DOI: https://doi.org/10.21203/rs.3.rs-2885951/v1
Background
Many COVID-19-infected patients have been observed to develop unexplained valvular heart disease (VHD), and the association between COVID-19 and VHD remains inconclusive. Therefore, we conducted a two-sample Mendelian randomization study to infer causality between COVID-19 and VHD from a genetic perspective using COVID-19 genetic tools.
Methods
This study used genetic variables and summary statistics from COVID-19 and VHD genome-wide association studies (GWAS). Single nucleotide polymorphisms (SNPs) were selected based on the assumption of instrumental variables (IVs). The inverse-variance weighted (IVW) method was used as the main analysis method to summarize the causal effects between exposure and outcome, while the weighted median and weighted mode methods were used as secondary methods. MR-Egger was used to test for horizontal pleiotropy, and the Q-test was used to test for heterogeneity. Sensitivity analysis was conducted using leave-one-out method. Scatterplots, forest plots, and funnel plots were used to visualize the results of MR analysis.
Results
In this study, seven COVID-19-related SNPs were selected as IVs, and the IVW [odds ratio (OR) = 1.16, 95% confidence interval (CI) = 1.04 − 1.28, P = 0.008], weighted median (OR = 1.21, 95% CI = 1.06 − 1.39, P = 0.006), and weighted mode (OR = 1.27, 95% CI = 1.05 − 1.54, P = 0.047) analysis methods suggested a causal effect of COVID-19 on CHD. MR-Egger indicated no evidence of horizontal pleiotropy (P = 0.589), and the Q-test suggested no heterogeneity (IVW, P = 0.349). Sensitivity analysis indicated robustness of the MR analysis results.
Conclusions
MR analysis revealed a causal effect of COVID-19 infection on the occurrence of VHD, indicating that patients with COVID-19 had a higher risk of VHD.
COVID-19 has become a major global public health challenge, and numerous studies have shown its significant impact on the cardiovascular system(1, 2). However, there is currently no research report on the causal relationship between COVID-19 and valvular heart disease (VHD). VHD is an important factor leading to physical disability, decreased quality of life, and shortened life expectancy(3). Due to the decrease in early mortality and population aging, the incidence of VHD is increasing(4, 5). We have observed several cases of VHD in patients with COVID-19 in clinical practice, and relevant case reports have also been reported by scholars(6, 7). Therefore, studying the causal relationship between COVID-19 and VHD is of great clinical significance. The aim of this study is to use Mendelian randomization (MR) to explore the causal relationship between COVID-19 and VHD.
MR study is a data analysis method that has been mainly applied to epidemiological causal inference in recent years(8). It can evaluate the causal effect of one factor on another factor without bias(9). We used two independent samples for MR analysis. The first sample came from “The COVID-19 Host Genetics Initiative”, and the other sample was a dataset of VHD from the FinnGen biological database(10). By combining these two samples, we can establish the causal relationship between COVID-19 and VHD.
In this study, we used a genetic marker-based MR method, which can avoid common confounding factors such as lifestyle and environmental factors and reduce the bias of causal effects(11). We selected COVID-19-related single nucleotide polymorphisms (SNPs) from the exposure dataset as instrumental variables (IVs) by analyzing SNPs and calculating the causal effect of COVID-19 on VHD. In addition, we also conducted a series of sensitivity analyses to evaluate the robustness and reliability of the results.
Herein, the study used the MR method to explore the causal relationship between COVID-19 and VHD. Our research results may provide new perspectives and ideas for the prevention and treatment of COVID-19-related cardiovascular complications.
Study design
This study utilized the MR approach to explore the causal relationship between COVID-19 and VHD at the genetic level. All data involved in this study were obtained from publicly available databases, which had obtained informed consent and completed ethical review prior to publication(12). Therefore, no additional ethical approval was required for this study. COVID-19 related SNPs that met the criteria were selected as IVs. The causal effect of COVID-19 on VHD was synthesized using inverse variance weighting (IVW) method and other methods. Sensitivity analyses were performed using Cochran’s Q test, MR-Egger, and leave-one-out analysis. Finally, the results of MR were visualized. The workflow is shown in Fig. 1.
Data source
The data for this study were obtained from a publicly available database (https://gwas.mrcieu.ac.uk/). The COVID-19 GWAS summary statistics data were selected from “The COVID-19 Host Genetics Initiative” project and the research results were released in 2020 (PMID: 32404885). The summary statistics data (id: ebi-a-GCST011073) included 38,984 case groups and 1,644,784 control groups of European population, with approximately 8.66 million variants that could be tested. The study cohort included both males and females. This is currently the most up-to-date GWAS research data for COVID-19. MR is based on three core assumptions(13). First, the genetic variant used as an instrument must be strongly associated with the COVID-19. This means that SNPs should be able to explain a large proportion of the variability in the exposure COVID-19. Second, the genetic variant used as an instrument must be independent of any confounders that may be associated with both COVID-19 and VHD. This means that SNPs should not be associated with any other factors that could influence the outcome, apart from its effect on the exposure. Finally, the genetic variant used as an instrument must only affect the outcome through its effect on the exposure. This means that the variant should not have any other direct or indirect effects on the outcome. If these assumptions are met, then MR can be used to provide evidence for a causal relationship between the exposure and the outcome. Thus, SNPs were selected using the following criteria: i) SNPs were extracted from the COVID-19-related GWAS dataset at the genome-wide significance level (p < 5×10− 8) to demonstrate a strong association between SNPs and the exposure COVID-19; ii) SNPs were selected for independence from other SNPs to avoid bias due to linkage disequilibrium (LD), and the LD of SNPs associated with COVID-19 was required to have an r2 < 0.001, with a window size of 10,000 kb; and iii) the correlation between instrumental variables and exposure was determined using the F-statistic(14). Generally, IVs with an F-statistic > 10 are considered unbiased. The F-statistic was calculated as (β/SE)2. The “TwoSampleMR” (version 0.5.6) software package was used for IVs filtering, MR, and the sensitivity analysis below in R version 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria)(15).
The GWAS data related to valvular heart disease are obtained from the FinnGen project, which can be accessed at https://gwas.mrcieu.ac.uk/datasets/finn-b-I9_VHD/. This study includes 38,209 cases with valvular heart disease and 156,711 control cases, with a total of 16,380,358 SNPs. All study participants are of European ancestry. The study cohort included both males and females. This is currently the most up-to-date GWAS research data for VHD. The study harmonizes the data to remove palindromic SNPs and ensure that the effect of the SNPs on the exposure and the effect of the SNPs on the outcome correspond to the same allele before MR estimates were calculated.
Mendelian randomization analysis
In order to assess the correlation between exposure and outcome, this study used three MR models to summarize the causal effects between COVID-19 and VHD: Inverse variance weighted (IVW), Weighted median, and Weighted mode. When there is neither horizontal pleiotropy nor heterogeneity in the IVs, the best analytical method is IVW, which can provide the most accurate estimation. The other two methods can be used to test its results. IVW is the most commonly used and primary analysis method in MR studies, which evaluates the causal effect of COVID-19 on VHD by combining the Wald ratio of each SNP causal effect(16). Weighted median can accurately calculate the causal association effect even when less than 50% genetic variation violate the core assumptions of MR(17). In cases where there is only heterogeneity and no pleiotropy, the results of the Weighted median method should be prioritized.
quality control
Horizontal pleiotropy refers to a phenomenon in which IVs affect the outcome through pathways other than exposure, which is a potential source of bias(18). Therefore, it is very important to test for horizontal pleiotropy, as its existence suggests that the IVs has not met the second assumption of MR. In this study, the MR-Egger method was used to assess horizontal pleiotropy. The MR-Egger method fits a regression model through the association effects of the gene-outcome and gene-exposure, and tests and corrects for bias generated by IVs pleiotropy(19). If the IVs do not exhibit pleiotropy, the intercept of the model should be zero, and therefore the intercept can be used as a test statistic to detect the presence of pleiotropy. The slope of the MR-Egger regression reflects the causal association effect of exposure on the outcome after correcting for pleiotropy bias. A P value greater than 0.05 indicates that the IVs have no pleiotropy.
The narrow sense of sensitivity analysis refers to the leave-one-out analysis. Leave-one-out sensitivity analysis is performed to determine whether the causal association is disproportionately affected by a single SNP. Each selected SNP is sequentially removed, and the effect size of the remaining SNPs is calculated and analyzed. If the removal of a particular SNP has a significant impact on the effect size, the result is deemed unsatisfactory. The results of the leave-one-out analysis are presented using a forest plot.
Heterogeneity is the variability in the causal estimates of each SNP. Even if all SNPs are valid, some heterogeneity may still exist. There may be apparent heterogeneity when there are outlier effects that may represent pleiotropic variation, or when the evidence for a causal effect relies on one or few variants(20). Low heterogeneity reflects the consistency of causal estimates for all SNPs, indicating improved reliability of MR estimates. In this study, the “TwoSampleMR” package was used to conduct Cochran’s Q Statistic to evaluate heterogeneity. A p-value greater than 0.05 was considered an ideal result.
Instrumental variables for COVID-19
We identified seven COVID-19-related SNPs as IVs: rs10936744, rs12482060, rs17078348, rs2271616, rs4971066, rs643434, and rs757405. All genetic instruments showed genome-wide significance with a p-value < 5×10− 8, and none of them were weak instruments (F > 10). The outcome-related SNPs were not influenced by the IVs. The characteristics of COVID-19-related SNPs are presented in Table 1.
| No. | SNP | Gene | Chr. | EA | OA | EAF. COVID-19 | EAF.VHD | β. COVID-19 | β. VHD | SE. COVID-19 | SE. VHD |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | rs10936744 | / | 3 | T | C | 0.359 | 0.363 | -0.063 | -0.013 | 0.010 | 0.010 |
| 2 | rs12482060 | IFNAR2 | 21 | G | C | 0.338 | 0.377 | 0.062 | 0.000 | 0.011 | 0.010 |
| 3 | rs17078348 | / | 3 | G | A | 0.100 | 0.088 | 0.092 | -0.005 | 0.016 | 0.016 |
| 4 | rs2271616 | SLC6A20 | 3 | T | G | 0.118 | 0.058 | 0.156 | -0.006 | 0.015 | 0.020 |
| 5 | rs4971066 | EFNA1 | 1 | G | T | 0.178 | 0.212 | -0.077 | -0.012 | 0.013 | 0.011 |
| 6 | rs643434 | ABO | 9 | A | G | 0.371 | 0.456 | 0.101 | 0.027 | 0.010 | 0.009 |
| 7 | rs757405 | OAS3 | 12 | A | T | 0.709 | 0.791 | 0.069 | 0.020 | 0.011 | 0.011 |
| SNP, single nucleotide polymorphism; COVID-19, the coronavirus disease 2019; EA, effect allele; OA, other allele; EAF, effect allele frequencies, VHD, valvular heart disease. | |||||||||||
Results of the Mendelian randomization analyses for valvular heart disease
This study employed various MR methods to estimate the causal effect of COVID-19 on VHD. The causal effect estimates of each SNP were calculated based on the Wald ratio and presented in a forest plot (Fig. 2). The MR estimates of COVID-19 on VHD were synthesized using IVW [odds ratio (OR) = 1.16, 95% confidence interval (CI) = 1.04 − 1.28, beta = 0.145, SE = 0.054, P = 0.008], weighted median (OR = 1.21, 95% CI = 1.06 − 1.39, beta = 0.192, SE = 0.070, P = 0.006), and weighted mode (OR = 1.27, 95%CI = 1.05 − 1.54, beta = 0.238, SE = 0.095, P = 0.047) and the results were presented in Table 2. The causal effects obtained from the three different MR methods exhibit consistent directionality, indicating the credibility of the research results. Therefore, we found that patients with COVID-19 had a higher risk of VHD (IVW, OR = 1.16, 95% CI = 1.04 − 1.28, P = 0.008). The MR study suggests that COVID-19-related SNPs should meet the condition of being independent of confounding factors associated with both COVID-19 and VHD. To verify this hypothesis, the study used MR-Egger regression analysis for pleiotropic analysis, and the results showed an intercept of 0.010 with a p-value of 0.589, which was not statistically significant. This represents that the SNPs used in this study did not exhibit pleiotropy. The study conducted heterogeneity tests using Cochran’s Q statistic on SNPs in the IVW (Q = 6.702, P = 0.349) method (Table 2). The results did not demonstrate statistical significance, indicating that the SNPs in this study do not exhibit heterogeneity.
| MR methods | nSnp | OR | 95%CI | Beta | SE | P Value | Heterogeneity Test | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Cochran’s Q Statistic | P Value | ||||||||||
| MR Egger | 7 | 1.03 | 0.68,1.56 | 0.027 | 0.212 | 0.904 | 6.285 | 0.279 | |||
| Weighted median | 7 | 1.21 | 1.06,1.39 | 0.192 | 0.070 | 0.006 | |||||
| Inverse variance weighted | 7 | 1.16 | 1.04,1.28 | 0.145 | 0.054 | 0.008 | 6.702 | 0.349 | |||
| Simple mode | 7 | 1.25 | 0.96,1.64 | 0.226 | 0.132 | 0.137 | |||||
| Weighted mode | 7 | 1.27 | 1.05,1.54 | 0.238 | 0.095 | 0.047 | |||||
| MR, mendelian randomization; COVID-19, the coronavirus disease 2019; VHD, valvular heart disease; Snp, single nucleotide polymorphism; OR, odds ratio; CI, confidence interval; SE, standard error. | |||||||||||
Visualization of Mendelian randomization results
Based on COVID-19-related SNPs, the effect of COVID-19 on VHD can be calculated. According to the principle of mediation effect, the effect of COVID-19-related SNPs on VHD is equal to the effect of COVID-19-related SNPs on COVID-19 multiplied by the effect of COVID-19 on VHD. Therefore, the causal relationship between COVID-19 and VHD can be expressed as the effect of COVID-19-related SNPs on VHD divided by the effect of COVID-19-related SNPs on COVID-19, which is the slope of the linear regression in the scatter plot (Fig. 3). The study used leave-one-out analysis to perform sensitivity testing, and the results were presented in a forest plot (Fig. 4). Each black dot represents the MR analysis using the IVW method after excluding a specific SNP. The overall analysis including all SNPs was also displayed for comparison. During the process of individually removing SNPs, the causal effect values changed very little, and there was no change in direction. This suggests that removing any SNP would not have a fundamental impact on the results, and no instrument variable strongly influences the estimation of causal effects as an outlier. A funnel plot was also used to assess heterogeneity, with greater dispersion indicating higher heterogeneity. The distribution of SNPs in the funnel plot is symmetrical with a vertical line as the axis of symmetry, indicating that SNPs do not exhibit directional horizontal pleiotropy, which is favorable for robust MR estimation (Fig. 5).
To our knowledge, this study employed a two-sample MR method for the first time to investigate the causal relationship between COVID-19 and VHD. In summary, we selected seven COVID-19 related SNPs, including rs10936744, rs12482060, rs17078348, rs2271616, rs4971066, rs643434, and rs757405, as IVs by screening the GWAS dataset of “The COVID-19 Host Genetics Initiative” project using criteria such as genome-wide association, LD, and weak IVs. We used the IVW, weighted median, and weighted mode methods to summarize the MR estimates based on Wald ratio. We found that patients with COVID-19 have a higher risk of VHD. Additionally, the MR-Egger method did not detect pleiotropy of IVs, and the Cochran’s Q statistic did not detect heterogeneity. Sensitivity analyses based on the leave-one-out method were also performed, and the symmetrical distribution of SNPs in the funnel plot and the regression results in the scatter plot suggested the robustness and validity of the study results. Overall, all analyses suggest a causal effect of COVID-19 on VHD.
COVID-19 is a global infectious disease that has had a significant adverse impact on public health and healthcare systems(21). Most researches on COVID-19 complications have focused on the respiratory and nervous systems, encompassing pneumonia, acute respiratory distress syndrome, coagulation disorders, and so on(22, 23). However, comparatively limited research has been conducted on the cardiovascular complications of COVID-19, especially on VHD. However, in clinical practice, there have been observations of COVID-19 patients who have developed unexplained VHD, such as significant mitral or tricuspid valve regurgitation, necessitating valvuloplasty or replacement to restore normal cardiac hemodynamics. Similar cases have been documented by other scholars as well. Consequently, we sought to employ MR to derive an unbiased conclusion that COVID-19 can lead to VHD. Our findings can supplement the understanding of the impact of COVID-19 on the cardiovascular system, providing a novel perspective for investigating the impact of COVID-19 on the cardiovascular system, particularly in patients with VHD. The causal relationship we determined to exist between COVID-19 and valvular heart disease has, to a certain extent, broadened pulmonologists’ awareness of cardiovascular complications in COVID-19 patients and alerted cardiologists to the importance of being attentive to COVID-19 infection. A previous history of COVID-19 infection may not alter the treatment plan for patients with VHD. However, early detection of cardiovascular function is advantageous in identifying changes in valve function early on and potentially improving patient prognosis(24).The mechanism by which COVID-19 causes valvular heart disease (VHD) is currently unclear. Heart valves are composed of dense connective tissue, endothelial cells, and valvular interstitial cells(25). The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) recognizes and attacks endothelial cells by binding to angiotensin-converting enzyme 2 (ACE2) that is expressed on the surface of these cells(26). This leads to large recruitment of immune cells which further attack the endothelial cells directly. The resulting endothelial cell dysfunction and inflammation due to SARS-CoV-2 infection can contribute to abnormal coagulation and actively participate in thrombo-inflammatory processes(27). These findings may support our research results. Ayesha Khanduri reported a case of severe acute mitral regurgitation in a middle-aged male COVID-19-positive patient who underwent emergency mitral valve replacement(28). Pathological evaluation confirmed that the valve damage was secondary to COVID-19 infection. This case highlights the possibility of direct damage to the mitral valve by SARS-CoV-2, resulting in severe mitral valve dysfunction. Yasser A Kamal described a 29-year-old female patient who developed moderate to severe tricuspid regurgitation during COVID-19 infection, with no previous causes of thrombosis(7). P Ryan Tacon reported a case of transcatheter aortic valve replacement (TAVR) following COVID-19 infection(6). A 90-year-old female with atrial fibrillation who underwent initial TAVR developed severe aortic regurgitation and valve thrombosis after COVID-19 infection. The patient’s valve function was restored after receiving valve-in-valve TAVR. Therefore, COVID-19-induced valvular dysfunction, and possibly VHD, may be promoted by attacking endothelial cells and promoting thrombosis. However, this requires further research to confirm. In future studies, whether COVID-19 induces acute or long-term VHD also deserves attention, which may have clinical significance.
This study has some limitations. Firstly, the GWAS data used in this study were derived from a pooled dataset(29). VHD can be classified by anatomical location, functional impairment, etiology, and severity, but this study could not conduct a more detailed analysis of subtypes of VHD or perform gender-specific analyses(30, 31). The dataset included all patients diagnosed with VHD. In addition, the dataset only included individuals of European descent. Therefore, future GWAS and MR studies can focus on specific subtypes of VHD, such as rheumatic or degenerative VHD, or include patients from different regions. Secondly, although the MR method used in this study has a certain level of evidence in evidence-based medicine, there is currently a lack of results from prior observational studies to support it. Therefore, more research is needed to validate the conclusions drawn from this study, such as large-scale randomized controlled trials with the highest level of evidence or in vitro experiments and basic research to explore the pathophysiological mechanisms of VHD caused by COVID-19, which could guide clinical diagnosis, treatment, and optimize clinical trial designs(32).
This study has several innovative aspects. Firstly, the MR study eliminates biases caused by external environmental factors, other confounding factors, and directionality issues, and uses SNPs as IVs for causal inference(33). Secondly, this study systematically analyzed the association between COVID-19 and VHD, and concluded that COVID-19 infection increases the risk of developing VHD for the first time, which has certain clinical implications and promotes more attention to cardiovascular complications caused by COVID-19. In addition, the results of the study remind us of the importance of the primary prevention of VHD. The reliability of causal association evidence derived from MR studies lies between that of observational epidemiological studies and experimental epidemiological studies. In other words, the evidence level provided by MR studies is higher than that provided by cohort studies and only slightly lower than that provided by RCTs. However, the high cost of RCTs makes them less feasible, and MR can provide strong evidence when RCTs cannot be performed, for example because of ethical considerations.
In future studies, it will be important to investigate the mechanism underlying the causal relationship established between COVID-19 and VHD and to explore potential therapeutic strategies aimed at reducing the risks of VHD. In causal inference, we are interested not only in the extent to which an exposure affects an outcome, but also in the mechanisms or pathways through which the exposure influences the outcome. The causal effect of COVID-19 on VHD is the total effect of the exposure on the outcome. The total effect can be decomposed into two parts: a direct effect of the exposure on the outcome and an indirect effect, where the exposure affects the outcome only through the mediator included in the model. Therefore, a mediation MR analysis can be performed to determine the causal pathways through which COVID-19 affects VHD and their relative importance. Some researchers have noted an association between pre-existing cardiovascular diseases and a higher risk of COVID-19 infection. Therefore, conducting a MR study on the causal effect of VHD on COVID-19 infection would be of significance. In addition, studies should be conducted in diverse populations and gender to evaluate the generalizability of the present findings.
MR analysis in this study found a causal effect of COVID-19 infection on the occurrence of VHD, indicating that patients with COVID-19 had a higher risk of VHD.
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.
Author Contributions
JR and ZW conceived and designed the study. JR and YY organized the collection and assembly of the data. JR and PW performed the statistical analysis. JR and YY wrote the first draft of the manuscript. All authors contributed to manuscript revision, and read and approved the submitted version.
Funding
No specific funding was obtained for this study.
Acknowledgments
This manuscript has been released as a pre-print at Research Square (doi:).
Data Availability Statement
Publicly available GWAS datasets (ID: ebi-a-GCST011073 and finn-b-I9_VHD) were accessed from https://gwas.mrcieu.ac.uk/.