Exploring the causal relationship between inflammatory cytokines and C-reactive protein levels and atopic dermatitis: a two-way two-sample Mendelian randomization study

doi:10.21203/rs.3.rs-4674994/v1

Download PDF

Article

Exploring the causal relationship between inflammatory cytokines and C-reactive protein levels and atopic dermatitis: a two-way two-sample Mendelian randomization study

https://doi.org/10.21203/rs.3.rs-4674994/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Objective: Atopic dermatitis (AD) is an allergic inflammatory skin disease. Changes in circulating inflammatory proteins are reflected in the entire process of AD progression, and its pathophysiology is still unclear. This Mendelian randomization study was conducted to further evaluate the role of circulating inflammatory proteins in AD.

Methods: This study investigated the potential causal relationship between circulating inflammatory proteins and AD. We used a two-sample Mendelian randomization (MR) method to analyze data from a large-scale genome-wide association study to explore the relationship between 91 circulating inflammatory proteins, 41 inflammatory factors, and CRP and AD. The inverse variance weighted method was mainly used to evaluate the causal relationship between exposure and outcome based on the effect indicator odds ratio (OR) and 95% confidence interval (CI). In addition, MR-Egger, weighted median, simple model, weighted model and MR-PRESSO multiple sensitivity analyses were applied to strengthen the final results. The leave-one-out method, heterogeneity test, and horizontal gene pleiotropy test were used to verify the stability and reliability of the results.

Results: Forward MR analysis showed that there was a significant correlation between AD risk and changes in the levels of multiple inflammatory proteins at different p-value thresholds, among which increased levels of interleukin-18 receptor 1 were found to increase the risk of AD, which was significant in all three groups of analysis (P IVW<0.05); increased levels of C-X-C motif chemokine 9 and Fms-related tyrosine kinase 3 ligand were found to reduce AD risk at P<5×10^-8 and p<5×10^-7 thresholds; increased levels of C-X-C motif chemokine 11 were found to be associated with a reduced risk of AD at P<5×10^-8and P<5×10^-6 thresholds (P IVW<0.05). Reverse MR analysis showed that increased AD risk was associated with decreased levels of AXIN-1, natural killer cell receptor 2B4, interleukin-1 receptor subunit α, and interleukin-33 (P IVW<0.05). In addition, increased AD risk was associated with increased Cystatin D levels (P IVW<0.05). In the 41 inflammatory factor data sets, increased AD risk may be associated with increased IL18 levels (P IVW=0.036) and MIG levels (P IVW=0.046). No significant heterogeneity and horizontal pleiotropy were observed in the analysis. After verification MR analysis, it was found that there was a significant association between the levels of inflammatory proteins such as Fms-related tyrosine kinase 3 ligand, interleukin 18 receptor 1, C-X-C motif chemokine 9, and tumor necrosis factor ligand superfamily member 14 and AD risk, and there was consistency between different P value thresholds. Bidirectional MR showed that there was a complex bidirectional causal relationship between interleukin 18 receptor 1 levels and AD. The leave-one-out analysis showed that the results were stable, there were no instrumental variables that had a strong impact on the results, and the leave-one-out method verified the robustness of the results. There was heterogeneity test and horizontal pleiotropy in the reverse causal relationship between the level of tumor necrosis factor ligand superfamily member 14 and the AD validation set.

Conclusion: The results of MR analysis indicate a potential causal relationship between circulating inflammatory proteins and AD. This study provides a new approach for exploring the biological mechanisms of AD in the future and proposes possible therapeutic targets. Further research is needed to confirm these results and understand the specific role of these proteins in AD, and to provide reference value for future studies on the relationship between circulating inflammatory proteins and AD.

Health sciences/Medical research

Health sciences/Molecular medicine

Health sciences/Risk factors

Mendelian randomization

inflammatory factors

atopic dermatitis

causal relationship

C-reactive protein level

Atopic dermatitis (AD) is a widespread chronic skin disease associated with allergic inflammation. AD is not just a pediatric disease, but also exists for a long time in adults. The most common criterion is a history of systemic dry skin. The prevalence of comorbidities is very high, which is a long-standing clinical problem that is difficult to solve and poses a serious threat to the life and health of patients ^[1]. However, the cellular and molecular mechanisms of AD are complex, and little is known about its damage. Due to its high recurrence and chronicity, it has brought further obstacles in determining therapeutic targets and clinical trial design ^{[2, 3]}. Therefore, how to effectively promote the healing of AD is the current research direction and focus of clinical work. Studies have shown that more and more circulating inflammatory proteins, such as IL1-β, IL12, IL13, IL18, etc., are related to the occurrence of AD ^[4–8], which indicates that circulating inflammatory proteins may play a key role in AD, but the causal relationship between different inflammatory proteins and AD needs to be further explored ^[9].

Mendelian randomization (MR) was proposed by Katan and is a genetic variable analysis that follows Mendel's law of inheritance ^[10]. MR introduces genetic variation and uses single-nucleotide polymorphisms (SNPs) as instrumental variables for MR analysis to assess the causal relationship between exposure and outcome ^[11]. Based on Mendel's second law of inheritance, that is, alleles are randomly assigned and fixed at conception, the process is similar to traditional RCT trials, where patients are randomly assigned to treatment and control groups. MR studies use the random distribution of genetic variation to eliminate confounding factors, reverse causality, and simulate the process of randomized controlled trials, which is not easily affected by reverse causality and confounding factors ^{[12, 13]}. Therefore, this study used a large-scale genome-wide association study (GWAS) dataset to analyze the causal association between circulating inflammatory proteins and AD through a two-sample bidirectional MR study.

Our study adopted a two-sample MR method and used multiple p-value thresholds to improve the accuracy of the research results while acknowledging the trade-offs. Lower p-values (e.g., < 5.0E − 08) are usually used to ensure the robustness of the association, reduce heterogeneity, and improve research precision. However, such strict criteria may also exclude potentially meaningful associations ^[14–16]. Therefore, by adopting different p-value thresholds, our analysis aims to strike a balance between minimizing false positives and not neglecting important associations that are crucial to understanding the pathophysiology of AD.

2.1 Research design

We used two-sample MR (TSMR) analysis to explore the causal relationship between circulating inflammatory proteins and AD. In our MR analysis, circulating inflammatory proteins were used as exposure factors, significantly associated SNPs were selected as IVs, and AD was used as the outcome variable. The two-sample MR method was used for causal analysis, and the Q test was used to test the heterogeneity of the results, and sensitivity analysis was performed to verify the reliability. In order to select appropriate IVs in the two-sample MR study, this study formulated three key assumptions: ① There is a significant association between IVs and circulating inflammatory proteins; ② IVs are independent of confounding factors and are not related to all confounding factors of circulating inflammatory proteins-AD; ③ Instrumental variables cannot directly affect the results, but can only indirectly affect the results through association with circulating inflammatory proteins. The research process is shown in the figure below (Fig. 1).

2.2 Data Sources

Supplementary Table 1 introduces the information of all datasets included in the studies. The datasets related to inflammatory cytokines are from 3 studies, and the data on 91 circulating inflammatory proteins are from GWAS data of 11 cohorts measured using the Olink Target Inflammation Panel, involving a total of 14,824 participants of European ancestry^[17]. Another inflammatory factor dataset comes from a comprehensive study that provides genetic variation information on 41 different inflammatory cytokines in 8293 Finnish individuals ^[18]. This survey coordinated data from the Young Finns Study (YFS) and the FINRISK survey, showing that the average age of YFS study participants was 37 years old and the average age of FINRISK survey participants was 60 years old. In addition, a dataset on C-reactive protein was included, which was derived from the ieu-b-35 dataset of the GWAS database, involving a total of 204,402 participants of European ancestry^[19].

The AD dataset comes from the ieu-a-996 dataset released by the EAGLE consortium, which includes 10,788 AD cases and 30,047 controls of European descent ^[20]. The AD data used as the validation set comes from the ukb-b-20141 dataset of the UKBB database, which involves a total of 462,933 participants of European descent, including 11,819 AD patients and 451,114 controls, including both males and females.

2.3 Instrumental variable selection

We selected SNPs (P < 5×10^− 6), (P < 5×10 − ⁷), and (P < 5×10^− 8) associated with the three inflammatory protein datasets in the whole genome for forward TSMR analysis. For reverse TSMR analysis, we used SNP (P < 5×10^− 6). In addition, PLINK clustering was used to calculate the linkage disequilibrium between SNPs for each exposure, and the parameter r² < 0.001 (cluster distance = 10000 kb) was set as the threshold for SNP linkage equilibrium to exclude the interference of linkage disequilibrium (LD). The F statistic was calculated using F = beta²/se², indicating the robustness of the instrumental variable, and the F values corresponding to single SNPs were all > 10, indicating that the results did not have weak IVs bias. This study used a variety of MR methods including IVW, weighted median, MR Egger, simple model and weighted model for analysis. IVW was selected as the main analysis method. Since IVW method has higher test efficiency than the other four MR methods, this study used IVW method to test causal effects. When P < 0.05, circulating inflammatory proteins and AD were considered to have a suggestive association ^[21]. Reverse MR analysis used a similar approach to evaluate the reverse effects of AD on inflammatory proteins. After screening positive results,We further only included inflammatory protein trait trait with at least three SNPs as IVs in downstream analyses.

Cochran's Q statistic was used to assess the heterogeneity between individual SNPs. If no significant heterogeneity was observed (p < 0.05), a fixed effect model was used ^[22]; otherwise, the causal relationship needs to be interpreted with caution. Sensitivity analysis was also performed to verify the robustness of our results. In addition, MR-Egger and MR-PRESSO methods were used to determine the presence of pleiotropy. The intercept obtained from MR-Egger regression was used to measure directional pleiotropy ^[23], and MR-PRESSO was used to enhance pleiotropy detection ^[24]. Leave-one-out sensitivity analysis was performed to determine whether a single SNP had a significant effect on the MR results.

2.4 Statistical analysis

All statistical analyses were performed using R software version 4.2.3. MR analysis was performed using the “TwoSample MR” R package, and MR-PRESSO was performed using the “MRPRESSO” R package.

2.5 External dataset validation

In the initial exploratory phase of the study, we did not use p-value correction to capture more potential associations. To ensure the reliability of the preliminary results, we chose to validate them using an external AD dataset. The ukb-b-20141 dataset from the MRC-IEU Consortium allows us to fully assess the generalizability of our findings. During the initial screening process, we included previously discovered potential key proteins, 17 inflammation-related proteins including CRP. By utilizing these datasets, we were able to validate the associations observed in our study.

3.1. Forward Mendelian randomization results

Using the SNP threshold (P<5×10^-8), instrumental variables of a total of 41 inflammatory proteins among 91 inflammatory proteins were extracted for TSMR analysis. The results showed that elevated levels of T cell surface glycoprotein CD5 were associated with reduced risk of AD (OR=0.766, PIVW=0.003), and the analysis showed no significant heterogeneity (MR Egger Q=2, Q p-value=0.651) , and there is no horizontal pleiotropy (P Egger Intercept = 0.984, P MR Presso=0.864); the study also found that the level of C-X-C motif chemokine 11 (OR=0.853, PIVW=0.043), C-X-C motif chemokine 11 Increased levels (OR=0.85, PIVW=0.043), increased levels of C-X-C motif chemokine 9 (OR=0.727, PIVW=0.001), and increased levels of Fms-related tyrosine kinase 3 ligand (OR=0.873, PIVW= 0.047) will also reduce the risk of AD, and no significant heterogeneity and horizontal pleiotropy were found (as shown in Table 1). Higher interleukin 18 receptor 1 levels were associated with increased risk of AD (OR = 1.164, PIVW = 0.001), and the analysis showed no significant heterogeneity (MR Egger Q = 4, Q p-value = 0.424) and no Horizontal pleiotropy (P Egger intercept=0.579, P MRPresso=0.626), in addition, increased cytokine levels induced by TNF-related activation were associated with increased risk of AD (OR=1.119, PIVW=0.045), with no significant difference Qualitative (MR Egger Q=5, Qp-value=0.795) or horizontal pleiotropy (P Egger Intercept)=0.47, P MR Presso=0.782). In the 41 inflammatory factor data set, under the SNP threshold (P<5×10-8), there are 12 inflammatory factors with no less than 3 SNPs. The instrumental variables of 12 inflammatory proteins were extracted for TSMR analysis. Positive results were found to be significantly correlated with the AD data set, and CRP was not found to be significantly correlated with the AD data set at the SNP threshold (P<5×10^-8). All positive results and instrumental variable specific details at the P<5×10^-8 threshold are listed in Supplementary Tables 2-3. The forest plot of all positive results is shown in Figure 2.

Some proteins did not yield SNPs due to the original selection criteria, leading us to relax the conditions. We selected SNPs with a threshold P<5×10^-7 as instrumental variables and performed TSMR analysis on 71 inflammatory proteins. The results were as follows: Elevated levels of C-X-C motif chemokine 9 were associated with reduced AD risk (OR=0.727, PIVW=0.001), and the analysis showed no significant heterogeneity (MR Egger Q=2, Q p-value=0.652), and there is no horizontal pleiotropy (P Egger Intercept=0.66, P MR Presso=0.813); the study also found that increased levels of Fms-related tyrosine kinase 3 ligand (OR=0.835, PIVW=0.004) will also In reducing the risk of AD, no significant heterogeneity (MR Egger Q=11, Q p-value=0.566) and horizontal pleiotropy (P Egger Intercept=0.296, P MR Presso=0.813) were found. It was also found that higher interleukin 13 levels were associated with increased risk of AD (OR=1.353, PIVW=0.034), and the analysis showed no significant heterogeneity (MR Egger Q=1, Q p value=0.288), involving With less than three SNPs, horizontal pleiotropy testing cannot be performed. In addition, higher interleukin 18 receptor 1 levels are associated with increased risk of AD (OR=1.164, PIVW=0.001), and the analysis showed no significant heterogeneity. (MR Egger Q=4, Q p-value=0.424) and no horizontal pleiotropy (P Egger intercept=0.579, P MRPresso=0.628). In the 41 inflammatory factor data set, under the SNP threshold (P<5×10-7), there are 17 inflammatory factors with no less than 3 SNPs. The instrumental variables of 17 inflammatory proteins were extracted for TSMR analysis. Positive results were found to be significantly associated with the AD dataset, CRP was not found to be significantly associated with the AD dataset at the SNP threshold (P<5×10^-7), All positive results with a threshold of P<5×10^-7 are listed in Supplementary Table 2, and the instrumental variable details are listed in Supplementary Table 4. A forest plot of all positive results is shown in Figure 3.

We selected SNPs with a threshold P<5×10^-6 as instrumental variables to enable TSMR analysis of all 91 inflammatory proteins. The results are as follows: increased levels of T cell surface glycoprotein CD5 are associated with reduced risk of AD (OR=0.832, P IVW=0.047), but the heterogeneity is obvious (MR Egger Q=17, Q p value=0.002), with Significant horizontal pleiotropy (P Egger Intercept=0.6, PMR Presso=0.003); the study also found that increased levels of C-X-C motif chemokine 11 (OR=0.878, P IVW=0.025) also reduced the risk of AD, There was no significant heterogeneity (MR Egger Q=20, Q p-value=0.473) or horizontal pleiotropy (P Egger Intercept=0.06, P MRPresso=0.337). At the same time, higher interleukin 18 receptor 1 levels were associated with increased risk of AD (OR = 1.158, PIVW = 0.001), and the analysis showed no significant heterogeneity (MR Egger Q = 20, Q p value = 0.555) There was no horizontal pleiotropy (P Egger intercept=0.379, P MRPresso=0.656). In addition, higher interleukin-6 levels were associated with an increased risk of AD (OR=1.282, PIVW=0.037), with no significant difference. Qualitative (MR Egger Q = 6, Q Pvalue = 0.09) or horizontal pleiotropy (P Egger Intercept = 0.194, P MRPresso=0.168); higher tumor necrosis factor ligand superfamily member 14 levels are associated with AD risk Elevated association (OR=1.154, PIVW=0.022), no significant heterogeneity (MR Egger Q=22, Q p-value=0.094) or level pleiotropy (P Egger Intercept=0.077, P MRPresso=0.052) ; In addition, higher TNF-a levels were also associated with increased risk of AD (OR=1.282, PIVW=0.037) without significant heterogeneity (MR Egger Q=6, Q p-value=0.09) or level pleiotropic effects Sex (P Egger Intercept=0.194, P MR Presso=0.168).

Under the SNP threshold (P<5×10^-6), TMSR analysis of all 41 inflammatory factors can be performed on the 41 inflammatory factor data set, and it was found that increased MIG levels will also reduce the risk of AD (OR=0.89, PIVW = 0.005), no significant heterogeneity (MR Egger Q=11, Q p-value = 0.35), horizontal pleiotropy (P Egger Intercept=0.295, P MRPresso=0.374)

In the case of SNP threshold (P<5×10^-6), no positive results were found that CRP was significantly related to the AD data set. When the parameters of SNP linkage balance were adjusted to r²<0.01 (clustering distance = 500kb), Positive results appeared, suggesting that crp levels may be related to AD (OR=1.13, PIVW=0.045), but the results were highly heterogeneous (MR Egger Q=169, Q Pvalue=<0.000) and significant The horizontal pleiotropy (P Egger Intercept) = 0.119, P MRPresso <0.002). The evidence is insufficient to confirm the correlation between crp and AD, and the results need to be further verified. All positive results at the P<5×10^-6 threshold are listed in Supplementary Table 2, and instrumental variable details are listed in Supplementary Table 5. The forest plot of all positive results is shown in Figure 4.

From the three sets of analyses, we can draw several conclusions:

Interleukin 18 receptor 1 levels: Elevated levels of interleukin 18 receptor 1 have been found to increase the risk of AD. This result was significant in all three sets of analyzes (P IVW<0.05). After analyzing the three different thresholds, the results showed no heterogeneity and were relatively stable.

Elevated levels of C-X-C motif chemokine 9 are associated with reduced risk of AD. This analysis showed no significant heterogeneity and horizontal pleiotropy; increased levels of Fms-related tyrosine kinase 3 ligand also reduced the risk of AD. No significant heterogeneity and horizontal pleiotropy were found. These two inflammatory factors were significantly associated with AD risk reduction in the first two sets of analyzes (P IVW<0.05). Elevated levels of C-X-C motif chemokine 11 were significantly associated with reduced risk of AD in both the first and third sets of analyses (P IVW<0.05).

Next, we performed a leave-one-out analysis on the above four key results (see Figure 5). This involves excluding each SNP in turn and estimating the effect size of the remaining SNPs. The analysis showed no significant difference in effect size before and after exclusion, indicating that no single SNP had a significant impact on the MR estimate.

Table 1 Heterogeneity and horizontal pleiotropy test of positive results of forward analysis under different thresholds

MR	Inflammatory factors	Heterogeneity test				Genetic pleiotropy
		IVW		MR-Egger		MR-Egger intercpet		MR-PRESSO
		IVW		MR-Egger		MR-Egger intercpet		global test
		Q	P	Q	P	Intercept	P	RSSobs	P
5.00E+08	T-cell surface glycoprotein CD5 levels	3	0.836	2	0.651	-0.002	0.984	1.417	0.864
	C-X-C motif chemokine 11 levels	3	0.364	2	0.237	0.018	0.693	5.824	0.483
	C-X-C motif chemokine 9 levels	3	0.773	2	0.652	-0.025	0.659	1.809	0.820
	Fms-related tyrosine kinase 3 ligand levels	8	0.718	7	0.637	-0.006	0.690	6.892	0.723
	interleukin-18 receptor 1 levels	5	0.516	4	0.424	0.049	0.579	5.181	0.626
	TNF-related activation-induced cytokine levels	6	0.811	5	0.795	0.017	0.470	4.440	0.782
5.00E+07	C-X-C motif chemokine 9 levels	3	0.773	2	0.652	-0.025	0.659	1.809	0.813
	Fms-related tyrosine kinase 3 ligand levels	12	0.545	11	0.566	-0.015	0.296	12.587	0.568
	Interleukin-13 levels	2	0.396	1	0.288	0.078	0.570
	interleukin-18 receptor 1 levels	5	0.516	4	0.424	0.049	0.579	5.181	0.628
5.00E+06	T-cell surface glycoprotein CD5 levels	18	0.002	17	0.002	0.013	0.600	43.979	0.003
	C-X-C motif chemokine 11 levels	21	0.307	20	0.473	0.028	0.060	25.847	0.337
	interleukin-18 receptor 1 levels	21	0.566	20	0.555	-0.010	0.379	19.971	0.659
	Interleukin-6 levels	7	0.038	6	0.090	-0.039	0.194	44.367	0.168
	Tumor necrosis factor ligand superfamily member 14 levels	23	0.041	22	0.094	0.021	0.077	40.589	0.052
	TNF-related activation-induced cytokine levels	27	0.690	28	0.734	0.003	0.771	24.460	0.757
	MIG	13.523	0.332	12.183	0.350	-0.023	0.295	15.532	0.374

3.2. Reverse Mendelian randomization results

In the reverse Mendelian randomization analysis involving AD and 91 inflammatory proteins, the following positive results were obtained: AXIN-1 level: Increased risk of AD was associated with decreased AXIN-1 level (OR=0.950, PIVW=0.011). The analysis showed no significant heterogeneity (MR Egger Q=37, Q p-value=0.783) and no evidence of horizontal pleiotropy (P Egger Intercept=0.419, P MR-Presso=0.754). Natural killer cell receptor 2B4 level: Increased risk of AD was associated with decreased natural killer cell receptor 2B4 level (OR=0.962, PIVW=0.032). No significant heterogeneity was observed in this analysis (MR Egger Q=37, Q P value=0.386), and there was no evidence of horizontal pleiotropy (P Egger Intercept= 0.851, P MR-Presso=0.452). Interleukin 10 receptor subunit alpha levels: Increased AD risk was associated with decreased levels of natural killer cell receptor 2B4 (OR=0.957, P IVW=0.030). No significant heterogeneity was observed in this analysis (MR Egger Q=37, Q p-value=0.452), and there was no evidence of horizontal pleiotropy (P Egger Intercept= 0.523, P MR Presso=0.460). Interleukin 33 levels: Increased AD risk was associated with decreased interleukin 33 levels (OR=0.960, PIVW=0.041). No significant heterogeneity was observed in this analysis (as shown in Table 2),and there was no evidence of horizontal pleiotropy (P Egger Intercept=0.141, P MRPresso=0.528). Cystatin D levels: In contrast, increased AD risk was associated with elevated cystatin D levels (OR=1.040, PIVW=0.029). The analysis also showed no significant heterogeneity (MR Egger Q =37, QP value=0.649) and no horizontal pleiotropy (P Egger Intercept=0.209, P MRPresso=0.642).

In the case of the SNP threshold (P<5×10^-6) of the 41 inflammatory factor datasets, 2 inflammatory proteins were significantly associated with the AD dataset, and an increased risk of AD may lead to increased IL18 levels (OR=1.078, PIVW=0.036). After analysis, it was found that the results had no significant heterogeneity (MR Egger Q=37, Q p value=0.379) and horizontal pleiotropy (P Egger Intercept)=0.465, P MRPresso=0.412). Increased AD risk was associated with increased MIG levels (OR=1.079, PIVW=0.046), with no significant heterogeneity (MR Egger Q=37, Q p value=0.166) and horizontal pleiotropy (P Egger Intercept=0.269, P MRPresso=0.175).

In the reverse TSMR analysis of the CRP dataset and the GWAS instrumental variables of AD, no positive results significantly associated with the AD dataset were found. All positive results of the reverse MR analysis under the threshold of P<5×10^-6 are listed in Supplementary Table 2, and the specific details of the instrumental variables are listed in Supplementary Table 6. The forest plot of all positive results is shown in Figure 6, and the sensitivity analysis performed by the leave-one-out method demonstrated robust results (see Figure 7).

3.3. Validation group results

We selected three SNP thresholds (P<5×10⁻⁸, P<5×10⁻⁷, P<5×10⁻⁶), took 17 inflammatory proteins as exposure objects, and AD as the outcome, and performed a two-sample Mendelian randomization analysis. The forest plot of the research results is shown in Figure 8. The clustering conditions were the same as the initial test set. The specific results are shown in Supplementary Tables 7-9. After analysis, it was found that interleukin 18 receptor 1 levels and Fms-related tyrosine kinase 3 were correlated with the AD validation set in the three different SNP threshold groups. This suggests that increased interleukin 18 receptor 1 levels may be a risk factor for AD, and increased Fms-related tyrosine kinase3 levels may be a protective factor for AD. Among them, interleukin 18 receptor 1 levels were significant for AD both in the initial exploration stage and in the validation set. When we performed reverse validation with AD as the exposure factor, we observed an opposite causal relationship between interleukin 18 receptor 1 levels and AD, and there was no obvious heterogeneity and horizontal pleiotropy. The specific results of the reverse validation analysis are shown in Supplementary Table 10. These results indicate that there is a complex interaction between interleukin 18 receptor 1 levels and AD, which may involve a bidirectional causal relationship. In the reverse validation, it was also found that higher levels of tumor necrosis factor ligand superfamily member 14 were significantly correlated with the AD validation set, but the results were subject to heterogeneity tests and horizontal pleiotropy tests. We also observed an opposite causal relationship in the initial GWAS set. These results indicate that there may be a bidirectional causal relationship between AD and tumor necrosis factor ligand superfamily member 14, but it still needs to be verified. In addition, the remaining positive results were not found to be significantly correlated with the AD validation data set. After validation analysis, CRP was also not found to be significantly associated with the AD validation set. All positive results of forward and reverse validation are shown in Supplementary Table 11.

This study used the TSMR approach to explore the potential causal relationship between circulating inflammatory proteins and AD. Our analysis revealed several important associations between inflammatory proteins and AD risk, providing new insights into the pathophysiology of AD and potentially revealing new therapeutic targets. This approach allows for a more comprehensive exploration of the potential causal relationship between inflammatory proteins and AD. Forward MR analysis revealed an association between elevated levels of specific inflammatory proteins and increased risk of AD, while reverse MR analysis provided evidence that increased risk of AD may lead to changes in levels of certain inflammatory proteins. These findings suggest that there may be a bidirectional causal relationship between inflammatory proteins and AD, further complicating their interactions.

In addition, forward MR analysis showed that at different p-value thresholds, there was a significant correlation between AD risk and changes in levels of multiple inflammatory proteins, such as IL18 receptor 1, C-X-C motif chemokine 9 levels, Fms-related tyrosine kinase 3 ligand levels, and C-X-C motif chemokine 11 levels. Reverse MR analysis showed that increased AD risk may lead to changes in the levels of certain inflammatory proteins, such as AXIN-1 levels, natural killer cell receptor 2B4 levels, interleukin 10 receptor subunit α levels, interleukin 33 levels, Cystatin D levels, IL18 levels, and MIG levels. These findings provide a new perspective for studying the role of these proteins in inflammation and may help understand and treat AD and related inflammatory skin diseases. In our study, multiple inflammatory proteins identified at different p-value thresholds showed significant positive results associated with AD risk. For example, increased levels of interleukin 18 receptor 1 were associated with increased AD risk, and Hu et al. found that plasma free IL-18 and free IL-18BP levels were higher in AD patients than in healthy controls. IL-18 can increase skin mast cells by up to 3.1 times ^[25]. Previous studies have shown that keratinocytes have a functional response to IL-18, with upregulation of MHC II and production of chemokines CXCL10/IP-10. These findings further support the important role of IL-18 in inflammatory skin diseases in the epidermis ^[26]. Natalija Novak et al. found that SNPs in the IL18 gene may be involved in the development of AE by causing dysfunctional IL-18 production in vivo. At the functional level, the amount of IL-18 in the supernatant of PBMCs of AE patients stimulated with Staphylococcus aureus enterotoxin B was significantly higher than that of healthy controls. At the same time, the amount of active IL-18 in the serum of AE patients increased when the disease worsened^{[8, 27, 28]}. Lind et al. proposed that IL-18-mediated CD11d-restricted iNKT cell activation and subsequent dysregulation play a role in the pathogenesis of human AE, with a decreased proportion of CD4(+) iNKT cells in peripheral blood and increased plasma IL-18 levels. In addition, a reduction in the CD4(+) iNKT cell pool is associated with increased plasma IgE levels, while plasma IL-18 levels are associated with total IgE and disease severity in AD patients^[27]. Increased levels of C-X-C motif chemokine 9 are associated with a reduced risk of AD ^[29]. Han et al. found that the levels of IFN-γ-induced chemokines CXCL9, CXCL10, and CXCL11 were elevated and associated with multiple disease activity markers. These interferon-γ-induced chemokines may contribute to the inflammatory response and skin manifestations in AOSD ^[30]. Our study found that increased levels of Fms-related tyrosine kinase 3 ligand also reduce the risk of AD ^{[31, 32]}. Currently, there are few studies on the relationship between Fms-related tyrosine kinase 3 ligand levels and AD. The rs10085109 SNP in the FLT4 gene is associated with AD susceptibility ^[33]. Yotaro Nishikawa et al. found that congenital defects in cDC not only exacerbate the pathogenesis of AD-like inflammation, but also may cause immune abnormalities through the disruption of the homeostatic feedback loop mediated by Fms-related tyrosine kinase 3 ligand (Flt3L), which is associated with the onset of AD ^[34]. Previous studies have shown that elevated levels of C-X-C motif chemokine 11 are associated with a reduced risk of AD ^{[35, 36]}. In addition, we investigated the bidirectional causal relationship between CRP and AD, and after verification, no significant association was found between the two ^[36].

In summary, our study based on GWAS data from European populations showed that IL18 receptor 1 levels have a protective effect on AD, which was verified in another AD dataset. In addition, we observed a complex bidirectional relationship between IL18 receptor 1 levels, tumor necrosis factor ligand superfamily member 14, and AD, but when AD was the exposure factor, the causal relationship had horizontal pleiotropy when tumor necrosis factor ligand superfamily member 14 was the outcome. The study by Huang et al. demonstrated that TNFSF14 has a potential mediating role in the effect of DHA on AD risk ^{[37, 38]}. Future studies can further investigate the specific roles of IL18 receptor 1 and tumor necrosis factor ligand superfamily member 14 in the pathogenesis of AD, evaluate their potential as biomarkers, and explore therapeutic strategies targeting them.

The results of this study relied on data from European populations, which means that the applicability of its conclusions may be limited. Although these inflammatory proteins showed significant associations for some AD patients in European populations, we must acknowledge that AD patients in other populations around the world may show different degrees of association and significance. Therefore, in order to fully understand the role of these inflammatory proteins and their differences in different populations, future studies should focus on collecting and analyzing data from more diverse populations. Such research efforts will help reveal the population specificity of AD pathogenesis, thereby laying the foundation for the discovery of universally applicable treatment strategies.

In conclusion, our study provides new insights into the role of circulating inflammatory proteins in AD and paves the way for future research and the development of therapeutic strategies. Future studies should focus on validating these findings and exploring the relationship between other potential inflammatory proteins and AD. In addition, a deeper investigation of the specific roles of these inflammatory proteins in the pathophysiology of AD will be crucial.

Data Availability Statement

The original contributions presented in the study are included in the article/supplementary materials, and further inquiries can be directed to the corresponding author.

Author Contributions

L.L.: Writing - original draft, Writing - review and editing. Y.M., Z.N.: Writing - review and editing. B.Z.: Data analysis. F.Y.: Conceptualization, Data curation, X.X.: Formal analysis, Funding acquisition.

Funding Statement

The authors declare that the research, authorship, and/or publication of this article received financial support. This study was supported by the National Natural Science Foundation of China (82174388); Beijing Hundred Thousand Talents Project Training Fund (2019A30); the Fifth Batch of National Excellent Clinical Talents in Traditional Chinese Medicine (National Traditional Chinese Medicine Office Human Education Letter [2021] No. 271); Beijing Municipal Hospital Management Center "Dengfeng" Talent Training Program (DFL20220801).

Conflict of Interest

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.

Maspero J, De Paula M R N, Zhang J, et al. Epidemiology of adult patients with atopic dermatitis in AWARE 1: A second international survey[J]. World Allergy Organ J, 2023,16(3):100724.
Mack M R, Brestoff J R, Berrien-Elliott M M, et al. Blood natural killer cell deficiency reveals an immunotherapy strategy for atopic dermatitis[J]. Sci Transl Med, 2020,12(532).
Girolomoni G, de Bruin-Weller M, Aoki V, et al. Nomenclature and clinical phenotypes of atopic dermatitis[J]. Ther Adv Chronic Dis, 2021,12:364051555.
Yamanaka K, Nakanishi T, Saito H, et al. Persistent release of IL-1s from skin is associated with systemic cardio-vascular disease, emaciation and systemic amyloidosis: the potential of anti-IL-1 therapy for systemic inflammatory diseases[J]. PLoS One, 2014,9(8):e104479.
Namkung J H, Lee J E, Kim E, et al. Association of single nucleotide polymorphisms in the IL-12 (IL-12A and B) and IL-12 receptor (IL-12Rbeta1 and beta2) genes and gene-gene interactions with atopic dermatitis in Koreans[J]. J Dermatol Sci, 2010,57(3):199-206.
Foerster J, Molęda A. Feasibility Analysis of Interleukin-13 as a Target for a Therapeutic Vaccine[J]. Vaccines (Basel), 2019,7(1).
Lee E, Kim J H, Lee S Y, et al. Association of IL13 genetic polymorphisms with atopic dermatitis: Fine mapping and haplotype analysis[J]. Ann Allergy Asthma Immunol, 2020,125(3):287-293.
Novak N, Kruse S, Potreck J, et al. Single nucleotide polymorphisms of the IL18 gene are associated with atopic eczema[J]. J Allergy Clin Immunol, 2005,115(4):828-833.
Ek W E, Karlsson T, Höglund J, et al. Causal effects of inflammatory protein biomarkers on inflammatory diseases[J]. Sci Adv, 2021,7(50):eabl4359.
Bowden J, Holmes M V. Meta-analysis and Mendelian randomization: A review[J]. Res Synth Methods, 2019,10(4):486-496.
Gupta V, Walia G K, Sachdeva M P. 'Mendelian randomization': an approach for exploring causal relations in epidemiology[J]. Public Health, 2017,145:113-119.
Ference B A, Holmes M V, Smith G D. Using Mendelian Randomization to Improve the Design of Randomized Trials[J]. Cold Spring Harb Perspect Med, 2021,11(7).
Ference B A. Mendelian randomization studies: using naturally randomized genetic data to fill evidence gaps[J]. Curr Opin Lipidol, 2015,26(6):566-571.
Bottigliengo D, Foco L, Seibler P, et al. A Mendelian randomization study investigating the causal role of inflammation on Parkinson's disease[J]. Brain, 2022,145(10):3444-3453.
Liu R, Shi X, Feng J, et al. Ischemic Stroke and Cerebral Microbleeds: A Two-Sample Bidirectional Mendelian Randomization Study[J]. Neurol Ther, 2023,12(4):1299-1308.
Ren F, Jin Q, Jin Q, et al. Genetic evidence supporting the causal role of gut microbiota in chronic kidney disease and chronic systemic inflammation in CKD: a bilateral two-sample Mendelian randomization study[J]. Front Immunol, 2023,14:1287698.
Zhao J H, Stacey D, Eriksson N, et al. Genetics of circulating inflammatory proteins identifies drivers of immune-mediated disease risk and therapeutic targets[J]. Nat Immunol, 2023,24(9):1540-1551.
Ahola-Olli A V, Würtz P, Havulinna A S, et al. Genome-wide Association Study Identifies 27 Loci Influencing Concentrations of Circulating Cytokines and Growth Factors[J]. Am J Hum Genet, 2017,100(1):40-50.
Ligthart S, Vaez A, Võsa U, et al. Genome Analyses of >200,000 Individuals Identify 58 Loci for Chronic Inflammation and Highlight Pathways that Link Inflammation and Complex Disorders[J]. Am J Hum Genet, 2018,103(5):691-706.
Paternoster L, Standl M, Waage J, et al. Multi-ancestry genome-wide association study of 21,000 cases and 95,000 controls identifies new risk loci for atopic dermatitis[J]. Nat Genet, 2015,47(12):1449-1456.
Bowden J, Del G M F, Minelli C, et al. A framework for the investigation of pleiotropy in two-sample summary data Mendelian randomization[J]. Stat Med, 2017,36(11):1783-1802.
Hemani G, Bowden J, Davey S G. Evaluating the potential role of pleiotropy in Mendelian randomization studies[J]. Hum Mol Genet, 2018,27(R2):R195-R208.
Burgess S, Thompson S G. Interpreting findings from Mendelian randomization using the MR-Egger method[J]. Eur J Epidemiol, 2017,32(5):377-389.
Bowden J, Davey S G, Haycock P C, et al. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator[J]. Genet Epidemiol, 2016,40(4):304-314.
Hu Y, Wang J, Zhang H, et al. Enhanced Expression of IL-18 and IL-18BP in Plasma of Patients with Eczema: Altered Expression of IL-18BP and IL-18 Receptor on Mast Cells[J]. Mediators Inflamm, 2017,2017:3090782.
Wittmann M, Purwar R, Hartmann C, et al. Human keratinocytes respond to interleukin-18: implication for the course of chronic inflammatory skin diseases[J]. J Invest Dermatol, 2005,124(6):1225-1233.
Lind S M, Kuylenstierna C, Moll M, et al. IL-18 skews the invariant NKT-cell population via autoreactive activation in atopic eczema[J]. Eur J Immunol, 2009,39(8):2293-2301.
Kim E, Lee J E, Namkung J H, et al. Association of the single-nucleotide polymorphism and haplotype of the interleukin 18 gene with atopic dermatitis in Koreans[J]. Clin Exp Allergy, 2007,37(6):865-871.
Brunner P M, Suárez-Fariñas M, He H, et al. The atopic dermatitis blood signature is characterized by increases in inflammatory and cardiovascular risk proteins[J]. Sci Rep, 2017,7(1):8707.
Han J H, Suh C H, Jung J Y, et al. Elevated circulating levels of the interferon-γ-induced chemokines are associated with disease activity and cutaneous manifestations in adult-onset Still's disease[J]. Sci Rep, 2017,7:46652.
Momenilandi M, Lévy R, Sobrino S, et al. FLT3L governs the development of partially overlapping hematopoietic lineages in humans and mice[J]. Cell, 2024,187(11):2817-2837.
Long J, Dang H, Su W, et al. Interactions between circulating inflammatory factors and autism spectrum disorder: a bidirectional Mendelian randomization study in European population[J]. Front Immunol, 2024,15:1370276.
Namkung J H, Lee J E, Kim E, et al. Single nucleotide polymorphism in the FLT4 gene is associated with atopic dermatitis in Koreans[J]. Cytokine, 2013,62(1):110-114.
Nishikawa Y, Fukaya T, Fukui T, et al. Congenital Deficiency of Conventional Dendritic Cells Promotes the Development of Atopic Dermatitis-Like Inflammation[J]. Front Immunol, 2021,12:712676.
Sinikumpu S P, Huilaja L, Auvinen J, et al. The Association Between Low Grade Systemic Inflammation and Skin Diseases: A Cross-sectional Survey in the Northern Finland Birth Cohort 1966[J]. Acta Derm Venereol, 2018,98(1):65-69.
Vekaria A S, Brunner P M, Aleisa A I, et al. Moderate-to-severe atopic dermatitis patients show increases in serum C-reactive protein levels, correlating with skin disease activity[J]. F1000Res, 2017,6:1712.
Huang X W, Pang S W, Yang L Z, et al. TNFSF14 mediates the impact of docosahexaenoic acid on atopic dermatitis: a Mendelian randomization study[J]. Eur Rev Med Pharmacol Sci, 2024,28(1):107-117.
Ikawa T, Ichimura Y, Miyagawa T, et al. The Contribution of LIGHT (TNFSF14) to the Development of Systemic Sclerosis by Modulating IL-6 and T Helper Type 1 Chemokine Expression in Dermal Fibroblasts[J]. J Invest Dermatol, 2022,142(6):1541-1551.

No competing interests reported.

Download PDF

Reviews received at journal
12 Oct, 2024
Reviewers agreed at journal
04 Oct, 2024
Reviews received at journal
18 Sep, 2024
Reviewers agreed at journal
03 Sep, 2024
Reviewers invited by journal
30 Aug, 2024
Editor assigned by journal
30 Aug, 2024
Editor invited by journal
17 Aug, 2024
Submission checks completed at journal
14 Aug, 2024
First submitted to journal
02 Jul, 2024

You are reading this latest preprint version

Exploring the causal relationship between inflammatory cytokines and C-reactive protein levels and atopic dermatitis: a two-way two-sample Mendelian randomization study

Status:

Version 1

Abstract

Figures

1. Introduction

2. Research methods and materials

2.1 Research design

2.2 Data Sources

2.3 Instrumental variable selection

2.4 Statistical analysis

2.5 External dataset validation

3. Results

4. Discussion

5. Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1