for strong instruments), and IVs missing in the outcome were replaced by proxy single-nucleotide polymorphisms (SNPs) when possible. Associations in the exposure and outcome were harmonized, and causality based on a single IV was estimated by the Wald ratio method. For weighted effects from multiple instruments, 4 MR methods of inverse variance weighted (IVW), Egger, weighted median, and weighted mode were applied. Post-MR analyses included the heterogeneity test (Q statistic) and the pleiotropy tests (MR-Egger intercept and Mendelian Randomization Pleiotropy RESidual Sum and Outlier). SNPs identified as outliers in Mendelian Randomization Pleiotropy RESidual Sum and Outlier were excluded from IV candidates. The workflow of MR analysis on SCA is demonstrated in Figure 1.

Figure 1.

Diagram of the Mendelian randomization analysis on sudden cardiac arrest. BMI = body mass index, DBP = diastolic blood pressure, GWAS = genome-wide association study, IV = instrument variable, IVW = inverse variance weighted, SBP = systolic blood pressure, TC = total cholesterol, TG = triglycerides.

2.4. Data analysis

Data analysis was conducted in the R statistical coding environment (version 4.4.1, R Core Team) run on a Microsoft Windows 11 platform (Microsoft Corporation, Redmond), with MR analysis primarily performed using the TwoSampleMR package (MRC Integrative Epidemiology Unit, University of Bristol, Bristol, UK). Core MR analysis was demonstrated in the supplemental R code file, Supplemental Digital Content 2. A significance level of α = 0.05 was used for null hypothesis tests. Positive MR effects between several exposures and the sole outcome of SCA were further confirmed for multiple-testing statistical significance by Benjamini and Hochberg adjustment.

3. Results

3.1. Identification of risk factors

Multiple risk factors for SCA have been previously identified, primarily based on associative evidence derived from observational studies. However, given the potential existence of uncontrolled/unmeasured confounders or due to the possibility of reverse causation, the causal links between these risk factors and SCA remain unchecked. MR uses qualified SNPs as IVs to query whether a causal link exists between a pair of exposure and outcome. To our knowledge, the FinnGen consortium represents the largest GWAS on cardiac arrest (defined as CARDARR within the International Statistical Classification of Diseases and Related Health Problems, 10th Revision category I9, “Diseases of the circulatory system”). The causal associations between each of the 12 exposures of interest and the single outcome of SCA were accordingly investigated by two-sample MR analysis, with causal estimates reported in Table 1, and visualized in the regression scatter plot (Fig. 2) and forest plot (Fig. 3).

Table 1.

Mendelian randomization analysis of the specified exposure and the outcome of sudden cardiac arrest.

Exposure	nSNP	IVW b (se), P	Egger b (se), P	Weighted median b (se), P
Diastolic BP	434	0.09 (0.01), 5 × 10−18	0.10 (0.02), 1 × 10−4	0.09 (0.02), 3 × 10−9
Systolic BP	431	0.06 (0.01), 5 × 10−25	0.07 (0.02), 1 × 10−5	0.07 (0.01), 1 × 10−13
Body mass index	478	0.36 (0.10), 3 × 10−4	0.21 (0.26), .41	0.24 (0.18), .19
Cardiac arrhythmia	11	10.16 (3.71), 6 × 10−3	8.64 (7.57), .28	9.54 (4.89), .051
Type 2 diabetes	167	0.02 (0.04), .57	−0.14 (0.09), .13	0.05 (0.07), .49
Total cholesterol	192	0.28 (0.10), 4 × 10−3	0.61 (0.13), 8 × 10−6	0.42 (0.13), 1 × 10−3
Total triglycerides	65	−0.01 (0.11), .94	−0.25 (0.17), .13	0.09 (0.14), .50
QRS duration #1	11	−0.003 (0.01), .84	0.03 (0.06), .64	−0.002 (0.02), .90
QRS duration #2	6	0.07 (0.28), .80	1.52 (1.01), .21	0.05 (0.29), .87
QT interval	13	−0.01 (0.01), .047	−0.003 (0.02), .86	−0.003 (0.01), .72
PR interval	11	−0.006 (0.01), .34	−0.01 (0.02), .59	−0.008 (0.01), .21
Atrial fibrillation #1	109	0.25 (0.05), 2 × 10−7	0.08 (0.09), .36	0.15 (0.08), .054
Atrial fibrillation #2	90	0.20 (0.07), .005	−0.02 (0.18), .91	0.11 (0.11), .34
Atrial fibrillation #3	10	0.08 (0.06), .17	0.07 (0.14), .62	0.04 (0.08), .61
Heart rate #1	10	−0.13 (0.35), .70	0.27 (0.92), .78	−0.16 (0.33), .63
Heart rate #2	6	−0.61 (0.52), .24	−0.70 (1.70), .70	−0.64 (0.63), .31

QRS, QT, PR – as defined in an electrocardiogram.

Due to space limitations, results of the weighted mode method were not displayed.

Multiple datasets were used for certain exposures (e.g., QRS duration #1 and #2).

The 95% CI of regression coefficient β could be calculated from sample-estimated statistics of b, se, and n. For example, the 95% CI of β_IVW for “DBP − SCA” is (0.09 − 1.96 × 0.01, 0.09 + 1.96 × 0.01) or (0.070, 0.110).

BP = blood pressure, CI = confidence interval, DBP = diastolic blood pressure, IVW = inverse variance weighted, SCA = sudden cardiac arrest, SNP = single-nucleotide polymorphism.

Figure 2.

Scatter plots of causality. Mendelian randomization analysis with the effects of various exposures on sudden cardiac arrest/death (cardiac‖id.mw28). The slope of each line corresponds to the estimated causal effect in different methods. Scatter plots show the association of instrument variables (IVs) with the exposure and the outcome (coefficients for point coordinates and standard errors for error bars). Line plots show regressed lines of multiple-IV estimates of inverse variance weighted (IVW), Egger, weighted median, and weighted mode. Panels of scatter-regression plots are sequentially displayed from the top-left corner as: BMI, DBP, SBP, T2D, total cholesterol, total triglycerides, QRS duration, QT interval, PR interval, atrial fibrillation, heart rate, and cardiac arrhythmia. BMI = body mass index, DBP = diastolic blood pressure, ECG = electrocardiogram, MR = Mendelian randomization, SBP = systolic blood pressure, SNP = single-nucleotide polymorphism, T2D = type 2 diabetes.

Figure 3.

Forest plot of 4 Mendelian randomization methods on 12 pairs of exposure–outcome causal analysis. The dot and extended bars depict the point estimate and 95% confidence interval of the causal effect. If the P value is <.05, the dots are solid; otherwise, the dots are hollow. BMI = body mass index, DBP = diastolic blood pressure, IVW = inverse variance weighted, MR = Mendelian randomization, SBP = systolic blood pressure, SCA = sudden cardiac arrest, T2D = type 2 diabetes.

3.2. Causal associations from MR analysis

3.2.1. Blood pressures and heart rates

Both systolic and diastolic blood pressures (SBP/DBP) are multigenetic traits with over 400 significantly associated SNPs. All 4 MR methods revealed a statistically significant positive causal link from blood pressures to SCA, suggesting that higher blood pressures increase the risk of SCA. Heart rates have much fewer IVs available and do not appear to lead to cardiac arrest.

3.2.2. Total cholesterol and triglyceride levels

Total cholesterol (TC) levels also displayed a consistently significant causal association with SCA. In contrast, total triglyceride levels showed no causal link to SCA.

3.2.3. BMI and type 2 diabetes

BMI, a common marker for obesity, was considered a risk factor for SCA in observational studies. However, only IVW-MR analysis suggested a possible causal link (P = 3 × 10⁻⁴), while other MR methods displayed no causal findings. Similarly, type 2 diabetes (T2D) did not show a modified risk for SCA from MR analysis.

3.2.4. Cardiac arrhythmia and electrophysiological factors

Apart from the above physiological exposures, we tested several cardiac factors for their associations with cardiac arrest. As a binary exposure, cardiac arrhythmia had large effect sizes compared to other exposures, but the causal link with SCA was only observed by the IVW method (P = 6 × 10⁻³). Despite being a related phenotype, atrial fibrillation (AF) also displayed a substantial causal link to SCA only in 2 out of 3 IVW scenarios. No significant causation was found between electrocardiogram measurements (QRS duration or PR interval) and SCA, and the causal effect of QT interval was also marginal only in the MR-IVW analysis (P = .047).

4. Discussion

In this study, we aimed to investigate the important question of whether associations between various potential exposures and the outcome of SCA were causal or not. Using two-sample univariate MR analysis, we screened several risk factors and demonstrated that 3 modifiable factors (SBP, DBP, and TC levels) are causal and thus could be applied to potentially reduce the risk of SCA. Other risk factors displayed negative or questionable (only significant with a certain MR method) causal links. Future endeavors to conduct two-sample multivariate MR analysis are worthwhile to further substantiate the causal links between the 3 exposures and the outcome of SCA.

4.1. Blood pressure: a pervasive and modifiable risk factor

Our MR analysis provides unequivocal evidence that both SBP and DBP exert causal effects on SCA risk. All 4 MR methods (IVW, Egger, weighted median, and weighted mode) consistently identified a positive causal relationship, with an increase in SBP or DBP associated with a higher SCA risk. This finding aligns with the Health in Men Study, in which hypertension was related to a 14% increase in risk for SCA per 10 mm Hg increment in SBP.^[10,11] Chronic hypertension induces myocardial remodeling, including fibrosis and electrical instability, which predisposes to ventricular arrhythmias. Experimental models show that angiotensin II-mediated oxidative stress accelerates apoptosis of cardiomyocytes, creating a substrate for reentry circuits.^[12] Additionally, hypertension elevates cardiac workload, increasing wall stress and oxygen demand – factors implicated in SCA cases with CAD.^[13] Discrepancies may exist in subgroup analyses. For example, a meta-analysis (n = 418,235) found an increased risk of SCD with hypertension diagnosis and increasing SBP,^[7,11] and also strong associations with middle and old ages,^[10] suggesting age-dependent mechanisms. Methodologically, we used clumping windows (10,000 kb) to prioritize independent SNPs, but this may exclude variants with moderate effects in gene–gene interaction networks.^[14] Future studies should explore nonlinear relationships in elderly populations.

4.2. Total cholesterol: a lipidic nexus of atherosclerosis and arrhythmogenesis

Elevated TC emerged as another robust causal risk factor for SCA, consistent with studies linking obesity and TC to myocardial infarction and SCD.^[7,15,16] TC promotes atherosclerotic plaque rupture through 2 pathways: low-density lipoprotein (LDL)–cholesterol levels exhibited a reverse J-shaped association with the incidence of SCA through atherosclerotic plaque^[15,16] and oxidized LDL impairs ion channel function, prolonging action potential duration and QT dispersion – a known precursor to torsades de pointes.^[17] Our finding contrasts with studies suggesting that high-density lipoprotein or LDL subfractions are more relevant.^[18] The observational study reported no association between TC and SCA after adjusting for high-density lipoprotein/LDL ratios.^[18] This discrepancy may reflect that TC-associated SNPs may influence liver metabolism rather than arterial biology. Future research should adopt multi-trait polygenic scores to disentangle these effects.

4.3. BMI: a complex mediator of SCA risk

While observational studies consistently link obesity (BMI) to SCA risk, our MR analysis revealed only nominal support for its causal role. This discrepancy highlights the multifaceted nature of obesity’s impact on cardiac health. Obesity may influence SCA risk through intermediate pathways such as adipose tissue secretes pro-inflammatory cytokines, which promote myocardial fibrosis and electrical instability,^[19] and obesity increases the prevalence of obstructive sleep apnea, a condition linked to nocturnal cardiac arrhythmias.^[20] For example, anthropometric measures and information on circumstances surrounding the cardiac arrest are rarely captured in genetic databases. Therefore, the underlying etiology and mechanism of death may be heterogeneous, and genetic associations are likely to be diluted.^[7] This limitation may explain why our IVW-MR analysis yielded only a marginally significant effect.

4.4. Type 2 diabetes: indirect associations and disease interactions

Similar to BMI, T2D showed no robust causal link to SCA in our MR analysis. This finding contrasts with clinical observations that diabetic patients have higher SCA rates,^[21] suggesting that T2D’s association is likely mediated by comorbid conditions. T2D and CAD act as genetic risk factors for SCA,^[16] creating a “collider bias” in MR. Our analysis likely diluted T2D’s true effect because the GWAS data included CAD patients without stratification by myocardial infarction history.

Emerging evidence suggests that insulin resistance drives cardiac electrophysiological dysfunction. Preclinical studies show that insulin signaling defects impair potassium channel function, prolonging action potentials and increasing reentry risk.^[22] However, most T2D GWASs prioritize glycemic traits, which may not capture insulin resistance’s electrophysiological effects.

4.5. Cardiac electrophysiology: QT interval as a key mediator

Prolonged QT interval emerged as the strongest electrophysiological predictor of SCA, underscoring its critical role in arrhythmogenesis. Mechanistically, QT prolongation facilitates early afterdepolarizations, which are premature depolarizations during phase 2 or 3 of the cardiac action potential. Abnormal action potential duration and amplification of the spatial dispersion of repolarization, coupled with EADs inducing triggered activity, are the primary causes of SCA. Therefore, genetically determined serum calcium levels and markers of ventricular repolarization may directly influence cardiovascular disease risk.^[23] In contrast, AF showed limited causal evidence for SCA in our analysis. Evidence suggests that while AF is a potential mediator of SCA, its association is primarily driven by heart failure and myocardial scar.^[23,24] Multivariable MR confirms that AF itself is not a direct cause of SCA.^[25]

4.6. Limitations and future directions

Our study has several limitations. First, the sample size for certain exposures may be insufficient to detect modest causal effects. Larger GWASs with more comprehensive genetic data are needed to address this issue. Second, potential horizontal pleiotropy of genetic variants could affect the validity of MR analysis. We used multiple methods to test for pleiotropy and excluded potentially pleiotropic SNPs, but residual pleiotropy cannot be completely ruled out. Finally, our study focused on genetic risk factors and did not consider gene–environment interactions, which may also play important roles in SCA risk.

Different methods have diverse sensitivity to IV assumptions. In this study, IVW was adopted as the primary analysis method, as it provides the highest statistical power under the assumptions of linearity and no pleiotropy. Furthermore, the effect directions estimated by Egger and weighted median estimation were largely consistent with those from the IVW method. The concordance among different methods provides robust evidence for the causal relationships identified. The limitations of MR in this study primarily lie in the fact that although we have conducted extensive sensitivity analyses, residual bias due to unmeasured confounders cannot be entirely ruled out.

Similar to the methodological design of this study, recent efforts were made using MR analysis to screen for nonclinical risk factors for cardiac arrest^[26] or investigate genetic predisposition in ischemic stroke-induced cardiac arrhythmias.^[27] Future studies should aim to include larger sample sizes, more comprehensive genetic data, and consider gene–environment interactions to provide a better understanding of the genetic architecture of SCA. Additionally, functional studies are needed to elucidate the biological mechanisms underlying the identified causal associations.

5. Conclusion

In conclusion, our study utilized MR analysis to evaluate the causal associations between various risk factors and SCA. We identified blood pressure and TC levels as significant genetic risk factors for SCA, while BMI, T2D, and several cardiac electrophysiological factors did not show convincing causal associations. These findings provide insights into the genetic architecture of SCA and have implications for risk stratification and preventive strategies.

Author contributions

Conceptualization: Huaye Wu.

Data curation: Fengling Hu, Lin Liu, Mengjun Wu, Mengqiao Wang.

Formal analysis: Huaye Wu, Mengjun Wu, Mengqiao Wang.

Funding acquisition: Huaye Wu, Mengjun Wu, Mengqiao Wang.

Investigation: Huaye Wu, Mengjun Wu, Yonghong Lin, Mengqiao Wang.

Methodology: Huaye Wu, Mengjun Wu, Mengqiao Wang.

Project administration: Mengjun Wu, Yonghong Lin, Mengqiao Wang.

Resources: Huaye Wu, Mengjun Wu, Mengqiao Wang.

Software: Huaye Wu, Fengling Hu, Mengqiao Wang.

Supervision: Yonghong Lin, Mengqiao Wang.

Validation: Mengjun Wu, Yonghong Lin, Mengqiao Wang.

Visualization: Huaye Wu, Mengqiao Wang.

Writing – original draft: Huaye Wu.

Writing – review & editing: Huaye Wu, Yonghong Lin.