Bidirectional causal association and shared anti-inflammatory target of asthma and atopic dermatitis: a Mendelian randomization and colocalisation study

Abstract

Background

To investigate the causal relationship between asthma and atopic dermatitis (AD), and to identify potential therapeutic targets for their co-treatment.

Methods

All summary statistics were sourced from publicly available genome‑wide association studies. Two‑sample Mendelian randomization (MR) analysis was employed to assess the causal relationship between asthma and AD, as well as the causal effects of 91 circulating inflammatory proteins on both conditions. Rigorous sensitivity analyses were conducted to verify the robustness of the findings. Colocalisation analysis was employed to determine whether the exposure and outcome were driven by shared single nucleotide polymorphisms. Additionally, molecular docking was used to validate the interactions of natural active ingredients with interleukin-6 (IL-6).

Results

In the discovery cohorts, the MR results revealed an estimated effect of asthma on AD (OR = 1.224, 95% CI = 1.172–1.278, P = 5.89e-20) and an estimated effect of AD on asthma (OR = 1.229, 95% CI = 1.167–1.294, P = 4.83e-15). The MR analysis of replication cohorts yielded consistent results. IL-6 was identified being associated with both asthma and AD. Colocalisation analysis revealed that a shared causal variant within the selected gene locus was significantly associated with IL-6 and AD. However, no evidence was found to support the association of the shared variant with AD and asthma, or with IL-6 and asthma. Four natural active ingredients that interact with IL-6 were identified through molecular docking. IL-6 and atractylenolide I exhibited the lowest binding energy (-7.50 kcal/mol).

Conclusion

Our study demonstrated a bidirectional genetic correlation between asthma and AD. IL-6 may represent a therapeutic target for the concurrent management of both conditions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12890-025-04042-9.

Introduction

Asthma is one of the most prevalent chronic and non-communicable diseases, characterized by variable respiratory symptoms and airflow limitation [1]. Atopic dermatitis (AD) represents a chronic and non-infectious dermatosis, characterized by persistent pruritus of the skin [2]. Both conditions are widely recognized as allergic disorders and are prevalent among children and young adults, leading to impaired quality of life and a considerable treatment burden [3]. Previous studies have indicated that asthma and AD exhibit a comorbid relationship. Specifically, the risk of AD is elevated by 55.0% in individuals with asthma [4]. Conversely, the overall pooled prevalence of asthma among patients with AD is 25.7% [5]. Patients with persistent or severe AD are at an increased risk of developing asthma [6]. Nevertheless, these observational studies are frequently constrained by cross-sectional design and are susceptible to potential confounding factors; thus, the causal association between asthma and AD remains to be determined.

The etiologies of asthma and AD reflect a complex and incompletely understood interplay between genetic predisposition and environmental factors [1, 2], posing significant limitations on prevention and treatment. Numerous studies have revealed that inflammation is involved in the pathogenesis of both asthma and AD. Eosinophilic type 2 airway inflammation is present in over 50% of adults with asthma [7]. Sputum cytology has provided evidence of eosinophilic, neutrophilic, and mixed inflammatory phenotypes in asthma patients [8, 9]. Similarly, cutaneous inflammation is central to the pathogenesis of AD [10]. The inflammatory profile in AD is complex and diverse, involving the activation of skin-resident inflammatory dendritic cells, innate lymphoid cells, and Langerhans cells [11]. Inflammation compromises the epidermal barrier, exacerbating the primary defects of the barrier [2]. Lesional AD skin exhibits dysregulated expression of a broad spectrum of genes, predominantly related to keratinocyte activity and T-cell infiltration [12]. Corticosteroids, the cornerstone of treatment for asthma and AD, have demonstrated efficacy in attenuating the inflammatory immune response [13–15]. In animal models, anti-inflammatory interventions have played a beneficial role in alleviating lung injury in asthma and enhancing the epidermal barrier in AD [16–19]. These findings suggest that inflammation plays a crucial role in the onset and maintenance of asthma and AD. However, whether asthma and AD share potential therapeutic targets in inflammation remains unexplored, which could be highly beneficial for achieving concurrent treatment for them.

Mendelian randomization (MR) is a methodology for evaluating the causality between an exposure and an outcome by employing genetic variants as instrumental variables (IVs) [20], given that genetic variants are single nucleotide polymorphisms (SNPs) independent of confounding factors and reverse causality [21]. The aim of this study was to investigate the bidirectional causal relationship between asthma and AD, and to explore shared anti-inflammatory targets for their treatment. Utilizing MR analysis, we identified the existence of a bidirectional genetic relationship between asthma and AD. Interleukin-6 (IL-6) emerged as a potential shared inflammatory target for the treatment of both asthma and AD. Furthermore, we identified four natural active ingredients that target IL-6 through molecular docking. Our study provides valuable insights for the development of more effective and targeted therapies for asthma and AD.

Materials and methods

Study design

This study comprised three major components (Fig. 1). Firstly, a two-sample MR analysis was conducted to investigate the bidirectional causal relationship between asthma and AD. Secondly, a two-sample MR analysis was utilized to detect the genome-wide association of 91 inflammatory proteins with asthma and AD. Finally, molecular docking was performed on the shared inflammatory proteins of asthma and AD to evaluate their binding energy and interaction patterns with natural active ingredients. This study was reported in accordance with the STROBE-MR guideline for the reporting of MR studies (Figure S1) [22].

Fig. 1.

Diagram of three major components in study design

Data sources

The datasets used in this study were retrieved from publicly accessible genome-wide association study (GWAS) datasets based on European population, including both males and females. The datasets were accessed on October 24th, 2024. The summary data for asthma (J10_ASTHMA_EXMORE, discovery cohort) and AD (L12_ATOPIC, replication cohort) were obtained from the FinnGen R10 (https://www.finngen.fi/en) [23]. GWAS summary statistics for AD discovery cohort, asthma replication cohort, and 91 circulating inflammatory proteins were sourced from previous studies [24–26]. Detailed information regarding the data can be found in the original articles and Table S1. The data employed in this study have been ethically approved, and informed consent has been obtained from participants in the original research. Ethics approval was not required for the current analysis, as all included GWAS data are publicly available.

Instrumental variables selection

A two-sample MR study was conducted based on three foundational assumptions [20]: (1) IVs are robustly associated with the exposure; (2) IVs are unrelated to any confounders; (3) IVs affect the outcome exclusively through the exposure. To ensure the selection of high-quality genetic IVs for each trait, a rigorous series of quality control measures were applied. Firstly, SNPs were selected using a threshold of p-value < 5.00e-8. Secondly, data from the European sample in the 1000 Genomes Project were filtered to retain SNPs based on linkage disequilibrium (LD) with a strict clump window (r² = 0.001 and kb = 10,000). Additionally, F-statistics were calculated to evaluate the strength of the IVs, utilizing the formula: F = R² × (N-2)/(1-R²), where N denotes the sample size and R² represents the proportion of the exposure explained by the IVs [27, 28]. R² was computed using the formula: R² = 2 × EAF × (1-EAF) × BETA², where BETA is the effect size of the SNPs and EAF is the effect allele frequency [29]. SNPs were considered strong instruments and retained if the F-statistic exceeded 10 [27–29].

During the harmonization of exposure and outcome data, palindromic SNPs with intermediate allele frequencies were removed [30]. The potential horizontal pleiotropy was evaluated using the MR pleiotropy residual sum and the MR-PRESSO global test [31]. Outliers were detected using MR-PRESSO and the Radial plot, and were subsequently removed one by one [31, 32]. The MR-PRESSO global test was repeatedly performed on the remaining SNPs until the p-value for the global test was insignifcant (p-value >0.05) [33]. The remaining SNPs were then used for the subsequent MR analysis.

Genetic correlation analysis

To elucidate the genetic relationship between asthma and AD, we performed a genetic correlation analysis using LD Score regression. This method quantifies the genetic contribution of each trait by examining the relationship between test statistics and LD [34]. Summary datasets were filtered according to the HapMap3 reference panel. SNPs were excluded if they were strand-ambiguous, duplicated, and had a minor allele frequency (MAF) of less than 0.01. For each variant, the Z-scores from Trait 1 were multiplied by the corresponding Z-scores from Trait 2. The genetic covariance was then estimated by regressing the product of these Z-scores against the LD score. The genetic correlation was obtained by normalizing the genetic covariance with respect to SNP-heritability.

Mendelian randomization analysis

The inverse-variance weighted (IVW) method with a random-effects model served as the primary approach to evaluate potential causal associations [35]. The IVW method estimates the causal effect for each SNP using the Wald Ratio method, achieving higher estimate accuracy and statistical power when the IVs adhere to the three core assumptions. Therefore, additional methods were employed to provide more robust estimates across a wider array of scenarios, including MR-PRESSO, MR Egger, weighted median estimator (WME), weighted model, and simple mode. MRPRESSO can identify pleiotropic outliers in multi-instrument summary-level MR test, particularly when horizontal pleiotropy is present in fewer than 50% of the IVs [31]. MR-Egger can detect some violations of the standard IVs assumption and allows all genetic variants to exhibit any form of pleiotropic effect [36]. WME can obtain a robust result when more than 50% of weights stem from invalid IVs [37]. Both weighted and simple mode require valid IVs contribute to the estimation of true causal effect; they exhibit greater robustness to a limited number of pleotropic variants compared to IVW and MR-Egger [37, 38]. The results of the MR analysis were considered to be consistent if they indicated the same direction of effect as estimated by each method (Figure S2).

Sensitivity analysis

The Cochran’s Q test was utilized to quantify the heterogeneity among the effects of IVs, a p-value >0.05 indicates the absence of heterogeneity [39]. MR-Egger regression intercept analysis was employed to evaluate horizontal pleiotropy [36]. A p-value >0.05 suggested the absence of pleiotropy, otherwise, the corresponding SNP was removed, and the causal estimation was reassessed. Given that the GWAS data for asthma and AD included mixed European cohorts (Table S1), MRlap was applied to reduce bias in causal effect estimation stemming from potential overlap between the exposure and outcome samples [40]. The Steiger test was conducted to infer the causal direction between the exposure and outcome, requiring an assumed direction and then evaluating whether this direction is consistent with causal effect [41]. Furthermore, a leave-one-out analysis was performed to assess the robustness of the causal effect [42].

Colocalisation analysis

Colocalisation analysis was conducted to assess whether the significant exposure and outcome share a common genetic causal variant within the gene region [43]. This analysis was performed by generating ± 200 kb windows centered around the SNPs used to instrument the exposure [44]. An approximate Bayes factor computation with default parameters was carried out to quantify the posterior probability that the identified exposure and outcome share the identical causal variant. The posterior probability (PP) was employed to quantify support for five hypotheses: PP.H0, no association with either exposure or outcome; PP.H1, association with exposure but not with outcome; PP.H2, association with outcome but not with exposure; PP.H3, association with both traits but driven by distinct variants; PP.H4, association with both traits and shared a causal variant [43]. A threshold of significance for colocalisation was established at a sum of PP for H3 and H4 (PP.H3 + PP.H4) ≥ 0.80 [45]. A PP for H4 (PP.H4) ≥ 0.80 was considered evidence supporting colocalisation with a shared causal variant [44].

Natural active ingredients selection

After conducting colocalisation analysis, IL-6 was identified as being associated with asthma and AD. Subsequently, natural active ingredients targeting IL-6 were retrieved from the TCMSP database (https://www.91tcmsp.com/#/database/). These ingredients were screened by pharmacokinetics with high gastrointestinal absorption and drug-likeness without violations in the SwissADME database (http://www.swissadme.ch/). The resultant candidates of natural active ingredients were selected for further processing.

Molecular docking

To further understand the effects of natural active ingredients on IL-6, we performed molecular docking at the atomic level to evaluate the binding energy and interaction pattern between these ingredients and IL-6. The 2D structure of the natural active ingredients were obtained from the PubChem Compound Database (https://pubchem.ncbi.nlm.nih.gov/). The 3D structure of IL-6 was downloaded from the Protein Data Bank (http://www.rcsb.org/). The IL-6 was imported into PyMOL software for dehydration and hydrogenation. The grid boxes were centered to encompass the structural domains of IL-6, allowing for unrestricted molecular mobility. All docking simulations were conducted using default parameters, and binding affinities were estimated based on total scores. A binding energy of < −7.00 kcal/mol was used as a criterion for screening the natural active ingredients in the analysis results [46]. The tightly connected docking results were imported into PyMOL software for visualization.

Statistical analysis

In this study, LD Score regression was performed using the LDSC software (version 1.0.1, https://github.com/bulik/ldsc). The causal estimates were presented as odds ratios (ORs) with 95% confidence intervals (95% CI). Two-sample MR analysis was conducted using R version 4.3.2 within R Studio (R Foundation for Statistical Computing, Vienna, Austria) with the TwoSampleMR package (version 0.5.7) and the MR-PRESSO package (version 1.0.0) [31, 47]. The MRlap analysis and Steiger test were performed using their respective R package [40, 41]. Colocalisation analysis was carried out using the R package coloc (version 5.2.3) with P1 = 1e-4, P2 = 1e-4, and P12 = 1eE-5 [43]. The entire molecular docking process was performed using AutoDockTools (verison 1.5.7, http://autodock.scripps.edu/) [48]. A conservative Bonferroni-corrected p-value threshold was established at 0.25/2 (0.025) for LD Score regression, at 0.05/4 (1.25e-2) for the bidirectional causal effect between asthma and AD, and at 0.05/132 (3.79e-4) for the causal effect of inflammatory proteins on asthma and AD. Additionally, a p-value < 0.05 was considered suggestive of significance [49].

Results

LD score regression analysis

We conducted LD Score regression analysis to assess the genetic correlation between asthma and AD. As presented in Table 1, the results demonstrated a statistically significant and positive genetic correlation between asthma and AD in both the discovery and replication cohorts.

Table 1.

Genetic correlations between asthma and AD

Cohort	Trait 1	Trait 2	r g	SE	p-value
Discovery	Asthma	AD	0.42	0.05	1.28e-16
Replication	Asthma	AD	0.47	0.04	1.42e-35

r_g genetic correlation estimate, SE Standard error

Causal effects of asthma on AD

In the discovery cohort, asthma (GWAS ID: J10_ASTHMA_EXMORE) was designated as the exposure, while AD (GWAS ID: GCST90244787) was identified as the outcome. Following the removal of outliers, 34 SNPs associated with asthma were selected as IVs (Table S2). The F-statistics for each IV exceeded 10, indicating the absence of weak IVs. The results derived from MR analysis revealed an estimated effect of asthma on AD (IVW: β = 0.202, OR = 1.224, 95% CI = 1.172–1.278, P = 5.89e-20) (Fig. 2; Figure S3A, B). In the replication datasets, asthma (GWAS ID: GCST90399686) and AD (GWAS ID: L12_ATOPIC) were selected as the exposure and outcome, respectively. Seventy SNPs related to asthma were included in the MR analysis, yielding a consistent result (IVW: β = 0.429, OR = 1.535, 95% CI = 1.411–1.671, P = 3.55e-23) (Fig. 2; Figure S3C, D). Estimates obtained from MR-PRESSO, MR Egger, WME, weighted model, and simple mode were concordant with these findings (Table S3). The results from the Steiger test also demonstrated that asthma is more likely to be the causal factor (Table S4). These results provide robust evidence that asthma is genetically associated with an increased risk of AD. No outliers were detected by leave-one-out analysis, suggesting that the genetic correlation between asthma and AD was not driven by a single SNP (Figure S3E, F). There was no evidence of heterogeneity in the Cochran’s Q test (Table S5). The funnel plot also visually excluded the presence of heterogeneity (Figure S3G, H). MR-Egger regression intercept analysis, MR-PRESSO global test, and Radial plot did not suggest any potential horizontal pleiotropy for the IVs (Table S6; Figure S3I, J).

Fig. 2.

Bidirectional relationship between asthma and atopic dermatitis. The IVW method was employed to evaluate the causal relationship between asthma and AD within both the discovery and replication cohorts. The findings revealed a statistically significant positive correlation between genetically predicted asthma and AD in these two cohorts. IVW: inverse variance weighed

Causal effects of AD on asthma

Reverse MR analysis was employed to assess the causal effects of AD on asthma. In the discovery cohort, 37 SNPs related to AD were designated as IVs. The IVW results revealed an estimated effect of AD on asthma (IVW: β = 0.206, OR = 1.229, 95% CI = 1.167–1.294, P = 4.83e-15) (Fig. 2; Figure S4A, B). During the replication phase, 19 SNPs were identified as IVs that were significantly associated with AD. The MR analysis corroborated the initial findings (IVW: β = 0.127, OR = 1.135, 95% CI = 1.104–1.167, P = 1.12e-18) (Fig. 2; Figure S4C, D). Consistent results were obtained using other MR methods, including MR-PRESSO, MR Egger, WME, weighted model, and simple mode (Table S3). The Steiger test results indicated that the variance explained by AD was significantly greater than that explained by asthma (Table S4). Leave-one-out analysis demonstrated no outliers that could potentially influence the genetic estimates (Figure S4E, F). No significant heterogeneity was detected using Cochran’s Q test or visualized through the funnel plot (Table S5; Figure S4G, H). The results of the MR-Egger regression intercept analysis, MR-PRESSO global test, and Radial plot provided no evidence of potential horizontal pleiotropy (Table S6; Figure S4I, J).

Bidirectional causal effects of asthma and AD evaluated by MRlap

GWAS datasets for AD (discovery cohort) and asthma (replication cohort) encompassed mixed European cohorts, including FinnGen (Table S1), thus MRlap was employed to mitigate bias in causal effect estimation due to potential sample overlap. The results are presented in Fig. 3. In the discovery cohort, the IVW-based effects of asthma on AD ranged from 0.105 for fully overlapping samples to 0.133 for non-overlapping samples, whereas the corrected effects exhibited an increase (0.116–0.157). A comparable pattern was discerned when evaluating the effects of AD on asthma, with IVW-based effects ranging between 0.409 for fully overlapping samples and 0.435 for non-overlapping samples. Following correction, the range of the estimated effect was augmented (0.432–0.474). In the replication cohort, when examining the causal effects of asthma on AD, the IVW‐based effect was 0.399 for fully overlapping samples, while the corrected results indicated an estimated effect of 0.426. For non-overlapping samples, the estimated effects were 0.459 and 0.517 for the IVW‐based effect and the corrected effects, respectively. When assessing the causal effects of AD on asthma, the IVW-based effects ranged from 0.242 for non-overlapping samples and 0.320 for fully overlapping samples, while the range of the estimated effect was elevated post-correction (0.269–0.349). Across all degrees of overlaps, the positive direction for the estimated causal effects of exposure on outcome was consistent with the two-sample MR analysis, suggesting that the bidirectional genetic relationship between asthma and AD were not affected by potentially overlapping samples.

Fig. 3.

Bidirectional causal effects of asthma and AD evaluated by MRlap. A The IVW-based observed causal effect estimates for the two traits are presented, comparing nonoverlapping samples (0%) with fully overlapping samples (100%). B The corrected causal effect estimates of the two traits are depicted, comparing nonoverlapping samples (0%) with fully overlapping samples (100%). Notably, both the p-values for the IVW-based observed causal effect and that for the corrected causal effect were less than 0.05. AD: atopic dermatitis

Colocalisation of asthma and AD

Colocalisation analysis serves as a method to evaluate whether an exposure and an outcome share the same causal SNP, particularly when SNPs exhibit robust associations with both [43]. We conducted a colocalisation analysis to further determine the probability of shared causal genetic variation underlying asthma and AD (Fig. 4A). When asthma was designated as the exposure, in both the discovery and replication cohorts, our findings indicated that two independent causal SNPs within the selected gene region (chr 6:32434347–32834347 and chr 6:32404425–32804425 for discovery and replication cohorts, respectively) were associated with both asthma and AD (PP.H3 = 1.00 and PP.H4 < 0.8). Comparable results were obtained when AD was selected as the exposure in the replication cohort (chr 5:132460977–132860977, PP.H3 + PP.H4 = 0.99 and PP.H4 = 0.30). However, when AD was designated as the exposure in the discovery cohort, a single variant within the selected gene region (chr 1:151979152–152379152) affected AD but was not associated with asthma (PP.H1 = 0.87 and PP.H3 + PP.H4 = 0.13). Nevertheless, the genetic associations between asthma and AD stem from distinct causal SNPs (PP.H4 < 0.80), reinforcing the robustness of our two-samples MR results and suggesting that they are not attributed to a locus where SNPs are in close linkage disequilibrium, thereby mitigating concerns of potential horizontal pleiotropy [50].

Fig. 4.

The results of colocalisation analysis. A Colocalisation analysis results show the probability of shared causal genetic variation between asthma and AD. B Colocalisation analysis results indicate the probability that SNPs associated with IL-6 share causal genetic variants between asthma and AD. PP.H0: no association with either exposure or outcome; PP.H1: association with exposure but not with outcome; PP.H2: association with outcome but not with exposure; PP.H3: association with both traits but driven by distinct variants; PP.H4: association with both traits and shared a causal variant. A threshold of significance for colocalisation was established at PP.H3 + PP.H4 ≥ 0.80. A PP.H4 ≥ 0.80 was considered evidence supporting colocalisation with a shared causal variant

Causal association of inflammatory proteins with asthma and AD

Both asthma and AD are often driven by chronic inflammatory mechanisms [1, 2]. Anti-inflammatory therapy utilizing corticosteroids reduces inflammatory responses and enhances symptom control [1, 51]. To investigate whether asthma and AD share common inflammatory factors, we performed a comprehensive MR analysis of 91 plasma inflammatory proteins on traits in the discovery cohort and identified several inflammatory proteins with suggestive significantly estimated effects on asthma and AD (Table S7). The results of the sensitivity analysis are presented in Table S8. Notably, IL-6 emerged as a shared inflammatory protein significantly associated with the risk of both asthma and AD. A one-unit increase in circulating IL-6 levels was causally associated with a 21.89% and 45.03% higher risk of asthma (Wald ratio: β = 0.198, OR = 1.219, 95% CI = 1.111–1.338, P = 3.05e-5) and AD (Wald ratio: β = 0.372, OR = 1.450, 95% CI = 1.328–1.584, P = 1.53e-16), respectively (Fig. 5). We conducted the Steiger test to assess the explained variance of instrumented SNPs on both exposure and outcome (Table S9), revealing that the variance explained by IL-6 was significantly greater than that explained by asthma and AD.

Fig. 5.

Causal effects of IL-6 on asthma and atopic dermatitis. The Wald ratio method was used to evaluate the causal relationship between IL-6 and two traits. The results revealed a statistically significant positive genetic association between IL-6 and both asthma and atopic dermatitis

Colocalisation analysis of IL-6 with asthma and AD

Elevated IL-6 expression has been observed in patients with asthma or AD [52, 53], and the suppression of IL-6 has been proved to alleviate symptoms of both asthma and AD [19, 54]. Subsequently, we conducted a colocalisation analysis to further determine the probability that SNPs associated with IL-6 share causal genetic variants with asthma and AD (Fig. 4B). The results indicated that two independent causal SNPs within the selected gene region (chr 1:154226970–154626970) were associated with IL-6 and asthma (PP.H3 + PP.H4 >0.80 and PP.H4 < 0.8). There was a 97.2% PP.H4 that the association between IL-6 and AD shared a causal variant within the selected gene region. These findings suggest that IL-6 may serve as an anti-inflammatory therapeutic target for patients with asthma and AD.

Natural active ingredients selection

Natural compounds have demonstrated significant efficacy in the prevention and treatment of asthma and AD [18, 54]. In our study, the TCMSP database was utilized to identify potentially effective active ingredients targeting IL-6 (TAR00351). A total of 30 natural active ingredients were retrieved and subsequently screened for pharmacokinetics and druglikeness using the SwissADME database. Ultimately, based on high gastrointestinal absorption and druglikeness without violations, 18 natural active ingredients were selected for further evaluation of their plausibility targeting IL-6 (Table S10).

Molecular docking

Molecular docking was performed to assess the binding affinity of the natural active ingredients for IL-6. AutoDockTools was employed to determine the binding sites and interactions between the natural active ingredients and IL-6, and to calculate the binding energy for each interaction. The docking results are presented in Table S11, revealing that each natural active ingredient interacts with IL-6 through visible hydrogen bonds and strong electrostatic interactions. Docking results with an energy less than − 7.00 kcal/mol indicate that the receptor and ligand can bind to each other under physiological conditions [46]. The docking energies for the interaction between IL-6 and four natural active ingredients met the scoring criterion, and the binding pocket of IL-6 was successfully occupied by these ingredients (Table 2; Fig. 6). Notably, IL-6 and atractylenolide I exhibited the lowest binding energy (−7.50 kcal/mol), indicating a highly stable binding interaction.

Table 2.

Docking results of IL-6 with four natural active ingredients

Target	Active ingredient	PubChem ID	Binding energy (kcal/mol)	Potential candidate
IL-6	Atractylenolide I	5,321,018	−7.50	Yes
	Morphine	5,288,826	−7.18	No
	Melanin	6,325,610	−7.12	No
	Sophocarpine	115,269	−7.03	Yes

Fig. 6.

Docking results of IL-6 and four natural active ingredients. A Atractylenolidei-1 has a binding energy of −7.50 kcal/mol; (B) Morphine has a binding energy of −7.18 kcal/mol; (C) Melanin has a binding energy of −7.12 kcal/mol; (D) Sophocarpine has a binding energy of −7.03 kcal/mol. The yellow dot line represents H bond

Discussion

This study comprehensively evaluated the genetic relationship between asthma and AD using a two-sample MR analysis, revealing a bidirectional genetic association between the two conditions. Subsequently, IL-6 was identified as a common inflammatory protein that has a genetic contribution to both asthma and AD, suggesting that IL-6 may serve as a therapeutic target for the management of the comorbidity of asthma and AD. Lastly, four natural active ingredients that interact with IL-6 were identified and subjected to molecular docking, showing IL-6 and atractylenolide I exhibited the lowest binding energy.

Patients with asthma exhibit a higher likelihood of experiencing a greater number of comorbidities compared to non-asthmatic individuals, including rhinosinusitis, sleep disorders, gastroesophageal reflux disease, and mental health disorders [4, 55]. Moreover, asthmatic patients with coexisting chronic conditions are more prone to asthma exacerbation, poor asthma control, and impaired quality of life [56]. The 2022 Global Initiative for Asthma (GINA) recommends the active management of comorbid conditions, particularly in cases of severe asthma [57], emphasizing that the assessment and treatment of comorbidities are imperative in the management of asthma [58]. A significant association was observed between AD and asthma compared to reference individuals (OR: 3.03, 95% CI: 2.64–3.47) [5]. Recent studies have revealed that asthma increases the incidence of AD (OR: 1.55, 95% CI: 1.40–1.73) [4], and the overall pooled prevalence of asthma among patients with AD was 25.7% (95% CI: 23.7–27.7%) [5]. Compared to asthma patients without AD, the risk of asthma exacerbation and hospitalization was significantly increased in asthmatic patients with AD [59]. A previous study revealed that the relative risks for asthma were higher in persistent AD compared to transient AD, and in severe AD compared to mild or moderate AD [6], suggesting that patients with persistent or severe AD are at a higher risk of developing asthma. These results indicate that actively targeting AD in patients with asthma leads to improvements in quality of life and asthma control. Our results demonstrate a bidirectional genetic association between asthma and AD, which is supported by sensitivity analyses and the MR-lap test, thus offering novel insights into the co-management of asthma and AD.

Despite the observation of a strong genetic correlation between asthma and AD through LD Score regression and bidirectional MR analysis, the specific genetic variants driving these associations may differ (PP.H3 + PP.H4 >0.8 and PP.H4 < 0.8). This suggests that the underlying genetic mechanisms may not be entirely overlapping. Genetic association reflects the degree of shared genetic influences between two traits at the overall genomic level [20, 34]. Colocalization analysis determines whether the causal variants driving the association are located within a specific genomic region [41]. Asthma and AD can exhibit a high correlation at the genomic level but be driven by distinct variants. These variants may be associated with shared functions or biological pathways, involving common cell types (such as epithelial cells) [60], inflammatory proteins (such as IL-4 and IL-6) [61], or immunologic patterns (such as type 2 immunity) [62]. Additionally, it is a relatively common situation for a single variant to influence the expression of different genes within a chromosomal region [41], reflecting tissue-specificity (e.g., lung vs. skin), disease onset (younger vs. elderly), and triggers (drug vs. non-drug). Future research, including cell-based fine-mapping, functional genomics validation, and cross-disease expression quantitative trait locus studies, will help elucidate how these distinct causal variants converge onto shared or interacting biological pathways, thereby providing a deeper understanding of the molecular basis underlying the comorbidity of asthma and AD.

Although the common underlying mechanisms have not been fully elucidated, it is well-established that inflammation plays a pivotal role in the pathogenesis and development of both asthma and AD [2, 63]. This is also supported by the fact that corticosteroid therapy is the most common and effective treatment for both asthma and AD [14, 51]. Various mechanisms for asthma drive eosinophilic and non-eosinophilic inflammation, leading to mixed granulocytic inflammation [1], which is associated with asthma exacerbation and airway remodeling [64]. Similarly, upon exposure to pro-inflammatory cytokines, keratinocytes can secrete various types of chemokines and cytokines, potentially initiating the inflammatory condition in AD [65]. Cytokine imbalance can alter immune responses and promote IgE-mediated hypersensitivity [66]. In the present study, through MR analysis, we identified several estimated causal effects of inflammatory proteins on asthma and AD. In particular, IL-6 was identified as a shared inflammatory protein significantly associated with an increased risk of both asthma and AD.

IL-6 stands as a prominent etiological factor in the pathogenesis of inflammatory diseases [67]. IL-6 has the capacity to amplify Th2-mediated tissue inflammation and cytokine responses [68]. Th2 cells-dominated inflammation plays an important role in the pathogenesis of allergic diseases, such as asthma and AD [62, 69]. Inhibition of the recruitment and differentiation of Th2 cells has been demonstrated to alleviate airway inflammation and AD-like skin lesions [70, 71]. Furthermore, IL-6 enhances the proliferation, maturation, and reactivity of mast cells (MCs) following FcεRI aggregation [72]. MCs are involved in inflammation and adaptive immunity. FcεRI-activated MCs release various cytokines and lipid mediators, which contribute to immune cell accumulation and tissue remodeling in asthma [73]. The activation of MCs and the subsequent release of inflammatory cytokines have been recognized as crucial mechanisms in the pathogenesis of AD [74, 75]. Reduction in MC accumulation and the release of pro-inflammatory mediators have been reported to improve symptoms in both asthma and AD [74, 76]. In addition, the activation of Th2 cells or MCs also results in the production of IL-6, which in turn mediates immune responses, inflammation, and acute-phase responses [77]. Patients exhibiting elevated IL-6 levels tend to present with more severe asthma and asthma exacerbation [53].Administration of an IL-6 neutralizing antibody effectively abrogated airway inflammation and remodeling in an asthma model induced by toluene diisocyanate [19]. In a murine model of AD-like dermal disorder, allergic inflammation of the skin was alleviated by treatment with Centella asiatica ethanol extract, which reduced the expression of pathogenic cytokines in mast cells, including IL-6 [54]. Collectively, these findings suggest that IL-6 represents a potential therapeutic target for patients with asthma and AD.

Colocalisation analysis revealed that a shared causal variant within the selected gene region was associated with IL-6 and AD, but the shared variant was not supported for IL-6 and asthma (PP.H3 = 0.73; PP.H4 = 0.26). Two distinct variants exist within the genetic region (chr 1:154226970–154626970). One influences circulating IL-6 levels. It remains to be established whether the potential second variant affects asthma susceptibility through an IL-6-related mechanism. Thus, within the selected gene region closely associated with IL-6, the genetic architecture underlying asthma risk may be more complex than a simple shared causal mechanism. Despite this genetic confounding factor, our study has identified chr 1:154226970–154,626,970 as a convergent region implicated in both IL-6 regulation and asthma pathogenesis. Future work utilizing expanded cohorts and high-resolution genomic tools will be required to precisely pinpoint the true causal variants within this region and experimentally clarify whether they influence asthma by regulating IL-6.

Additionally, natural drugs have demonstrated efficacy in the management of asthma and AD. Baiheqingjin decoction, a traditional Chinese herbal formula comprising seven herbs, is commonly employed to treat asthma by attenuating the release of inflammatory cytokines, including IL-6 [78]. Pi-pa-run-fei-tang exhibited lung-protective effects in an ovalbumin-induced asthma model by decreasing IL-6 levels and modulating the pathway related to IL-6 [18]. Topical application of quercetin proved beneficial in controlling AD symptoms in mice with AD-like skin lesions by suppressing IL-6 expression [17]. Illicium verum has been identified as a promising potential therapeutic candidate for inflammatory skin diseases, owing to its ability to inhibit IL-6 expression [65]. In the present study, four potential candidates targeting IL-6 were identified: atractylenolide I, morphine, melanin, and sophocarpine. These compounds are isolated from Atractylodes macrocephala, opium, Nigella sativa, and Sophora alopecuroides, respectively [79–82]. Atractylenolide I demonstrated a dose-dependently inhibitory effect on angiogenesis by downregulating IL-6 activity in chronic inflammation [83]. Additionally, atractylenolide I ameliorated cancer cachexia by decreasing IL-6 production [79]. Sophocarpine exerted anti-cachectic effects through the inhibition of IL-6 [81], and alleviated LPS-induced lung injury by suppressing IL-6 [84]. However, previous studies have observed that morphine and melanin induce an enhanced production and secretion of IL‑6 [82, 85–87], suggesting that they should not be considered as candidate drugs for the treatment of asthma and AD. Consequently, atractylenolide I and sophocarpine emerge as potential active ingredients for the co-treatment of asthma and AD (Table 2). Notably, IL-6 and atractylenolide I exhibited the lowest binding energy (−7.50 kcal/mol), indicating an extremely stable binding interaction.

Limitations

This study is subject to several limitations that warrant consideration. First, although the estimated effect sizes of exposure on outcome determined by MR analysis were statistically significant, their magnitudes were relatively moderate. For instance, the OR of asthma on AD in discovery cohort was 1.22. MR analysis may not fully recapitulate the conditions of real-world clinical trials. As a result, the effect sizes observed in MR studies may not precisely correspond to those seen in practical clinical settings [20]. Second, the reliance on 91 plasma circulating inflammatory proteins for MR testing presents challenges in pinpointing the most effective tissue for treatment. Given that different tissues may exhibit distinct genetic regulatory mechanisms, focusing solely on blood biomarkers may not provide a comprehensive understanding of the disease and their potential treatment. Third, the accuracy of molecular docking analysis is highly dependent on the quality of the protein structures and ligands. Therefore, further experimental validation and rigorous clinical trials are necessary to confirm the safety and therapeutic potential of these natural active ingredients. Finally, the generalizability of the study is limited by its predominant focus on individuals of European descent. To extrapolate these findings to individuals of other ethnicities, additional research and validation are necessary.

Conclusion

In conclusion, our study demonstrated a bidirectional genetic relationship between asthma and AD through MR analysis. IL-6 may serve as an anti-inflammatory therapeutic target for individuals suffering from both asthma and AD. The natural active ingredient atractylenolide I, which targets IL-6, has exhibited potential value for the concurrent management of asthma and AD. This study provides novel insights into the comorbidity of asthma and AD, as well as the possibility for co-treatment by targeting the shared inflammatory cytokine IL-6.

Authors’ contributions

Jirui Bi, Jihong Bi, and Xing Ni conceived and designed the study. Jirui Bi, Jihong Bi, Xing Ni, Lin Lu, Tingting Yin, Shaohua Mao, Run Zhang, and Juan He collected and analyzed the data. Jirui Bi and Jihong Bi drafted the manuscript. Jirui Bi and Xing Ni edited the manuscript. All authors reviewed and approved the final manuscript.

Funding

This work was supported by the Special Project for the Translation of Clinical Medical Research of Anhui Provincial Key Research and Development Project (No. 202304295107020097, Xing Ni).

Data availability

The GWAS datasets of asthma for discovery cohort and replication cohort generated and/or analysed during the current study are available in the repositories of FinnGen R10 (J10_ASTHMA_EXMORE) and GWAS Catalog (GCST90399686), respectively, (J10_ASTHMA_EXMORE: https://www.finngen.fi/en/access_results; GCST90399686: https://www.ebi.ac.uk/gwas/). The GWAS datasets of atopic dermatitis for discovery cohort and replication cohort generated and/or analysed during the current study are available in the repositories of GWAS Catalog (GCST90244787) and FinnGen R10 (L12_ATOPIC), respectively, (GCST90244787: https://www.ebi.ac.uk/gwas/; L12_ATOPIC: https://www.finngen.fi/en/access_results). The GWAS datasets of 91 circulating inflammatory proteins generated and/or analysed during the current study are available in the GWAS Catalog (GCST90274758-GCST90274848) repository, (https://www.ebi.ac.uk/gwas/). The 3D structure of IL-6 generated and/or analysed during the current study is available in the UniProt (P05231) repository, (https://www.uniprot.org/), which was downloaded from the Protein Data Bank (1ALU) repository, (https://www.rcsb.org/).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.