Atrial fibrillation (AF) is the most common cardiac arrhythmia in clinical practice, with an incidence rate of 2.3%–3.4%, which increases with age (1). AF can increase the risk of stroke and systemic thromboembolism in patients, and the risk of heart failure increases by 1.5–2.0 times (2). It has brought a heavy burden to patients, society and economy (3). Although many studies have been done in recent years to explore the pathophysiological mechanism and treatment methods of AF, the prevention and treatment of AF is still a major problem in clinical medicine.

Metabolites are considered as the final reaction of the biological system to inherited or environmental variations, and their levels directly reflect the physiological state of the body (4). Human blood metabolites (intermediates of metabolism), vary greatly among different individuals. Cell-free metabolites (from plasma or blood) are usually used to explore their biological significance and function because of the convenience of sample collection.

Metabolomics is an important part of system biology (5). It explores biomarkers related to disease diagnosis and prognosis by detecting small molecular substances in biological samples, which has been widely used in clinical and experimental studies (6, 7).

There is now growing interest in the relationship between metabolites and AF. Many studies have found that there is a potential relationship between metabolites and AF, indicating that some metabolites take part in the pathogenesis and development of AF (8–12). Harskamp et al. found that medium chain acylcarnitines, short chain dicarboxylacylcarnitines and long chain acylcarnitines were associated with AF (8). Yan et al. found that metabolites such as potassium and sodium ion were associated with AF (9). Harju et al. found 61 metabolites associated with AF, including energy, histidine, glutathione, purine, sugar, lipid and so on (10). However, as far as we know, a comprehensive and systematic research evaluating the causal effects of blood metabolites on AF is still lacking.

Mendelian randomization (MR) is an epidemiological method that uses single nucleotide polymorphisms (SNPs) as exposure-related instrumental variables (IVs) to assess the causality between exposure and outcome (13). In the absence of randomized clinical trials (RCT), MR research is a significant alternative method to analyze the causal relationship, because MR analysis, based on the random classification theory in the process of meiosis in genetics, carried out random grouping similar to RCT in the population (14). Diseases usually occur after the formation of genotype, so genotype is difficult to be affected by diseases progression, so it is less susceptible to reverse causality (13), thus overcoming the inherent bias and confounding factors of observational studies (15, 16).

The study used the statistical data of the genome-wide association study (GWAS) as the data source and the two-sample MR analysis as the research method to evaluate the causality between blood metabolites and AF, selected the candidate blood metabolites related to AF, providing a novel perspective for the early diagnosis and treatment strategy of AF.

Among 486 blood metabolites, we first identified 5 metabolites that had potential causal relationships with AF, including 3 known metabolites (tryptophan betaine, uridine, lactate) and 2 unknown metabolites (X-12189, X-12717). Then we performed the replication analysis on another GWAS dataset of AF to verify the results, and finally found 2 overlapping metabolites, among which tryptophan betaine was statistically significant and uridine was nominally significant.

AF is a common, complex and age-related arrhythmia that occurs in almost all organic and non-organic heart diseases. The high incidence and recurrence rate of AF have caused a great burden on people, so early diagnosis and prevention of AF are very important. Although many studies have explored the pathogenesis and pathological mechanisms of AF, the specific mechanism is not yet fully understood due to its complexity.

Tryptophan betaine, also known as lenticin or hypaphrine, is essentially an indole alkaloid, which is the N-methylated form of tryptophan. As an important active ingredient, it has played a very good role in anti-inflammatory, especially in endothelial cell injury. Sun et al. found that tryptophan betaine could inhibit endothelial cell inflammation induced by lipopolysaccharide (32–34). Ding et al. found that tryptophan betaine inactivated p38/JNK signal pathway by up-regulating DUSP1, thus preventing lipopolysaccharide-mediated inflammatory response (35). In addition, some studies found that tryptophan betaine analogues could prevent the changes of blood potassium and sodium ion concentrations during coronary artery occlusion/reperfusion injury in myocardial ischemia rats, and could effectively reduce myocardial reperfusion injury and infarction area (36). AF is the result of multiple mechanisms, including oxidative stress, autonomic nervous response, inflammation response and so on, and inflammation response is one of the key factors leading to the pathogenesis of AF (37, 38). Our results suggested that tryptophan betaine was negatively associated with the risk of AF. However, there were few previous studies on the correlation between tryptophan betaine and AF. According to previous research results about tryptophan betaine, it is speculated that tryptophan betaine may reduce the risk of AF through anti-inflammatory mechanism, but the exact biological mechanism needs to be clarified by further experimental studies.

Uridine, composed of uracil and D-ribose, is a component of RNA and belongs to pyrimidine nucleosides. As a necessary nucleoside, uridine exists in many biological fluids. Uridine is involved in some metabolic disorders, such as dhydropyrimidinase deficiency, mitochondrial neurogastrointestinal encephalopathy, beta-ureidopropionase deficiency and Lech-Nyhan syndrome, etc. Some studies found that uridine had protective effects on cardiovascular diseases. Early studies by Krylova et al. found that uridine and uridine-5′-monophosphate could reduce lactic acid, improve energy metabolism, and playe a protective role in myocardial ischemia and arrhythmias (including ventricular premature, ventricular tachycardia and ventricular fibrillation) (39, 40). Recent studies found that uridine could prevent myocardial injury in ischemia and reperfusion rat models by activating mitochondrial pathway, prevented the decrease of cell energy supply and antioxidant system activity, thus improving energy supply (41). Our study found that uridine may be negatively associated with AF, which is consistent with previous findings indicating the cardioprotective effect of uridine. Myocardial ischemia is an important cause of AF (42), and mitochondrial dysfunction is an important pathogenesis of AF (43), so uridine may improve mitochondrial function by regulating energy metabolism, thus improving myocardial ischemia to reduce the risk of AF. Nevertheless, it is still lack of sufficient experimental evidence.

Lactate was found to have a possible causal relationship with AF in the initial MR analysis. Although it was not finally verified in the replication analysis, its relationship with AF is still worth exploring. Lactate is an organic acid that is a chiral molecule composed of two optical isomers, L-lactic acid and D-lactic acid. Lactate is continuously produced from pyruvate through lactate dehydrogenase (LDH) during normal metabolism and exercise fermentation. Plasma lactate is the final product of anaerobic metabolism of glucose and a sign of the imbalance between oxygen supply and demand in the body. Previous studies (44, 45) showed that plasma lactate was closely associated with AF, which played a potential role in the maintenance and recurrence of AF. When the plasma lactate increases excessively, the heart is in a state of hypoxia and the energy supply of myocardial cells is insufficient. This can stimulate the production of reactive oxygen species, lead to oxidative stress and mitochondrial controlled apoptosis, cause myocardial structural remodeling (46), and promote the occurrence and development of AF. In our study, lactate was positively correlated with AF, which was consistent with the above studies.

The mechanisms of AF may be a combination of multiple factors. This study found a causal relationship between some blood metabolites and AF, which helps us understand the pathogenesis of AF. However, we cannot ignore other risk factors related to AF. For example, in the Brugada syndrome, the underlying mechanism of AF may include genetic or acquired factors that may affect autonomic nervous function, atrial structure, conduction velocity, etc. (47). In the hypertrophic cardiomyopathy, AF is linked to advanced diastolic dysfunction, left atrial dilation, and remodeling (48). Moreover, AF may be precipitated by hypertension, hyperthyroidism (49), lifestyle factors such as endurance sport (50), alcohol consumption, and even smoking (51) (see Figure 5).

Our study aims to investigate whether there is a causal relationship between blood metabolites and AF. Through relatively rigorous MR analysis, we obtained some blood metabolites that have a potential causal relationship with AF, and determined the possible metabolic pathways through metabolic pathway analysis. This is exploratory research and analysis, which lays a solid foundation for further targeted metabolomics research in the future.

The study had some strengths. First, this study covered a wide range of human blood metabolites (486 in total) as exposure factors for MR analysis, aiming to study the metabolic characteristics leading to AF. Second, MR method effectively overcame the defects of traditional observational studies, and avoided reverse causation and residual confounding to a large extent. Third, after MR analysis of two different GWAS datasets of AF, overlapping blood metabolites were screened out, which could better explain the causal effect of these metabolites on AF, providing more reliable evidence. Fourth, the study used Bonferroni's corrected P value due to multiple testing, thus the results were more scientific and rigorous.

However, the study had several limitations. First, because the number of SNPs meeting genome-wide effect was limited, the P value was relaxed, but the F statistic of all SNPS was greater than 10 to exclude weak IVs, which is a commonly used strategy. Second, given the classification of the original datasets, it was impossible to further subdivide the clinical types of AF, so only an overall analysis could be performed. Third, all of the GWAS databases in this article are from Europe, so it is not clear whether the same results would be produced and extended to other ethnic groups. Further research is needed in the future to evaluate its generality in other ethnic groups. Fourth, although MR method has been proven to be an effective tool for assessing the causality between the exposure and the outcome, the findings need to be further studied depending on some experimental observations to reveal their significance in the development of AF.

In summary, this study used the two-sample MR analysis to analyze the effects of 486 blood metabolites on the risk of AF. The results indicated that there was a significant negative correlation between tryptophan betaine and the risk of AF. While uridine might have a potential causal relationship with AF. This study strengthens our understanding of the relationship between blood metabolites and AF, hoping to provide more useful information for future research on the pathogenesis of AF.