Genetically Determined Birthweight Associates With Atrial Fibrillation: A Mendelian Randomization Study

Atrial fibrillation is a common arrhythmia, characterized by irregular electrical activity in the upper chambers of the heart (atria). Large population studies estimate a worldwide prevalence of ≈0.5% with >33 million affected individuals.¹ The irregular cardiac conduction resulting from atrial fibrillation can lead to significant complications, including cardioembolic stroke, heart failure, and death. Many chronic cardiometabolic risk factors contribute to the development of atrial fibrillation, including hypertension, diabetes mellitus, heart failure, thyroid disease, chronic kidney disease, smoking, obesity, cardiac surgery, coronary heart disease, and valvular heart disease.² Genetic risk factors also contribute to the development of atrial fibrillation. Heritability of atrial fibrillation is estimated at 20%, and family linkage studies and genome-wide association studies have identified both rare, coding, large-effect monogenic variants, and more common, small-effect variants that contribute to the development of atrial fibrillation.^3–10

Birthweight has been identified as a risk factor for a number of cardiometabolic diseases, including atrial fibrillation, coronary heart disease, obesity, hypertension, and diabetes mellitus.^11–14 The relationship between birthweight and risk of atrial fibrillation has previously been investigated in a number of population studies, including the Women’s Health Study,¹⁵ Atherosclerosis Risk In Communities,¹⁶ and the HBCS (Helsinki Birth Cohort Study).¹⁷ However, the cumulative results from this collection of studies have been conflicting. Both low and high birthweight have been associated with increased risk for atrial fibrillation (Table I in the Data Supplement).^15–18 We hypothesized that a Mendelian randomization approach may help clarify the relationship between these 2 traits.

Birthweight, like many anthropometric traits, is highly heritable. Large-scale birthweight GWAS have identified over 60 genome-wide significant loci and have explained 15% of the variation in birthweight.^11,12,19 Genetic studies of birthweight and other health-related traits have identified a negative genetic correlation between birthweight and type 2 diabetes mellitus, systolic blood pressure, and coronary artery disease, and a positive association with anthropometric and obesity-related traits such as body mass index.¹³

In the current study, we utilized birthweight genetics to study the impact of birthweight on atrial fibrillation. We used 2-sample Mendelian randomization to consider the impact of birthweight on incident atrial fibrillation using summary data from the Early Growth Genetics Consortium GWAS of birthweight (N=153 781) and a large biobank-based GWAS of atrial fibrillation (n=60 620 cases and n=970 216 controls).^4,12

The data, analytic methods, and study materials may be made available to other researchers for purposes of reproducing the results or replicating the procedure. The study was approved by the Institutional Review Board of the University of Pennsylvania. Full Methods are available in the Data Supplement of the article.

We performed 2-sample Mendelian randomization (Tables II and III in the Data Supplement), using summary statistics from a genome-wide association study of atrial fibrillation including >1 million individuals.⁴ Mendelian randomization leverages genetic instrumental variables to provide unconfounded causal estimates of the effects of one or exposure on another. We constructed a genetic instrument composed of 42 single nucleotide polymorphisms (SNPs) that had previously been strongly associated with birthweight at a genome-wide level of significance (P<5×10⁻⁸; Table IV in the Data Supplement). After removal of heterogenous variants (n=3), inverse variance-weighted modeling revealed a significant (odds ratio, 1.27 [95% CI, 1.14–1.41]; P=1×10⁻⁵) association between increasing birthweight and increased risk of developing atrial fibrillation (Figure 1; Figures I through III in the Data Supplement). This association remained significant in sensitivity analysis using weighted median and weighted mode. The intercept from Egger regression was −0.001 (P=0.8), suggesting lack of significant pleiotropic bias. In additional sensitivity analysis, an instrument composed of 39 SNPs associated with birthweight was investigated for association with atrial fibrillation using summary statistics from the AFGen consortium genome-wide association study of atrial fibrillation, with similar results (Table V in the Data Supplement).

Asymmetry analysis was performed to detect any bias from reverse causality. A genetic instrument comprised of 98 SNPs strongly associated with atrial fibrillation (P<5×10⁻⁸) was tested for association with birthweight. No significant associations were identified by any Mendelian randomization method (inverse variance-weighted, weighted median, weighted mode, simple mode, and MR-Egger).

We queried published GWAS studies to determine whether variants included in the genetic instrument for birthweight were associated with risk factors for atrial fibrillation. Variants included in the instrument had previously been associated with anthropometric, metabolic, lipid, and pulmonary traits, among others (Table VI in the Data Supplement).

A prior Mendelian randomization study¹³ previously identified associations between birthweight and known atrial fibrillation risk factors including coronary heart disease,²⁰ diabetes mellitus,²¹ body mass index,²² and LDL-cholesterol.²³ We performed multivariable Mendelian randomization to determine whether shared genetic risk among these traits may mediate the relationship between birthweight and atrial fibrillation. After adjustment for shared genetic risk factors, birthweight remained significantly associated with risk of atrial fibrillation (odds ratio, 1.32 [95% CI, 1.2–1.5]; P=5×10⁻⁶; Figure 2).

The current study highlights the use of human genetics to clarify prior epidemiological observations that birthweight is significantly associated with the risk of developing of atrial fibrillation. A number of large epidemiological studies including Atherosclerosis Risk in Communities, WHS (Women’s Health Study), and HBCS previously investigated the relationship between birthweight and atrial fibrillation. We used a genetic approach to further refine this association by employing 2 sample Mendelian randomization using summary statistics from large genome wide association studies of birthweight and atrial fibrillation including >1 million individuals. Our results show that the genetic risk of increased birthweight, as assessed using a genetic instrument composed of 42 independent genome-wide significant SNPs, was strongly associated with an increased risk of atrial fibrillation.

Our findings are consistent with those of the 2010 analysis of the WHS. Analysis of data from 27 982 women with median follow-up of 14.5 years identified a significant linear relationship between self-reported birthweight and atrial fibrillation, with women in the highest birthweight category (>4.5 kg) at 71% increased risk of incident atrial fibrillation compared with women in the lowest birthweight category (<2.5 kg), after adjustment for age, hypercholesterolemia, smoking, exercise, alcohol consumption, education, race/ethnicity, and hormone replacement therapy.¹⁵

Mendelian randomization analysis applied to birthweight and cardiovascular outcomes has previously been investigated in the UK BioBank, finding strong associations between low birthweight and type II diabetes mellitus and coronary artery disease.¹³ In that same study, Mendelian randomization was also used to examine the association between birthweight and atrial fibrillation, but the genetic instruments and outcomes were derived from older genome-wide associations of birthweight and atrial fibrillation.^12,24 Using summary statistics from larger studies of birthweight and atrial fibrillation, we refine the magnitude and confidence around the effect estimate (inverse variance weighted [IVW] odds ratio, 1.27 [95% CI, 1.14–1.41]) versus (IVW odds ratio, 1.15 [95% CI, 0.95–1.39]).

Our results must be interpreted with caution. In MR-Egger and MR-PRESSO analysis, used to assess for and account for violations of the horizontal pleiotropy assumption of Mendelian randomization, we found no significant evidence of pleiotropic bias.^25–27 However, because of the complex relationship between parental genetics and diseases/risk factors associated with atrial fibrillation, birthweight may not necessarily be the causal exposure through which our genetic instrument increases the risk of atrial fibrillation.²⁸ Although we did not find evidence of horizontal pleiotropy in our MR analysis, our query of published GWAS results revealed associations between birthweight variants and multiple anthropometric and cardiometabolic risk factors, which may in part mediate the relationship between birthweight and atrial fibrillation. A combined observational and Mendelian randomization study of birthweight and cardiometabolic disease in UK Biobank found that increased birthweight protects from some traditional atrial fibrillation risk factors like coronary artery disease and diabetes mellitus, while also being positively associated with other atrial fibrillation risk factors like body mass index.¹³ In multivariable Mendelian randomization analysis, birthweight remained significantly associated with atrial fibrillation despite adjusting for the shared genetic component among these risk factors. While horizontal pleiotropic effects may plausibly exist between birthweight and atrial fibrillation, we were unable to identify these effects despite extensive Mendelian randomization sensitivity analyses. In sum, these results highlight the complex relationship between the genetic and clinical risks birthweight, cardiometabolic, and anthropometric risk factors that converge to increase atrial fibrillation risk.

Birthweight is only one proxy for fetal growth and the intrauterine environment. Instrumental variables reflecting more specific intrauterine exposures would help refine our understanding of the developmental origins of atrial fibrillation.²⁹ As maternal and fetal genetic variation are correlated, further study of maternal-fetal pairs with both genetic and longitudinal outcome data of the offspring may further clarify this association. Recent work by Warrington et al^30,31 identify the contributions of fetal and maternal genetic influences on birthweight may also help parse the direct genetic effect of variants contributing to both birthweight and atrial fibrillation from the indirect maternal effects.

Our Mendelian randomization findings remained robust to sensitivity analysis and showed no evidence of horizontal pleiotropy.^25,26 Our findings were limited to individuals of European ancestry because of lack of large-scale genome-wide association studies of birthweight in populations of diverse ancestry, which may also limit generalizability. Studies in these populations are clearly needed.

Overall, we used genetic analysis to provide clarity into the relationship between birthweight and risk of atrial fibrillation, finding a robust positive association between these traits. Our results provide further support to the growing body of evidence that intrauterine growth has a lifelong impact on health.