The Causal Relationship between Plasma Myeloperoxidase Levels and Respiratory Tract Infections: A Bidirectional Mendelian Randomization Study

Background Observational researches reported the underlying correlation of plasma myeloperoxidase (MPO) concentration with respiratory tract infections (RTIs), but their causality remained unclear. Here, we examined the cause–effect relation between plasma MPO levels and RTIs. Materials and Methods Datasets of plasma MPO levels were from the Folkersen et al. study (n = 21,758) and INTERVAL study (n = 3,301). Summarized data for upper respiratory tract infection (URTI) (2,795 cases and 483,689 controls) and lower respiratory tract infection (LRTI) in the intensive care unit (ICU) (585 cases and 430,780 controls) were from the UK Biobank database. The primary method for Mendelian randomization (MR) analysis was the inverse variance weighted approach, with MR-Egger and weighted median methods as supplements. Cochrane's Q test, MR-Egger intercept test, MR pleiotropy residual sum and outliers global test, funnel plots, and leave-one-out analysis were used for sensitivity analysis. Results We found that plasma MPO levels were positively associated with URTI (odds ratio (OR) = 1.135; 95% confidence interval (CI) = 1.011–1.274; P=0.032) and LRTI (ICU) (OR = 1.323; 95% CI = 1.006–1.739; P=0.045). The consistent impact direction is shown when additional plasma MPO level genome-wide association study datasets are used (URTI: OR = 1.158; 95% CI = 1.072–1.251; P < 0.001; LRTI (ICU): OR = 1.216; 95% CI = 1.020–1.450; P=0.030). There was no evidence of a causal effect of URTI and LRTI (ICU) on plasma MPO concentration in the reverse analysis (P > 0.050). The sensitivity analysis revealed no violations of MR presumptions. Conclusions Plasma MPO levels may causally affect the risks of URTI and LRTI (ICU). In contrast, the causal role of URTI and LRTI (ICU) on plasma MPO concentration was not supported in our MR analysis. Further studies are needed to identify the relationship between RTIs and plasma MPO levels.


Introduction
Myeloperoxidase (MPO), a member of the heme peroxidasecyclooxygenase superfamily [1], demonstrates elevated expression levels in neutrophil granulocytes, while exhibiting comparatively lower expression levels in monocytes and macrophages [2][3][4].It not only catalyzes hypochlorous acid (HOCl) and reactive oxygen species (ROS) to stimulate neutrophils to phagocytose bacteria and other microorganisms [5] but also actively participates in inflammation regulation, immune cell recruitment at infection sites, and influences inflammation resolution [6].
Respiratory tract infections (RTIs) are the most common infections seen in primary care and the single most significant contributor to the overall burden of disease worldwide [7].In the most severe instances, RTIs can progress rapidly into sepsis, multiorgan failure, and even death.Therefore, it is necessary to identify potential risk factors for RTIs and thus improve global public health.The role of MPO in inflammation, particularly in RTIs, has received more attention in recent years [8].Several studies have reported an elevation in plasma MPO levels among patients with various respiratory conditions, such as influenza, SARS-CoV-2 infection, exacerbations of chronic bronchitis with airway obstruction, frequent exacerbations of chronic bronchitis with airway obstruction, acute respiratory distress syndrome (ARDS), and sepsis when compared to individuals without these conditions [9][10][11][12][13][14][15].However, whether individuals exhibiting heightened levels of plasma MPO are more susceptible to RTIs is uncertain.Consequently, it is essential to investigate the causal effects of plasma MPO levels on RTIs at the scale of the entire population.
Mendelian randomization (MR) is an epidemiological tool using genetic variations (single-nucleotide polymorphisms, SNPs) as instrumental variables (IVs) to estimate the causal relationship between exposures and outcomes at a population level [16].Genetic variation is randomly assigned to offspring during conception, making it less susceptible to interference from inverse associations and confounding factors compared to traditional randomized controlled trials (RCTs) [17].A recent MR study genetically determined that elevated plasma MPO levels are causally associated with increased risks of ischemic stroke, cardioembolic stroke (CES), heart failure (HF), and atrial fibrillation (AF) [18].However, no MR research has been conducted on establishing a causal link between plasma MPO levels and RTIs.
Therefore, the objective of this current study is to conduct a two-sample bidirectional MR study using summarylevel data from a genome-wide association study (GWAS), aiming to elucidate the potential causal bidirectional relationship between plasma MPO levels and RTIs, including upper respiratory tract infection (URTI) and lower respiratory tract infection (LRTI) in the intensive care unit (ICU).

Study Design.
The study design for bidirectional MR analysis is illustrated in Figure 1.Three fundamental assumptions underlie the design of MR are as follows: (1) genetic variants directly influence exposures; (2) genetic variants are not associated with potential confounders; and (3) genetic variants affect outcomes only via the effects on exposures [16].All original studies acquired ethical approval and obtained informed consent from the participants.The data utilized in this research were readily accessible within the public domain, thus obviating the necessity for ethical approval and informed consent in accordance with the study's design.Meanwhile, the results of this study were reported in adherence to the Strengthening the Reporting of Observational Studies in Epidemiology-Mendelian Randomization (STROBE-MR) guidance from 2021 [19].

GWAS Data Summary for Exposures and Outcomes.
After searching the IEU OpenGWAS project (https://gwas.mrcieu.ac.uk/), we identified two European-descent GWASs [20,21] with the SNPs of plasma MPO levels and the UK Biobank database with the SNPs of URTI and LRTI (ICU).Details of these GWASs are displayed in Table 1.Folkersen et al. [20] collected 13,138,585 SNPs from 21,758 Europeans, while the INTERVAL study [21] collected 10,534,735 SNPs from 3,301 participants.Among them, the Folkersen et al. [20] study is the primary dataset, and the INTERVAL study [21] is the supplementary dataset.URTI (2,795 cases and 483,689 controls) and LRTI (ICU) SNPs with F > 10.The F statistics for each SNP were computed using the following formula:  [24,25].Leave-one-out analysis to evaluate whether a single SNP drove or biased the MR estimate [16].
2.6.Statistical Analysis.The odds ratio (OR), beta, and 95% confidence interval (CI) were utilized to present the causal estimates, and a P value of less than 0.05 was deemed statistically significant.The statistical significance threshold of sensitivity analyses was set at P >0:05.The forest plots, scatter plot, leave-one-out plot, funnel plot, and all statistical analyses performed in this study were performed in R (version 4.2.3) using the "TwoSampleMR" package (version 0.4.20;https:// github.com/MRCIEU/TwoSampleMR).

Genetic Prediction of Plasma MPO Levels for RTIs Risk
3.1.1.IV Selection.We first selected the IVs for plasma MPO levels to assess the causal impact of plasma MPO levels on the risk of URTI and LRTI (ICU).Initially, 33 SNPs closely linked to plasma MPO levels were retrieved from the primary dataset [20] and 13 from the supplementary dataset [21].Following strict implementation of the IV selection procedure (as described in Section 2), 28 SNPs were used as IVs between plasma MPO levels and URTI, and 29 SNPs were used as IVs between plasma MPO levels and LRTI (ICU) in the primary MR study (Supplementary Tables S1  and S3).In the supplementary MR study, 11 SNPs were identified as IVs between plasma MPO levels and URTI, as well as between plasma MPO levels and LRTI (ICU) (Supplementary Tables S2 and S4).The F statistic values were all ≥10 (Supplementary Tables S1-S4), suggesting that the possibility of weak instrument bias is slight.Supplementary Tables S8  and S9 provide the SNP filtering process.

Plasma MPO Levels and URTI.
In the primary MR study, based on the IVW method with the random effect model, a significant association was observed between plasma MPO levels and the risk of URTI (OR = 1.135, 95% CI = 1.011-1.274,P ¼ 0:032).The weighted median results were consistent with IVW (OR = 1.261, 95% CI = 1.003-1.586,P ¼ 0:048).While the MR-Egger analysis did not show a statistically significant relationship between plasma MPO levels and the risk of URTI (OR = 1.198, 95% CI = 0.892-1.610,P ¼ 0:242), the direction of effect aligned with the main analysis, especially IVW (Figure 2(a)).The supplementary MR study had a similar causal effect of plasma MPO levels on URTI.The combined OR estimated through the IVW method with the random effect model was 1.158 (95% CI: 1.072-1.251,P <0:001) (Figure 2(a)).The scatter plot and forest plot showed the overall causal effect estimation and the causal effect estimation of individual SNPs (Figures 3(a) and 3(c), Figures 4(a) and 4(c)).We performed several sensitivity analyses to evaluate the robustness of the causal effect estimates of plasma MPO    2).The leave-one-out sensitivity analysis, as shown in Figures 3(d) and 4(d), demonstrated that any SNPs had little effect on the overall effect of causal relationships.Last, the funnel plots for MR analysis in Figures 3(b) and 4(b) revealed that the data points were equally distributed around the funnel, indicating that no substantial asymmetry existed and that there was no evidence of horizontal pleiotropy.

Plasma MPO Levels and LRTI (ICU).
In the primary MR study, the IVW method with the random effect model suggested a significant association between plasma MPO levels and the risk of LRTI (ICU) (OR = 1.323, 95% CI = 1.006-1.739,P ¼ 0:045).While the weighted median (OR = 1.341, 95% CI = 0.733-2.453,P ¼ 0:349) and the MR-Egger analysis (OR = 1.483, 95% CI = 0.929-2.368,P ¼ 0:099) did not demonstrate a statistically significant relationship between plasma MPO levels and the risk of LRTI (ICU), the direction of effect was consistent with the IVW method with the random effect model (Figure 2 Additionally, no evidence of heterogeneity or horizontal pleiotropy was observed in these analyses (all P >0:05).(Table 2).The leave-one-out sensitivity analysis, as shown in Figures 5(d) and 6(d), demonstrated that the overall estimates were not disproportionately affected by any individual SNP.Additionally, the funnel plots in Figures 5(b) and 6(b) show no evidence of horizontal pleiotropy.

Discussion
Previous observational and clinical studies have shown an association between plasma MPO levels and RTIs, but the exact causal relationships have yet to be well established.In this study, we conducted a bidirectional MR analysis to systematically explore the causative relationships among plasma MPO levels and RTIs based on summary-level data from large-scale GWASs.We showed a direct causality between higher plasma MPO levels and a higher URTI and LRTI (ICU) risk for the first time.In contrast, the causal role of URTI and LRTI (ICU) on plasma MPO concentration was not supported in our MR analysis.
Observational studies have provided abundant evidence for the association between MPO and the risk of RTIs.For instance, in an analysis based on three pediatric patients with ARDS infected with H5N1 influenza and 31 non-H5N1 influenza-infected ARDS children, plasma MPO levels were Horizontal pleiotropy analyses were conducted by MR-Egger regression and MR-PRESSO methods, and the results showed that there is no evidence of horizontal pleiotropy in IVs of infectious diseases (all P >0:05).Heterogeneity tests were conducted by MR-Egger regression and IVW with the fixed model, and the results showed that there is no evidence of heterogeneity in IVs of infectious diseases (all P >0:05).PRESSO, MR-Pleiotropy RESidual Sum and Outlier; IVW, inverse variance weighted method with the fixed effect model.MPO * , MPO-related SNPs from the Folkersen et al. [20] study; MPO#, MPO-related SNPs from INTERVAL study [21].
Mediators of Inflammation Mediators of Inflammation higher in pediatric patients with ARDS infected with H5N1 influenza than in non-H5N1 influenza-infected ARDS children [13].In addition, a prospective study of 279 individuals showed that plasma MPO levels in nonsevere (NS), severe (S), and postacute phase (PAP) COVID-19 patients were significantly different from the levels in healthy individuals, and plasma MPO levels had high diagnostic power for the disease severity of COVID-19 [26].Moreover, Chang and Yao [12] conducted a prospective cohort study on adult acute exacerbation of chronic obstructive pulmonary disease (AECOPD) patients, suggesting that plasma MPO levels in patients with frequent AECOPD exacerbations were significantly higher than in patients with infrequent AECOPD exacerbations.Similar associations were reported for MPO levels in respiratory specimens and the risk of respiratory infection [27][28][29][30].Our findings further highlight these associations.We found that plasma MPO concentration was genetically associated with increased risks of respiratory infections (URTI: OR = 1.135, 95% CI = 1.011-1.274,P ¼ 0:032; LRTI in the critical care units: OR = 1.323, 95% CI = 1.006-1.739,P ¼ 0:045).Unfortunately, null causal effects of URTI and LRTI (ICU) risk on plasma MPO concentration were observed in our MR study (URTI: beta = 0.021, 95% CI = −0.005-0.046,P ¼ 0:111; LRTI in the critical care units: beta = 0.013, 95% CI = −0.001-0.026,P ¼ 0:068).Several possible pathophysiological mechanisms underlying the detrimental effect of plasma MPO levels on the risks of RTIs have been suggested by previous studies, including impaired neutrophil function, oxidative stress, and altered immune regulation [31][32][33].MPO is an essential component of neutrophils, but excessive release of MPO can impair their function.Thus, elevated levels of MPO may lead to neutrophil dysfunction, making it more difficult for the immune system to clear respiratory pathogens effectively [31].MPO and its oxidative products can directly harm the respiratory epithelium, increase airway epithelial permeability, compromise the integrity of the mucosal barrier, and decrease the mucociliary clearance mechanism, making it easier for pathogens to establish infections in the airways [32].In addition, MPO has been shown to modulate the activity of various immune cells, such as macrophages and lymphocytes [33].Dysregulation of immune responses due to elevated MPO levels can lead to an imbalance in the immune system, potentially impairing the body's ability to defend against respiratory pathogens effectively.Therefore, the detailed mechanism underlying the different associations of plasma MPO levels with RTIs and their subtypes warrants further study.
To the best of our knowledge, this study is the first to assess plasma MPO levels as a causal risk factor for RTIs using the MR design with data from a substantial number of individuals.There are several important public health significances and clinical implications.In the present MR study, we demonstrated the potential causal relationships between plasma MPO levels and incidence of URTI and LRTI (ICU) from the genetic insights, which might provide novel clues for preventing RTIs.According to our findings, plasma MPO levels could be a promising biomarker for identifying high-risk individuals for active surveillance and early intervention of RTIs.Furthermore, investigating whether targeting MPO could reduce the risk of RTIs is of clinical interest.
However, several limitations were also present.First, all GWAS data came from the European population.Whether our described findings would be consistent in other populations remains to be investigated.Second, the plasma MPO levels may be affected by both genes and the environment.However, our results can only explain the relationship between the changes in plasma MPO levels caused by genetic variation and infections [34].Third, the GWAS database did not include detailed demographic characteristics or clinical data.Therefore, subgroup analysis cannot be further performed.Last, as mentioned in our previous method, three assumptions must be met to use variables as genetic instruments in MR analysis.If the first assumption is not satisfied, a "weak instrument problem," such as weak statistical power and increased bias due to pleiotropic effects, occurs.The first assumption is tested by checking whether the F-statistic exceeds 10.Our study's F-statistic values were all ≥10, indicating no relevance assumption violation.However, due to the relatively small number of SNPs related to MPO we obtained, we did not validate Hypotheses 2 and 3, which may have led to some bias in our results.

Conclusion
Our two-sample MR study provides strong evidence for a causal relationship between plasma MPO levels and RTIs.We found that plasma MPO levels increased the risk of URTI and LRTI (ICU).However, there was no evidence of reverse causation.Our findings prompt future studies to investigate and confirm the role of MPO as a clinical biomarker that regulates the risk of RTIs and its potential role in therapeutic interventions.

FIGURE 3 :FIGURE 4 :
FIGURE 3: Primary MR analysis results of the causal effect of plasma MPO levels on URTI: (a) comparison of the five MR analysis methods employed; (b) funnel plot assessing directional horizontal pleiotropy; (c) forest plot displaying the effect estimates; (d) leave-one-out analyses detecting outliers.
(a)).Analogous causal effects of plasma MPO concentration on LRTI (ICU) were observed in the supplementary MR study.The combined OR estimated through the IVW method with the random effect model was 1.216 (95% CI: 1.020-1.450,P ¼ 0:030) (Figure2(a)).The MR regression slopes and individual causal estimates of each SNP are illustrated in Figures5(a) and 5(c), Figures 6(a) and 6(c).

FIGURE 5 :FIGURE 6 :
FIGURE 5: Primary MR analysis results of the causal effect of plasma MPO levels on LRTI (ICU): (a) comparison of the five MR analysis methods employed; (b) funnel plot assessing directional horizontal pleiotropy; (c) forest plot displaying the effect estimates; (d) leave-one-out analyses detecting outliers.
Study design for the MPO-RTIs two-sample bidirectional MR analyses.SNP, single-nucleotide polymorphism; URTI, upper respiratory tract infection; LRTI (ICU), lower respiratory tract infection in the intensive care unit.

TABLE 1 :
Details of the GWASs included in the Mendelian randomization.
ðbÞ FIGURE 2: MR analysis of the effect of plasma MPO levels on RTIs (a) and MR analysis of the effect of RTIs on plasma MPO levels (b).IVs, instrumental variables; OR, odds ratio; OR (95% CI), 95% confidence interval of odds ratio.Beta (95% CI), 95% confidence interval of beta.

TABLE 2 :
Pleiotropy and heterogeneity test of the bidirectional Mendelian randomization study.