Cardiovascular diseases (CVDs) remain the leading cause of morbidity and mortality globally (1). The most common clinical manifestations of CVDs include heart failure (HF), coronary artery disease (CAD), cerebrovascular disease and atrial fibrillation (AF). Although the nature of CVDs is complex and not fully understood, common risk factors, such as hypertension (2), obesity (3), hyperlipidemia (4), and smoking (5), have been recognized as contributing to the early management of CVDs. In addition, observational studies have further suggested that atopic dermatitis (AD) is a potential risk factor for CVDs (6).

AD (also known as atopic eczema) is a common inflammatory skin disease, which typically begins in childhood and affects 5–10% of adults (7). A meta-analysis of longitudinal cohort studies showed that participants with AD experienced an increased risk of HF, myocardial infarction (MI), and stroke compared to those without AD (6). Cross-sectional studies have further revealed that CVDs were more prevalent in patients with AD (8). In addition, the risk of CVDs appeared to increase in patients with more severe and recalcitrant AD (6). However, the results from observational studies are inconclusive regarding the causal effects of AD on CVDs given to biases such as residual confounding, misclassification, and reverse causality (9). In addition, observational studies may not reveal the association of life-long exposure to AD with CVDs risk.

Recognition of the causal links between genetic predisposition to AD and CVDs may provide new avenues for the management of CVDs. In the present study, we conducted a Mendelian randomization (MR) study to clarify the causal associations between AD and CVDs. MR is a powerful approach to evaluate the causal associations between exposures and outcomes by using genetic variants as instrumental variables (IVs) (10). Given the genetic variants that are randomly assorted and fixed at conception, the approach diminishes the influence of potential confounding factors and reverse causality, thereby strengthening the causal inference (10). Here, we sought to explore the causality in the association of genetically determined AD with CVDs (CAD, MI, AF, HF, and stroke).

Characteristics of the genetic associations of IVs with AD and CVD outcomes are shown in Supplementary Table 2. The included 19 SNPs were estimated to account for 5.63% of the phenotypic variability of AD. All of the SNPs had F-statistics higher than the threshold of 10, indicating a minor weak instrument bias in the present study.

In the standard IVW analyses, genetic predisposition to AD was causally associated with a higher risk of HF in the HERMES (odds ratio (OR), 1.06; 95% confidence interval (CI), 1.02–1.10; P = 0.004; Figures 1, 2). The association of AD with HF risk was replicated in the UK Biobank Heart Failure GWAS (OR, 1.10; 95% CI, 1.02, 1.18; P = 0.010; Figures 1, 2) and persisted in the meta-analysis (OR, 1.07; 95% CI, 1.03, 1.10; P = 1.11 × 10⁻⁴; Figures 1, 2). In the complementary analyses, the causal effect of AD on HF remained consistent in the weighted median and MR-Egger methods (Table 2).

For other CVDs (AF, CAD, MI, AS, IS, LAS, CES, and SVS), patients with AD may not have an increased risk of these outcomes compared to the controls (Supplementary Figure 1). The null effects of AD on these diseases were further supported by the complementary analyses results (Supplementary Figure 1).

For all considered outcomes, the MR-Egger intercept was close to zero (P_intercept > 0.05), indicating a low level of directional pleiotropy in the present study (Supplementary Table 4). No evidence of heterogeneity was found except for AS (PCochran′sQ = 0.03; Supplementary Table 4). Importantly, the MR-PRESSO method did not detect any horizontal pleiotropic outliers in all reported results (Supplementary Table 4).

Furthermore, rs61813875 was found to be different from the rest of the SNPs (Figure 2). It had the largest association with AD and HF. Therefore, a sensitivity analysis without this SNP was carried out to test the robustness of the results. The standard IVW analyses showed that the total effect of AD on HF remained significant after excluding rs61813875 (OR, 1.05; 95% CI, 1.01, 1.09; P = 0.009; Supplementary Figure 2). The association remained directionally consistent in the complementary analyses, albeit with wider CIs (Supplementary Table 5).

In the present study, two-sample MR analyses demonstrated that genetic predisposition to AD is causally associated with a higher risk of HF. However, no significant associations between AD and AF, CAD, MI, stroke, and stroke subtypes were found. The results were consistent and robust in complementary analyses using the weighted median and MR-Egger methods. To the best of our knowledge, this is the first MR study to explore the causal associations between AD and CVDs.

The observation that AD may be associated with CVDs and their risk factors (8, 30, 31) has prompted investigators to look for more evidence to confirm this association. The present results regarding the AD-HF association were consistent with a previous cohort study that identified AD as an independent risk factor for HF (hazard ratio (HR) 1.46, 95% CI, 1.10, 1.93) (32). The positive association was further supported by a large cohort study which reported that patients with AD experienced a higher risk of HF (HR 1.19, 99% CI, 1.09, 1.30) independent of common cardiovascular risk factors (33). Mechanistic work suggested that AD may exert a procoagulant effect (34), contributing to CAD and MI and ultimately leading to HF. However, the associations between AD and CAD or MI were inconsistent in the literature. An analysis of three cross-sectional surveys in U.S. populations found a 2.5-fold increased prevalence of CAD in patients with AD compared to their controls. The associations remained significant after adjustment for body mass index, smoking, alcohol consumption, and physical activity (8). Conversely, studies from Canada, Germany, Denmark, and the U.S.A. found either no associations or negative associations between AD and MI (35–38). In addition, the present MR study suggested that AD was not causally associated with MI, CAD, or AF, leaving the question of how AD can lead to HF open.

Disrupted barrier function was commonly observed in affected and unaffected skin of atopic subjects (39). This condition does not only lead to typical skin dryness and physical damage from scratching, but also induces inflammatory responses and overproduction of reactive oxygen species (39, 40). These pathophysiology drivers can promote and interact with others. Also, chronic skin inflammation and oxidative stress can cause other cutaneous and non-cutaneous diseases (40–43). As inflammation and oxidative stress are also involved in the pathogenesis of HF (44, 45), these conditions are likely to mediate the causal effect of AD on HF. On the other hand, clinical data showed that hypertension was a common comorbidity of AD (46). Prolonged pressure overload-induced by hypertension can lead to pathological cardiac hypertrophy and progression to HF (47). Therefore, hypertension is likely to, at least in part, explain the observed association between AD and HF.

The correlation between AD and stroke remains inconclusive so far. A large cohort study showed AD to be associated with an increased risk of stroke (HR 1.10; 99% CI, 1.02, 1.19) (33). However, a meta-analysis including eight related studies reported that AD may not be a relevant risk factor for stroke (OR 1.15; 95% CI 0.95–1.39) (48). The results of non-significant associations were in agreement with those obtained in the present MR study. Genetic liability to AD was not causally associated with the risk of stroke or its subtypes. The positive results observed in some observational studies may be attributed to residual confounders or reverse causality.

Anti-inflammatory therapy and barrier function improvement are recommended for the prevention and treatment of AD (39). Therapies like anti-hypertension and lipid-lowering and lifestyle changes, such as smoking cessation and appropriate sleep duration, can improve the management of HF and other CVDs (49–52).

The present study had several notable strengths. First, the MR approach was less susceptible to potential confounding factors and reverse causation, thus reinforcing the causal inference. Second, complementary analyses returned effect estimates similar to those obtained in the main analysis, thus strengthening the evidence. Third, a meta-analysis combining the results from two GWASs increased the power to detect an AD-HF association. Finally, the bias introduced by population structure was less likely to affect the results as the analyses were restricted to individuals of European ancestry.

However, there were several limitations to consider. First, the population in the present study, as mentioned above, was restricted to European ancestry. Therefore, the generalizability of study conclusions might be limited. Future studies on other populations are therefore warranted. Second, some psychiatric traits (such as major depressive disorder (MDD) and neuroticism) shared genetic liability with AD and CVDs (53–55). Specifically, there is a bidirectional causal association between MDD and AD (53), and MDD can lead to a higher risk of CVDs (54). Therefore, the influence of potential pleiotropy remains a concern in this MR study, despite the lack of evidence from MR-Egger regression and MR-PRESSO methods. Third, due to the low variance explained by the SNPs and low sample size for some of the outcomes, the statistical power for the null associations may be insufficient. Thus, these negative results should be interpreted with caution. Finally, since the GWAS conducted by the EAGLE eczema consortium did not classify the severity of AD (12), we were not able to explore whether the risk of CVDs increases with increasing severity of AD, as several observational studies have suggested (6, 33, 56).

The present MR study provided genetic evidence showing a causal association between genetic predisposition to AD and HF. Prevention and early diagnosis of AD may help prevent HF. No causal associations between AD and the risk of AF, CAD, MI, stroke, and stroke subtypes were found. Further studies are warranted to gain more insight into the underlying mechanisms.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

The studies involving human participants were reviewed and approved by Local Ethics Committees of the UK Biobank project, CARDIoGRAMplusC4D, Nielsen et al. GWAS, HERMES consortium, the MEGASTROKE project, and the EAGLE eczema consortium. No additional ethical approval are required.

HC: study conception and design. HC and CZ: data analyses and draft preparation. LZ: supervision of the study. All authors contributed to the article and approved the submitted version.

This work was supported by the National Natural Science Foundation of China [Grant Number 81873484], the Youth Program of National Natural Science Foundation of China [Grant Number 82000316], and the Nature Science Foundation of Zhejiang Province [Grant Number LZ16H020001].

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.