Introduction

Heart failure (HF) and atrial fibrillation (AF) are major cardiovascular diseases that pose a global challenge. Heart failure and AF often co-occur in the same patients, with more than a third of AF patients also having HF,¹ and vice versa, up to half of HF patients also having AF. The coexistence of these 2 conditions can worsen each other's prognosis.² However, the causal relationship between AF and HF is still unclear in many cases. Although some observational studies have shown an association between AF and HF.^3-5 these findings are prone to confounding and reverse causation.

Recent evidence has highlighted the survival benefits of maintaining sinus rhythm through interventions such as catheter ablation in patients with HF, suggesting a potential therapeutic advantage over medical therapy.⁶ This underscores the importance of rhythm control, particularly in the context of HF, where the interplay between AF and HF can be particularly detrimental. Furthermore, the necessity for repeat procedures to maintain sinus rhythm in patients with HF has been recognized, indicating a dynamic approach to the management of these conditions.⁶ In addition to pulmonary vein (PV) isolation, the role of extrapulmonary triggers—such as those originating from the left atrial posterior wall, left atrial appendage, ligament of Marshall, coronary sinus, superior vena cava, and crista terminalis—has become increasingly evident. These non-PV triggers are significant contributors to AF recurrence, especially in patients with persistent AF, and their management is crucial for the long-term maintenance of sinus rhythm.⁷

Mendelian randomization (MR) is a method that can help overcome these limitations.^8,9 In this case, MR uses genetic variants related to the exposure as instrumental variables (IVs), similar to randomized controlled trials, which randomly assign alleles at conception according to Mendel’s second law.¹⁰ This way, confounding factors are randomly distributed across the population. After identifying the common variants in the outcome data, the basic principles of MR can be used to evaluate the effect of the suspected exposure factors on disease risk. Thanks to the rapid increase in genome-wide association studies (GWASs) in the past decade, other studies have used MR to establish causal relationships between blood pressure and AF,¹¹ body mass index (BMI) and AF,¹² and AF and HF in European ancestry.¹³ However, evidence for a causal relationship between AF and HF in East Asian ancestry remains scarce.

In this study, we conducted a bidirectional 2-sample MR (TSMR) analysis for the first time to assess causal associations between AF and HF in East Asian ancestry.

Methods

Data Source

Our analysis relied on GWAS summary data that is curated and centralized by the Medical Research Council Integrative Epidemiology Unit open GWAS database (https://gwas.mrcieu.ac.uk), These data are available through the R package “TwoSampleMR.”^14,15 A total of 8180 cases and 28 612 controls for AF from the GWAS for East Asian descents were used in this study (https://gwas.mrcieu.ac.uk/datasets/bbj-a-71/).¹⁶ The outcome variable summary-level association results for the final sets of AF-related SNPs were extracted from the largest meta-analysis of GWASs for HF, which tested associations between 8 885 805 genetic variants and HF in a total of 9413 cases and 203 040 controls of East Asian descents(https://gwas.mrcieu.ac.uk/datasets/bbj-a-109/) (Table 1).

To ensure the robustness of our MR analysis, we verified the independence of the 2 samples. The AF-related SNP data were derived from individuals without HF, and conversely, the HF-related SNP data were sourced from individuals without documented AF. This independence is critical to the validity of our MR findings, as it prevents potential bias that could arise from overlapping samples.

Genetic Instrumental Variable Selection

To estimate the causal effect of AF on the risk of HF, we selected genetic variants as IVs in line with the following assumptions: (1) they have a predictive effect for AF, (2) they are independent of confounders, and (3) they do not affect the outcome through any other pathways than through AF.¹² Single nucleotide polymorphisms (SNPs) that reached genome-wide significance (P < 5 × 10⁻⁸) were chosen as IVs. To ensure the independence of genetic variants, these SNPs then underwent linkage disequilibrium pruning (distance threshold = 10 000 kb, r ² < 0.001). We subsequently excluded the SNPs associated with potential confounders of the outcome. In our study, coronary artery disease, hypertension, valvular heart disease (VHD), BMI, chronic obstructive pulmonary disease (COPD), alcohol and tobacco consumption were identified as confounders for HF (as identified at http://www.phenoscanner.medschl.cam.ac.uk/) Supplementary Table 1). The strength of the relationship between the SNP and AF was assessed with an F-statistic calculated as F = R² (n − K − 1)/(1 − R²)  × K. This calculation factors in exposure variance (R²), sample size (n), and the number of SNPs (K). An F-statistic > 10 indicates a significant association between the selected IVs and AF.¹⁷

Mendelian Randomization Analysis

We used the inverse variance weighting (IVW) method as the main analysis to assess the causal relationship between AF and HF in our TSMR study. The IVW method calculates the exposure effect of each SNP using the Wald ratio method and then performs a weighted linear regression with a forced 0 intercept. When the IV meets 3 basic assumptions, it achieves higher accuracy and power in estimation.¹⁸

To account for the possible confounding by unknown and unmeasurable factors, we performed the robust adjusted profile score (MR-RAPS), the MR-Egger regression (MR-Egger), the weighted mode, and the weighted median estimation to test the robustness of our results. Mendelian randomization-RAPS is an extension of IVW that allows for very weak instruments.¹⁹ The weighted median assumes that at least 50% of the instruments are valid.²⁰ For the mode-based estimations, simple mode and weighted mode provide a consistent result when there is no pleiotropy among all instruments.²¹ Mendelian randomization-Egger method provides estimates that are corrected for pleiotropic effects, but with lower statistical power.²²

In addition to these analyses, we also conducted a bidirectional MR study. This reverse MR analysis was performed to investigate the potential causal effect of HF on the risk of developing AF.

Sensitivity Analysis

We conducted several sensitivity analyses to further explore the potential heterogeneity and pleiotropy between exposure and outcome. We evaluated the IV using the Cochran’s Q test, with P _h < .05 indicating heterogeneity.²³ As a measure of horizontal pleiotropy, we used the Egger intercept in the MR Egger regression analysis.²² We performed multi-instrument summary-level MR tests using MR pleiotropy residual sum and outlier test (MR-PRESSO) to detect horizontal pleiotropic outliers.²⁴ We also performed IVW radial variants and MR-Egger radial variants, which are similar to conventional IVW and MR-Egger regressions, to improve the visualization of the IVW and MR-Egger estimates.²⁵ Our sensitivity analysis included leave-one-out analysis, accompanied by IVW, to determine the combined effect of the remaining SNPs. The MR analysis is not overly influenced by any single SNP if the combined effect is consistent with the main effect result.

We performed all analyses with R software (version 4.2.1) using R packages (“Two Sample MR,” “MR-PRESSO,” and “Radial MR”). Statistical significance was defined as a 2-sided P-value of less than .05.

Power Calculations

We used the publicly available mRnd web tool, we evaluated the power of our study using a non-centrality parameter approach (http://cnsgenomics.com/shiny/mRnd/).²⁴ For the binary outcome (HF), we roughly estimated the minimum detectable OR after entering the desired parameters in mRnd (in this study, α = 0.05; R ² = 0.36).

Results

Instrumental Variables Validity

We used 17 SNPs as IVs for estimating the causal effect of AF on HF. These SNPs explained 35.6% (R ²) of the variation in AF. In our MR analysis, all the SNPs had high F-statistics ranging from 266.43 to 6110.92, indicating that they were not weak instruments. Table 2 shows the effect estimates of each SNP on both AF and HF. Based on power calculations, our study had 100% power to detect ORs between 1.10 and 1.19 for the association between AF and HF.

Mendelian Randomization Analysis

The results of the 2-sample MR analyses are presented in Table 2. Using the random-effect IVW models, we found that AF was associated with a significant increase in the risk of HF (OR = 1.14, 95% CI: 1.10-1.19, P < .001). This association was consistent across the MR-RAPS, MR-Egger, weighted mode, and weighted-median analysis (Table 3, Figure 1).

Additionally, to assess the potential reverse causality, a bidirectional MR analysis was conducted. The results, presented in Supplementary Table 2, did not reveal any significant causal effect of HF on the risk of developing AF, suggesting the observed relationship is unidirectional from AF to HF. This finding further strengthens the evidence for a causal role of AF in the development of HF within the East Asian population studied.

Validation of Sensitivity Analysis

We performed several sensitivity analyses to assess the potential heterogeneity and pleiotropy between exposure and outcome. Neither the Cochran’s Q-test nor the MR-Egger regression analysis detected any heterogeneity or horizontal pleiotropy (P _h = .27, P _intercept = .28) (Table 3). Moreover, the MR-PRESSO results and the radial plot did not identify any outlier SNPs (Figure 2). The leave-one-out analysis showed that the overall effect of AF on HF was not influenced by any single SNP (Figure 3).

Discussion

Epidemiological studies have identified several cardiovascular risk factors that are strongly associated with the development of HF. One of the most common and potent risk factors for HF is AF. However, the association between AF and HF may be confounded by several factors, such as aging, CVD, hypertension, VHD, BMI, and COPD, which are prevalent in both conditions. Using MR, we demonstrated for the first time a causal relationship between AF and HF in East Asian populations. Our sensitivity analysis showed that the causal effect was not significantly affected by other established risk factors for both conditions, such as CVD, hypertension, and VHD. According to the last randomized trial EAST-AFNET4 and CABANA sub-analysis early rhythm control performed with catheter ablation of AF improves survival and prevents HF.²⁶ Our results are in line with other MR studies and support the hypothesis that HF can be prevented by controlling AF. Previous studies have shown that AF can lead to left-ventricular systolic dysfunction and increase the mortality and morbidity of HF.⁵ Therefore, our findings imply that public health interventions are needed to emphasize the importance of adequate AF management in reducing the global burden of HF and its severe complications.

The mechanisms underlying the pathogenesis of HF in AF patients may involve several pathways. Atrial fibrillation is associated with adverse hemodynamic changes, such as loss of atrial systole, ventricular rate irregularity, and chronotropic incompetence. Under normal sinus rhythm, atrial contraction contributes 20%-25% of the total left ventricular stroke volume.²⁷ The loss of atrial contractility can precipitate HF, especially in cases where the ventricular filling is compromised, such as in CVD or hypertension. Atrial fibrillation also causes a decrease in cardiac output, which is accompanied by an increase in neurohumoral vasoconstrictors. The activation of neurohumoral systems is a hallmark of HF and is associated with left ventricular dysfunction.²⁸ Tachycardia-mediated cardiomyopathy is a common consequence of AF.^29,30 Increased heart rates are associated with abnormal calcium signaling between the surface membranes of cardiomyocytes and the sarcoplasmic reticulum, as well as reduced sarcoplasmic calcium levels. This results in decreased myocardial contractility and dilatation. The effect of AF on atrial function has also been extensively studied. Chronic AF causes intracellular oxidative stress, which leads to calcium overload and the initiation of the inflammatory cascade. This causes both the persistence of the arrhythmia and the remodeling of the atrium, resulting in fibrosis. This chronic condition is known as “atrial cardiomyopathy” in the medical literature.³¹ Patients with highly symptomatic AF have been shown to have diffuse left ventricular (LV) fibrosis.³² There was a positive correlation between the degree of ventricular fibrosis and AF burden, independent of age or systolic dysfunction. Even patients with lone AF had diffuse LV fibrosis compared to healthy controls. Based on our findings, we suggest that physicians can take several measures in clinical practice to improve patient outcomes. Specifically, physicians should be aware that patients with AF have an increased risk of developing HF and should monitor them closely for signs and symptoms of HF. Moreover, physicians can work to identify and treat the underlying causes of AF, such as hypertension, diabetes, and obesity. Treating these underlying conditions may help to prevent or delay the onset or progression of HF in patients with AF. Furthermore, physicians can consider early intervention strategies for patients with AF, such as anticoagulation therapy and rhythm control, which may help to reduce the risk of HF and other adverse outcomes. Overall, our findings suggest that physicians should take a proactive approach to managing patients with AF in order to reduce the risk of HF and improve patient outcomes. Further studies are needed to confirm these findings and explore the clinical implications in more detail.

Our study has several notable strengths. First, we conducted a comprehensive analysis of incident HF and robust GWAS using a very large sample size to obtain genetic instruments for MR analysis. Moreover, our MR method yielded more reliable effect estimates than conventional observational studies by minimizing the confounding and reverse causation. Last, a rigorous process was applied to select and validate the IVs, reducing the bias caused by inappropriate IVs.

However, our study also has some limitations. First, due to the scarcity of data resources, we could not perform stratified analyses or adjust for covariates. Second, since MR assumes a linear relationship between exposure and outcome, the nonlinear association between AF and HF risk could not be evaluated. Third, the dissemination of our findings to other populations was hampered by the East Asian ancestry of the samples.

Conclusion

Our TSMR analysis substantiates a genetic causal relationship between AF and HF risk among East Asian populations. The findings reinforce the clinical relevance of sinus rhythm preservation in individuals predisposed to HF. Additionally, our bidirectional MR analysis did not demonstrate a causal effect of HF on AF, suggesting a unidirectional causality from AF to HF. These insights necessitate further research to validate and elucidate the clinical ramifications of our study comprehensively.

Data Availability

This study used publicly available GWAS summary data that can be accessed from https://gwas.mrcieu.ac.uk/.

Footnotes

Ethics Committee Approval: No ethics approval was required as this study used public summary data.

Informed Consent: The data for this study is sourced from public datasets, and the patients’ written informed consent forms have already been signed in the original research.

Peer-review: Externally peer-reviewed.

Author Contributions: Conception – X.W.; Data Curation – X.W., Y.M.; Analysis – X.W., Y.M.; Investigation – F.Y., X.Y.; Methodology – X.W., F. Y.; Supervision – X.W.; Original Draft – X.W., Y.M.; Review and Editing – X.W.

Acknowledgments: The authors are grateful to all the researchers who shared their GWAS data publicly.

Declaration of Interests: The authors have no conflicts of interest to declare.

References

1. Santhanakrishnan R, Wang N, Larson MG. Atrial fibrillation begets heart failure and vice versa: temporal associations and differences in preserved versus reduced ejection fraction. Circulation. 2016;133(5):484-492. https://doi.org/10.1161/CIRCULATIONAHA.115.018614
2. Sossalla S, Vollmann D. Arrhythmia-induced cardiomyopathy. Dtsch Ärztebl Int. 2018;115(19):335-341. https://doi.org/10.3238/arztebl.2018.0335
3. Wang TJ, Larson MG, Levy D. Temporal relations of atrial fibrillation and congestive heart failure and their joint influence on mortality: the Framingham Heart Study. Circulation. 2003;107(23):2920-2925. https://doi.org/10.1161/01.CIR.0000072767.89944.6E
4. Pabel S, Sossalla S. Atrial fibrillation and heart failure: novel insights into the chicken and egg dilemma. Eur Heart J. 2022;43(36):3376-3378. https://doi.org/10.1093/eurheartj/ehac257
5. Willems S, Meyer C, de Bono J. Cabins, castles, and constant hearts: rhythm control therapy in patients with atrial fibrillation. Eur Heart J. 2019;40(46):3793-3799c. https://doi.org/10.1093/eurheartj/ehz782
6. Şaylık F, Çınar T, Akbulut T, Hayıroğlu Mİ. Comparison of catheter ablation and medical therapy for atrial fibrillation in heart failure patients: a meta-analysis of randomized controlled trials. Heart Lung. 2023;57():69-74. https://doi.org/10.1016/j.hrtlng.2022.08.012
7. Nesapiragasan V, Hayıroğlu Mİ, Sciacca V, Sommer P, Sohns C, Fink T. Catheter ablation approaches for the treatment of arrhythmia recurrence in patients with a durable pulmonary vein isolation. Balk Med J. 2023;40(6):386-394. https://doi.org/10.4274/balkanmedj.galenos.2023.2023-9-48
8. Smith GD, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations. Int J Epidemiol. 2004;33(1):30-42. https://doi.org/10.1093/ije/dyh132
9. Smith GD, Timpson N, Ebrahim S. Strengthening causal inference in cardiovascular epidemiology through Mendelian randomization. Ann Med. 2008;40(7):524-541. https://doi.org/10.1080/07853890802010709
10. Smith GD, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease. Int J Epidemiol. 2003;32(1):1-22. https://doi.org/10.1093/ije/dyg070
11. Yang S, Pudasaini R, Zhi H, Wang L. The relationship between blood lipids and risk of atrial fibrillation: univariable and multivariable Mendelian randomization analysis. Nutrients. 2021;14(1):-. https://doi.org/10.3390/nu14010181
12. Ma M, Zhi H, Yang S, Yu EYW, Wang L. Body mass index and the risk of atrial fibrillation: a Mendelian randomization study. Nutrients. 2022;14(9):-. https://doi.org/10.3390/nu14091878
13. Kwok MK, Schooling CM. Mendelian randomization study on atrial fibrillation and cardiovascular disease subtypes. Sci Rep. 2021;11(1):18682-. https://doi.org/10.1038/s41598-021-98058-w
14. Hemani G, Zheng J, Elsworth B. The MR-Base platform supports systematic causal inference across the human phenome. eLife. 2018;7():-. https://doi.org/10.7554/eLife.34408
15. Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 2017;13(11):e1007081-. https://doi.org/10.1371/journal.pgen.1007081
16. Low SK, Takahashi A, Ebana Y. Identification of six new genetic loci associated with atrial fibrillation in the Japanese population. Nat Genet. 2017;49(6):953-958. https://doi.org/10.1038/ng.3842
17. Pierce BL, Burgess S. Efficient design for Mendelian randomization studies: subsample and 2-sample instrumental variable estimators. Am J Epidemiol. 2013;178(7):1177-1184. https://doi.org/10.1093/aje/kwt084
18. Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37(7):658-665. https://doi.org/10.1002/gepi.21758
19. Zhao Q, Wang J, Hemani G, Bowden J, Small DS. Statistical inference in two-sample summary-data Mendelian randomization using robust adjusted profile score. Ann Statist. 2020;48(3):1742-1769. https://doi.org/10.1214/19-AOS1866
20. Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol. 2016;40(4):304-314. https://doi.org/10.1002/gepi.21965
21. Zheng J, Baird D, Borges M-C. Recent developments in Mendelian randomization studies. Curr Epidemiol Rep. 2017;4(4):330-345. https://doi.org/10.1007/s40471-017-0128-6
22. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512-525. https://doi.org/10.1093/ije/dyv080
23. Greco MF, Minelli C, Sheehan NA, Thompson JR. Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome. Stat Med. 2015;34(21):2926-2940. https://doi.org/10.1002/sim.6522
24. Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(5):693-698. https://doi.org/10.1038/s41588-018-0099-7
25. Sanderson E, Davey Smith G, Windmeijer F, Bowden J. An examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings. Int J Epidemiol. 2019;48(3):713-727. https://doi.org/10.1093/ije/dyy262
26. Palamà Z, Nesti M, Robles AG. Tailoring the ablative strategy for atrial fibrillation: a state-of-the-art review. Cardiol Res Pract. 2022;2022():9295326-. https://doi.org/10.1155/2022/9295326
27. Meisner JS, Keren G, Pajaro OE. Atrial contribution to ventricular filling in mitral stenosis. Circulation. 1991;84(4):1469-1480. https://doi.org/10.1161/01.cir.84.4.1469
28. Hartupee J, Mann DL. Neurohormonal activation in heart failure with reduced ejection fraction. Nat Rev Cardiol. 2017;14(1):30-38. https://doi.org/10.1038/nrcardio.2016.163
29. Dandamudi G, Rampurwala AY, Mahenthiran J, Miller JM, Das MK. Persistent left ventricular dilatation in tachycardia-induced cardiomyopathy patients after appropriate treatment and normalization of ejection fraction. Heart Rhythm. 2008;5(8):1111-1114. https://doi.org/10.1016/j.hrthm.2008.04.023
30. Jais P, Sanders P, Hsu LF, Hocini M, Haissaguerre M. Catheter ablation for atrial fibrillation. Heart (Br Card Soc). 2005;91(1):7-9. https://doi.org/10.1136/hrt.2003.030205
31. Bergau L, Bengel P, Sciacca V, Fink T, Sohns C, Sommer P. Atrial fibrillation and heart failure. J Clin Med. 2022;11(9):-. https://doi.org/10.3390/jcm11092510
32. Ling LH, Kistler PM, Ellims AH. Diffuse ventricular fibrosis in atrial fibrillation: noninvasive evaluation and relationships with aging and systolic dysfunction. J Am Coll Cardiol. 2012;60(23):2402-2408. https://doi.org/10.1016/j.jacc.2012.07.065