Causal Effects of COVID-19 on the Risk of Thrombosis: A Two-Sample Mendel Randomization Study

Abstract

Background Coronavirus disease 2019 (COVID-19) and thrombosis are linked, but the biomolecular mechanism is unclear. We aimed to investigate the causal relationship between COVID-19 and thrombotic biomarkers.

Methods We used two-sample Mendelian randomization (MR) to assess the effect of COVID-19 on 20 thrombotic biomarkers. We estimated causality using inverse variance weighting with multiplicative random effect, and performed sensitivity analysis using weighted median, MR-Egger regression and MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) methods. All the results were examined by false discovery rate (FDR) with the Benjamin and Hochberg method for this correction to minimize false positives. We used R language for the analysis.

Results All COVID-19 classes showed lower levels of tissue factor pathway inhibitor (TFPI) and interleukin-1 receptor type 1 (IL-1R1). COVID-19 significantly reduced TFPI (odds ratio [OR] = 0.639, 95% confidence interval [CI]: 0.435–0.938) and IL-1R1 (OR = 0.603, 95% CI = 0.417–0.872), nearly doubling the odds. We also found that COVID-19 lowered multiple coagulation factor deficiency protein 2 and increased C–C motif chemokine 3. Hospitalized COVID-19 cases had less plasminogen activator, tissue type (tPA) and P-selectin glycoprotein ligand 1 (PSGL-1), while severe cases had higher mean platelet volume (MPV) and lower platelet count. These changes in TFPI, tPA, IL-1R1, MPV, and platelet count suggested a higher risk of thrombosis. Decreased PSGL-1 indicated a lower risk of thrombosis.

Conclusion TFPI, IL-1R, and seven other indicators provide causal clues of the pathogenesis of COVID-19 and thrombosis. This study demonstrated that COVID-19 causally influences thrombosis at the biomolecular level.

Keywords

COVID-19 - Mendelian randomization - thrombosis - TFPI - IL-1R

Introduction

Coronavirus disease 2019 (COVID-19) has posed a major threat to public health and lives.[1] However, with the increased coverage of the COVID-19 vaccine, the focus of the COVID-19 work has shifted from reducing mortality to managing complications,[2] which are essential to lower the risk of infection. Due to the diverse pathogenic mechanisms of COVID-19, patients may experience various complications, including but not limited to thrombosis, sepsis, acute respiratory distress syndrome (ARDS), kidney injury, and respiratory failure.[2] [3] [4] Among these, thrombotic complications are especially prominent and prevalent in COVID-19 deaths and infections: venous thromboembolism occurs in 4.8 to 46% of hospitalized patients with COVID-19,[5] with the highest incidence in critically ill patients admitted to intensive care units.[6] Thrombotic complications have both short- and long-term effects on various systems throughout the body, and their onset is often insidious, requiring attention and further investigation.

Recent retrospective studies have shown that patients with COVID-19 have a high rate of thrombosis, resulting in a higher risk of thrombotic events in severe cases.[7] [8] Current evidence suggests that COVID-19 is associated with thrombocytopenia.[9] Wibowo et al conducted a meta-analysis showing that people with COVID-19 had elevated von Willebrand factor (vWF) levels.[10] Retrospective studies have identified common thrombotic biomarkers in COVID-19 patients, including interleukin (IL)-6, vWF, and D-dimer, which are helping to unravel the hidden mechanisms of COVID-19-associated thrombosis.[11] Deeper research requires understanding the causes of thrombosis at the molecular level, but there is a lack of comprehensive clinical and basic research on the relationship between other thrombotic markers and COVID-19. However, many other biomarkers associated with thrombosis suggest that a study could explore a deeper and novel molecular relationship between COVID-19 and thrombosis. Some studies have shown the beneficial role of tissue factor pathway inhibitor (TFPI) in reducing arterial thrombosis and the potential of P-selectin glycoprotein ligand 1 (PSGL-1) levels to increase adhesion at the site of thrombosis.[12] [13] A Mendelian randomization (MR) study found an association between genetic variation in IL-1 receptor type 1 (IL-1R1) levels and severity of COVID-19.[14] Plasminogen activator tissue type (tPA) is used for treating COVID-19-induced ARDS and the associated potential physiological toxic effects,[15] suggesting that it may reduce the risk of thrombosis. In addition, low levels of the C–C motif chemokine 3 (CCL3) are associated with a high risk of venous thromboembolism.[16] There are previous MR studies showing that Mean platelet volume (MPV), platelet count, and mean platelet width all correlate with thrombosis.[17] But there is a lack of MR studies of thrombosis associated with common indicators related to different coagulation factors. Therefore, we plan to comprehensively select 20 biomarkers involved in different mechanisms of thrombosis for COVID-19 to fill the gap in this area of research. Randomised control trials (RCTs) are a credible way to test if a risk factor causes a certain outcome,[18] but they are not always practicable because they can be very costly, lengthy, and sometimes unethical. When an RCT is not an option to study the effect of a risk factor on an outcome, using genetic variable methods might be a good substitute.

MR uses genetic instrumental variables (IVs) to determine the genetic association between exposures and outcomes, thereby excluding potential confounders. MR studies have been conducted using summary statistics from genome-wide association studies (GWASs) in the European population and have replicated results in individuals of East Asian and other descent.[19] In this study, we established the causal relationship from COVID-19 to thromboses at the molecular level and demonstrated that thrombotic complications in COVID-19 are mediated by multiple etiological factors.[20] It aimed to provide new perspectives and insights into the diagnosis, management, and prevention of COVID-19 complications. This study followed the STROBE-MR criteria for transparent reporting of MR studies.

Method

Study Overview

We conducted a two-sample MR study using genetic variation to assess whether COVID-19, in-hospital COVID-19, and severe COVID-19 (sCOVID-19) are risk factors for thrombotic biomarker formation. The current study used IV analysis, in which single nucleotide polymorphisms (SNPs) systematically identified by GWASs were used as IVs to represent exposures, allowing the observational study to mimic the design of RCTs. Based on the STROBE-MR report, we outline the analysis flowchart ([Fig. 1]).

Fig. 1 Process of MR. CCL3, C–C motif chemokine 3; FIII, FVII, FVIII, FX, FXI, coagulation factors; IL-1R1, interleukin-1 receptor type 1 levels; IL-1R2, interleukin-1 receptor type 2 levels; MCFD2, multiple coagulation factor deficiency protein 2 levels; MPV, mean platelet volume; PAI-1, plasminogen activator inhibitor 1; PSGL-1, P-selectin glycoprotein ligand 1 levels; TFPI, tissue factor pathway inhibitor; tPA, plasminogen activator, tissue type; VWF, von Willebrand factor.

Data for Genetic Epidemiology of COVID-19

For this analysis, we used GWAS data from the COVID-19 Host Genetics Initiative (HGI; RELEASE 5) with 1054664-1877672 individuals, including COVID-19, hospitalized COVID-19, severe-a COVID-19, and severe-b COVID-19 ([Supplementary Table S1], available in the online version). We performed MR analyses of COVID-19 and thrombotic biomarkers using data from the COVID-19 HGI. We obtained data from a GWAS of reported SARS-COV-2 infection (COVID-19 case) in 38,984 European participants from 37 studies and from a GWAS meta-analysis of hospitalized COVID-19 (hCOVID-19) in 9,986 European participants from 22 cohorts. We also used data from two GWAS meta-analyses of sCOVID-19 in critically ill patients, conducted by the HGI in 5,101 (severe-a) and 4,792 (severe-b) European participants from 15 and 14 studies, respectively. Their definition of severe case is the same. In addition, other information is similar, which is aimed to avoid analytical errors associated with a single sample as severe cases are the focus of the analyses and have to be treated with more care. We define each COVID-19 level as follows: (1) sCOVID-19 case (critically ill case): laboratory-confirmed SARS-CoV-2 infection AND hospitalized for COVID-19 AND (death OR ventilator support). (2) Hospitalized COVID-19 case: laboratory-confirmed SARS-CoV-2 infection AND hospitalized for COVD-19. (3) COVID-19 case (reported case of SARS-CoV-2 infection): laboratory-confirmed SARS-CoV-2 infection OR EHR/ICD coding/physician-confirmed COVID-19 OR self-reported COVID-19 by questionnaire. COVID-19 data were obtained from the COVID-19 HGI.[21] Due to the absence of complete clinical information, we do not know the complete incidence of thrombosis. In addition, it is rather unfortunate that the COVID-19 data did not include information on anticoagulation therapy at the time of enrolment. However, the lack of these data does not fundamentally affect the accuracy of our analysis as we use genetic instruments to perform this IV analysis. All study populations were European. Gender was not reported in the cohort.

Data for Genetic Epidemiology of Thrombotic Biomarker

For outcomes, 20 thrombotic biomarkers were included: 350,470 participants in platelet distribution width, 166,066 participants in platelet count, 1,313 participants in multiple coagulation factor deficiency protein 2 (MCFD2) levels, 21,758 participants in thrombomodulin levels, 10,737 participants in MPV, 982 participants in CCL3, 204,402 participants in C-reactive protein level, 3,394 participants in tPA, 21,758 participants in PSGL-1, 1,323 participants for IL-1R1, 1,323 participants for IL-1 receptor type 2 (IL-1R2), 1,323 participants for TFPI, 3,394 participants for coagulation factor III (FIII), coagulation factor VII (FVII), 3,301 participants in coagulation factor VIII (FVIII), 3,301 participants in coagulation factor X (FX), coagulation factor XI (FXI), 1,323 participants in vWF, 1,323 participants in plasminogen activator inhibitor 1 (PAI-1), 3,301 participants in D-dimer ([Supplementary Table S2], available in the online version). First, we prioritize large-sample databases for outcomes with huge sample size differences when choosing databases. Second, the outcome databases which are similar in sample size would be used for initial analyses and the selection of the one with more SNPs for higher confidence. We also consider heterogeneity and horizontal pleiotropy to choose the most suitable databases, not just following these rules. More relevant information is available in [Supplementary Table S2] (available in the online version).

Selection of Instruments

The genetic variants used in this MR study meet three key assumptions: (1) the SNPs under investigation must exhibit a significant degree of correlation with COVID-19. (2) The observed SNPs must demonstrate a lack of association with any potential confounding variables, such as horizontal pleiotropy, that may be linked to COVID-19 thrombotic biomarkers. (3) The effect of genetic variation on outcomes is specific to COVID-19 and is not mediated by other causal pathways. When these assumptions are met, MR provides a valid causal estimate of the effect of the exposure on the outcome.[22] The assumptions underlying MR design are depicted in [Fig. 1]. First, we selected genetic variants with a genome-wide significant p-value (p < 5 × 10⁸) and an F-statistic greater than 10,[23] which satisfies the first assumption, whose violation might introduce the so-called “weak instruments bias”.[24] Second, to mitigate issues with linkage disequilibrium, we will adjust for an r ² of 0.001 and a window size of 10,000 kb. This would avoid the violation of the second or the third IV assumption (similar to violations due to pleiotropy). Concerns about population stratification can be mitigated by restricting the study population to those with the same ethnic background (European).

MR Analysis

MR studies were conducted using summary statistics of GWASs in the European population and replicated the results in individuals of East Asian descent and other ones. On the harmonized sets of genetic instruments, we estimated the causal effects of exposures on outcomes using two-sample MR methods. Interference from confounding factors and reverse causation may disturb traditional epidemiological findings. MR uses genetic IVs to determine the genetic association between exposures and outcomes, thereby excluding potential confounders from interfering. The primary method used was the inverse variance weighted (IVW) method, which provides the greatest statistical power. Weighted median and MR-Egger methods were used for secondary analyses. p-Values below 0.05 were considered statistically consistent. Three key principles guided the selection of the MR method. First, in cases where there was neither heterogeneity nor pleiotropy, IVW estimates were chosen as the preferred option. Second, in cases where there was heterogeneity but no pleiotropy, the weighted median approach was the preferred method of analysis. In addition, when multiple homogeneous variants were identified, the MR-Egger method was favoured.[25]

Sensitivity Analyses

First, to assess heterogeneity, Cochran's Q-test was performed for both the IVW method and the MR-Egger method (p < 0.05 is positive). In addition, we used the false discovery rate (FDR) with Benjamin and Hochberg method for this calculation to the correction to minimize false positives[26] (p < 0.05 is positive). Second, MR-Egger regression and MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) analysis were used to assess the signs of cross-sectional pleiotropy.[27] We can observe the directional horizontal polymorphism by looking at the cut-off distance of the MR-Egger regression. If it does not fall at 0, there is horizontal pleiotropy. MR-PRESSO was used for two purposes: (1) the global test is used to detect the presence of horizontal pleiotropy. (2) To mitigate the effects of horizontal polymorphism, the outlier test was applied, where potential SNPs are progressively eliminated based on a p-value threshold of 0.05.[27] Third, to see whether the association of a genetic variant with the outcome is attenuated after adjustment for exposure, we performed a leave-one-out analysis, which progressively eliminates each SNP, calculates the meta-effects of the remaining SNPs, and observes whether the results change after eliminating each SNP. The effectiveness and reliability of the genetic IV in representing COVID-19 can be confirmed if the F statistic is greater than 10, as specified by the hypotheses. The F statistic indicates the intensity of the IVs on COVID-19, while R2 was used to represent the variance of COVID-19 explained by the IV.

Statistical Analysis

All statistical analyses were performed with the TwoSampleMR package in R (version 0.5.6).

Result

A Decrease in TFPI Levels Is Associated with COVID-19

Our results show significant associations between COVID-19 and a decrease in TFPI (odds ratio [OR] = 0.639, 95% confidence interval [CI] = 0.435–0.938, FDR.P = 0.029), as well as between hospitalized COVID-19 (OR = 0. 800, 95% CI = 0.647–0.990, FDR.P = 0.039), severe-a COVID-19 (OR = 0.855, 95% CI = 0.751–0.974, p = 0.029) and severe-b COVID-19 (OR = 0.848, 95% CI = 0.739–0.973, p = 0.029) and TFPI. The number of genetic tool variables extracted from COVID-19 varied from 5 to 8 ([Supplementary Tables S4–S7], available in the online version). In addition. we found no pleiotropy or heterogeneity ([Fig. 2] and [Table 1]). We use the F-statistic to assess weak IV bias, while all F-statistics are greater than 10.

Fig. 2 Forest plot for total causal effects of COVID-19 on TFPI. The size of the black dot represents the size of the SE. CI, confidence interval; IVW, inverse-variance weighted; OR, odds ratio.

Table 1 F and MR-PRESSO results of COVID-19 with TFPI

	F	MR-PRESSO
COVID-19	55.42 (32.49–107.43)	0.904
Hospitalized COVID-19	87.57 (34.43–240.13)	0.760
Severe-a COVID-19	67.14 (35.61–197.98)	0.982
Severe-b COVID-19	73.24 (30.65–202.73)	0.971

Abbreviations: F, F-statistics; MR-PRESSO, MR-PRESSO global test.

Decrease in IL-1R1 and Interleukin-1 Receptor Type 2 Levels after COVID-19

In the MR analysis of the causal effects of different COVID-19 on the IL-1 receptor outcomes, we found that COVID-19 showed a causal effect on IL-1R1 (OR = 0.603, 95% CI = 0.417–0.872, FDR.P = 0.024), hospitalized COVID-19 showed a causal effect on IL-1R1 (OR = 0. 765, 95% CI = 0.621–0.942, FDR.P = 0.024), severe-a COVID-19 showed a causal effect on IL-1R1 (OR = 0.867, 95% CI = 0.757–0.993, FDR.P = 0.039), and severe-b COVID-19 showed a causal effect on IL-1R1 (OR = 0.870, 95% CI = 0.761–0.993, FDR.P = 0.039) ([Fig. 3]). [Table 2] shows the F-statistic and MR-PRESSO global tests (F > 10). We cannot find any significant interception in the MR-Egger way from the scatter plots, which do not represent pleiotropy ([Fig. 4]). All SNPs are shown in [Supplementary Tables S4] to [S7] (available in the online version).

Fig. 3 Forest plot for total causal effects of COVID-19 on IL-1R1. The size of the black dot represents the size of the SE. CI, confidence interval; IVW, inverse-variance weighted; OR, odds ratio.

Table 2 Effect of COVID-19 to thrombosis biomarkers

Class of COVID-19	Biomarker	Effect
All class	TFPI	–
All class	IL-1R1	–
Severe COVID-19	IL-1R2	–
COVID-19	MCFD2	–
COVID-19	CCL3	+
Hospitalized COVID-19	tPA	–
Hospitalized COVID-19	PSGL-1	–
Severe COVID-19	Platelet count	–
Severe COVID-19	MPV	+

Abbreviations: CCL3, C–C motif chemokine 3; IL-1R1, interleukin-1 receptor type 1 levels; IL-1R2, interleukin-1 receptor type 2 levels; MCFD2, multiple coagulation factor deficiency protein 2 levels; MPV, mean platelet volume; PSGL-1, P-selectin glycoprotein ligand 1 levels; TFPI, tissue factor pathway inhibitor; tPA, plasminogen activator tissue type.

Note: +: increase; −: decrease.

Fig. 4 Scatter plots of COVID-19 with IL-1R1. The X-axis displays the SNP effect and SE on each of COVID-19 IVs, while the Y-axis shows the SNP effect and SE on IL-1R1. The regression line for MR-Egger, weighted median, IVW, simple mode, and weighted mode is presented. (A–D) Results of COVID-19, severe-a COVID-19, severe-b COVID-19, and hospitalized COVID-19 on IL-1R1.

In addition, COVID-19 and hospitalized COVID-19 did not show a causal effect on IL-1R2. While severe-a COVID-19 showed a causal effect on IL-1R2 (OR = 0.863, 95% CI = 0.763–0.978, p = 0.021), severe-b COVID-19 showed a causal effect on IL-1R2 (OR = 0.870, 95% CI = 0.762–0.993, p = 0.039) ([Fig. 5]). However, when we added FDR validation, the p-value changed to 0.078 (>0.05). So, it cannot be considered as true positive results. The F-statistic was greater than 10 and MR-PRESSO global tests showed no lateral polymorphism ([Table 3]). To assess heterogeneity, Cochran's Q test was performed for both the IVW and MR-Egger methods. It showed no heterogeneity in the incidental effect of COVID-19 on IL-1R2 ([Fig. 5]). We cannot find any significant interception in the MR-Egger way from the scatter plots, which do not represent pleiotropy ([Fig. 6]).

Fig. 5 Forest plot for total causal effects of COVID-19 on IL-1R2. The size of the black dot represents the size of the SE. CI, confidence interval; IVW, inverse-variance weighted; OR, odds ratio.

Table 3 F and MR-PRESSO results of COVID-19 with IL-1R1

	F	MR-PRESSO
COVID-19	64.09 (32.82–107.43)	0.623
Hospitalized COVID-19	78.78 (34.43–240.13)	0.559
Severe-a COVID-19	60.91 (34.75–197.98)	0.394
Severe-b COVID-19	64.57 (30.65–202.73)	0.412

Abbreviations: F, F-statistics; MR-PRESSO, MR-PRESSO global test.

Changes in Other Thrombotic Markers following COVID-19

In order to better understand COVID-19-related thrombosis, we performed MR analysis to elucidate the causal effects of different COVID-19 severity on other thrombotic biomarker outcomes. First, in the course of our study, we unexpectedly observed a robust causal association between COVID-19 and MCFD2 levels (OR = 0.676, 95% CI = 0.468–0.977, FDR.P = 0.046). There was also a causal association between COVID-19 and CCL3 (OR = 0.493, 95% CI = 1.033–1.406, FDR.P = 0.046) ([Fig. 7A]). However, genetically predicted hospitalized COVID-19 showed a strong association with tPA (OR = 0.959, 95% CI = 0.924–0.995, FDR.P = 0.017) and PSGL-1 (OR = 0.902, 95% CI = 0.834–0.976, FDR.P = 0.017) ([Fig. 7B]). Furthermore, our test showed that severe-a COVID-19 showed a statistically significant association with platelet count (OR = 0.978, 95% CI = 0.962–0.994, FDR.P = 0.007), while severe-b COVID-19 showed a causal relationship with MPV (OR = 1.084, 95% CI = 1.006–1.168, FDR.P = 0.034) and platelet count (OR = 0.970, 95% CI = 0.943–0.997, FDR.P = 0.034) ([Fig. 7C]). The F-statistics were all greater than 10, while MR-PRESSO global tests showed that the causal relationship between COVID-19 and platelet count had horizontal pleiotropy. We then performed an outlier test, which showed a significant improvement compared to the original result (the p-value of the MR-PRESSO bias test is less than 0.05) ([Table 4]).

Fig. 6 Scatter plots of COVID-19 with IL-1R2. The X-axis shows the SNP effect and SE on each of COVID-19 IVs. The Y-axis shows the SNP effect and SE on IL-1R2. (A, B) MR-Egger results of severe-a COVID-19 and severe-b COVID-19 on IL-1R1. The regression line for MR-Egger, weighted median, IVW, simple mode, and weighted mode is shown. SNP, single nucleotide polymorphism.

Fig. 7 Forest plot for total causal effects of COVID-19 on other thrombotic biomarkers. (A) MR results of common infection COVID-19 on thrombosis, (B) MR results of hospitalized COVID-19 on thrombosis, and (C) MR results of severe COVID-19 on thrombosis. The size of the black dot represents the size of the SE. CI, confidence interval; IVW, inverse-variance weighted; MR, Mendelian randomization; OR, odds ratio.

Table 4 F and MR-PRESSO results of COVID-19 with IL-1R2

	F	MR-PRESSO
Severe-a COVID-19	60.91 (34.75–197.98)	0.754
Severe-b COVID-19	64.57 (30.65–202.73)	0.507

Abbreviations: F, F-statistics; MR-PRESSO, MR-PRESSO global test.

Table 5 F and MR-PRESSO results of COVID-19 with other biomarkers

	F	MR-PRESSO
		Global test	Outlier test
COVID-19
MCFD2	64.09 (32.82–107.43)	0.693
CCL3	64.09 (32.82–107.43)	0.092
Hospitalized COVID-19
tPA	82.98 (34.43–240.13)	0.156
PSGL-1	78.78 (34.43–240.13)	0.202
Severe-a COVID-19
Platelet count	60.91 (34.75–197.98)	0.081
Severe-b COVID-19
MPV	64.57 (30.65–202.73)	0.261
Platelet count	64.57 (30.65–202.73)	<0.0003	0.003

Abbreviations: F, F-statistics; MR-PRESSO, MR-PRESSO global test.

Overall, in IVW and sensitivity analysis, COVID-19 correlates with a decrease in levels of MCFD2 and CCL3 ([Fig. 7A]). Hospitalized COVID-19 is correlated with a decrease in tPA and PSGL-1 ([Fig. 7B]). In addition, severe-a COVID-19 showed a notable association with a decrease in platelet count levels and severe-b COVID-19 showed an apparent causal effect leading to a decrease in platelet count levels and an increase in MPV ([Fig. 7C]). Sensitivity analysis of the association showed that the results were indeed reliable. Literature analysis shows that the above causal relationship leads to a further increased risk of thrombosis, indicating that COVID-19 may easily lead to thrombosis, which has some guiding significance for clinical management ([Table 2]).

Negative Results

Our analysis of FX, FVIII, FXI, FIII, and FVII did not show positive results, and there are two possible reasons for this. First, during the coagulation process, the consumption rate of coagulation factors may exceed their production rate in the liver, resulting in minimal fluctuations in coagulation factor levels. Second, clotting factors primarily aggregate to form clots rather than increasing in quantity to cause clot formation. Similarly, COVID-19 has not shown a strong causal relationship with coagulation regulators such as thrombomodulin, PAI-1, vWF levels, and D-dimer ([Supplementary Table S3], available in the online version).

Sensitive Analysis

Extensive sensitivity analyses were performed to validate the causal relationship and robustness of all results. Heterogeneity was assessed using IVW and MR-Egger tests (Cochran's Q test), with a p-value of <0.05 indicating heterogeneity in the studies. The MR-PRESSO R software package was used to assess whether there were discrepancies between the MR analysis results before and after correction. For the associations of COVID-19 and platelet distribution width, COVID-19 and C-reactive protein, heterogeneity was observed and the MR-PRESSO global test showed horizontal pleiotropy. In this case, the ME-Egger method should be preferred to observe causality. Both platelet distribution width and C-reactive protein showed a negative result in the ME-Egger method. We used the MR-PRESSO bias test to remove the outliers (rs2271616, rs10936744, and rs35508621) in the associations of COVID-19 and platelet distribution width, but no significant improvement was detected. Outliers (rs35508621, rs643434, and rs757405) in the association between COVID-19 and C-reactive protein were also removed in the same way. There was no significant improvement in the chance effect compared to before outlier removal. In the association between hospitalized COVID-19 and tPA, heterogeneity was observed and the MR-PRESSO global test showed no horizontal pleiotropy. In this case, the weighted median method should be preferred to observe causality. The weighted median method showed a positive effect. We used the MR-PRESSO bias test to remove the outlier (rs74956615) in the associations of severe-b COVID-19 and platelet count ([Table 5]), with a significant improvement detected (p = 0.00233). Other associations showed neither heterogeneity nor horizontal pleiotropy. The leave-one-out results suggest that the entire meta-effect was not driven by a single IV. The results of each SNP was available in [Supplementary Table S8] (available in the online version).

Discussion

This study showed that COVID-19 induces thrombogenesis through a multi-factorial mechanism rather than a single factor. It is primarily characterized by a decrease in TFPI and IL-1-1 receptor levels, accompanied by a reduced trend of tPA and PSGL-1. The validity of the decrease in platelet count and increase in MPV after COVID-19 was then verified. However, it was found that MCFD2 was also reduced after COVID-19 and, together with the reduction of PSGL-1, their effect on thrombosis remains incompletely elucidated and they may be indirect factors in the formation of thrombosis.

Main COVID-19-Mediated Thrombotic Mechanisms

TFPI is a potent inhibitor of TF-mediated coagulation and thrombin.[28] It has been observed that monocytes generally release pulmonary TF, which is upregulated in SARS-CoV-2 disease.[29] [30] According to Hackeng et al's proposal, the inhibitory effect of the proteins on tissue factor activity is achieved by facilitating the interaction between full-length TFPI and FXa.[31]

A recent MR study found that IL-1R1 is a risk factor for the severity of respiratory COVID-19.[14] During the COVID-19 epidemic, anakinra is commonly used as an immunotherapy to suppress IL-1R.[32] However, many studies indicate that the effect of anakinra is not as expected.[33] [34] IL-1 is an abbreviation for two cytokines: IL-1 alpha and IL-1 beta.[35] They interact with their co-receptor, which consists of IL-1R1 and a helper protein (IL-1Racp).[36] The complexes then trigger a violent inflammatory response. IL-1R2 is distinct from other components of the IL-1 family, not only because it does not have a TIR structural domain, but also because it has a unique mode of action in down-regulating IL-1 activity. Specifically, IL-1R2 binds to IL-1 alpha and IL-1 beta with a strong attraction, functioning as a molecular trap to inhibit their activity and block signal transduction.[37] Under physiological and pathological conditions, the up-regulated expression of IL-1R2 is deemed to reflect the activation of negative feedback loop regulating cytokine activity or the response to anti-inflammatory therapy, whereas a deficit in IL-1R2 expression is related to the etiopathogenesis of disease, which is easy to form an inflammatory thrombus.[38]

The defect in IL-1R2 expression mainly, which occurred mainly in sCOVID-19, is consistent with venous thrombosis in critically ill COVID-19 patients. However, the result became negative (FDR.P = 0.078) after FDR correction, which is closed to 0.05 that does not exclude the possibility of a true negative. The decreased trend of IL-1R1 (FDR.P = 0.039/0.024) in COVID-19 infection suggests that the combination of anakinra and IL-1R would decrease and result in a poor effect of immunotherapy. Therefore, it provides new ideas and insights for the development of new drugs targeting the IL-1R. We need further studies to validate these findings in order to improve the efficacy of immunotherapy drugs.

Mechanisms Relative to Other Thrombotic Biomarkers

tPA can activate plasminogen to mediate the formation of plasmin to facilitate thrombolysis. It has also been shown to play a critical role in facilitating cell migration within the extracellular matrix, while acting as a critical mediator of cellular signaling pathways.[39] [40] [41] Judicious administration of tPA may bolster the recovery of patients with ARDS, subsequently diminishing mortality associated with COVID-19.[42] [43] The causal relationship between hospitalized COVID-19 and tPA highlights the increased risk of thrombotic events that these patients may face.

Research has indicated that PSGL-1 blocks the replication of coronavirus by damaging the SARS-CoV-2 spike glycoprotein and intercepting pseudoviral adhesion.[44] But it is not clear whether the coronavirus has a mechanism to antagonize PSGL-1. From this perspective, our finding shows that coronaviruses are antagonistic to PSGL-1. It may suggest that COVID-19 and PSGL-1 are antagonistic to each other, which may be seen as a way for the body to self-regulate.

Elevated MPV associated with a decrease in platelet count was found to be a result of COVID-19 infection in our study. It was correlated with another study in 2022.[17]

Meaning and Limitation of the Study

Our findings have important clinical implications. First, using two-sample MR analysis, we found that nine thrombotic indicators have a causal relationship with COVID-19, five of which are novel biomarkers and seven of which suggest that COVID-19 leads to thrombosis formation. Second, our study found that COVID-19 infection may be associated with a decrease in IL-1R2, which may mediate the inflammatory response to induce thrombosis. Third, MR analysis revealed that COVID-19 infection leads to a decrease in IL-1R1, which may explain why the effect of immuno-antagonist therapy was not as expected.

We acknowledge some limitations in our research. First, the relatively small number of variables in our tool may affect the generalizability of our analysis results. Second, the timing of sampling is unknown, although it was mainly divided as at discharge, on admission, and hospitalization, the accurate proportion is unknown as it was not open access. However, this will not influence the analysis of our study. Third, we lack data on thrombosis screening as it was not open access. In addition, conventional markers of COVID thrombosis risk were not the main biomarkers in our study as some observational studies have been done on it and some of the mechanisms have been pointed out.[11] Our aim is to explore the mechanisms of other markers with COVID-19 besides conventional molecular markers. It is possible that this will differ from conventional studies. It is regrettable that we could not establish a causal relationship between D-dimer and COVID-19, which may be due to the incompleteness of the database content. Moreover, some biases and confounding factors may influence the validity and accuracy of our analysis in real-world settings. Lastly, the study population is all European and may not be representative of other populations. Future analysis should explore the molecular mechanisms behind the thrombosis formation and provide better suggestions for therapy methods.

Conclusion

We used a two-sample analysis and found that COVID-19 showed a causal relationship with nine thrombotic indicators, represented by a reduction in TFPI directly suggesting that COVID-19 leads to thrombosis. Our study showed that COVID-19 infection is associated with a decrease in IL-1R1, which may explain why the effect of immuno-antagonist therapy was not as expected. In addition, COVID-19 may also be associated with a decrease in IL-1R2, which will mediate the inflammatory response to induce thrombosis.

Conflict of Interest

None declared.

Acknowledgment

We thank all the authors for their contributions to this manuscript.

Data Availability Statement

The data generated and codes used in the current study are available in this published article and [Supplementary File] (available in the online version) associated with it. The data underlying this article are available in IEU OpenGWAS project, at https://gwas.mrcieu.ac.uk/.

Ethical Approval Statement

The data for this study came from a public database, so no such approval was required.

Authors' Contribution

M.Z. collected and organized the data. Z.L. analyzed the data and finished the writing. All authors participated in the writing and review of the article.

^* These authors contributed equally to the research.

• References

• 1 Hu B, Guo H, Zhou P, Shi ZL. Characteristics of SARS-CoV-2 and COVID-19. Nat Rev Microbiol 2021; 19 (03) 141-154
• 2 Suthar AB, Wang J, Seffren V, Wiegand RE, Griffing S, Zell E. Public health impact of covid-19 vaccines in the US: observational study. BMJ 2022; 377: e069317
• 3 Wu Z, McGoogan JM. Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention. JAMA 2020; 323 (13) 1239-1242
• 4 Zhou F, Yu T, Du R. et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet 2020; 395 (10229): 1054-1062
• 5 Pellicori P, Doolub G, Wong CM. et al. COVID-19 and its cardiovascular effects: a systematic review of prevalence studies. Cochrane Database Syst Rev 2021; 3 (03) CD013879
• 6 Kwee RM, Adams HJA, Kwee TC. Pulmonary embolism in patients with COVID-19 and value of D-dimer assessment: a meta-analysis. Eur Radiol 2021; 31 (11) 8168-8186
• 7 Vincent JL, Levi M, Hunt BJ. Prevention and management of thrombosis in hospitalised patients with COVID-19 pneumonia. Lancet Respir Med 2022; 10 (02) 214-220
• 8 Antic D, Milic N, Chatzikonstantinou T. et al. Thrombotic and bleeding complications in patients with chronic lymphocytic leukemia and severe COVID-19: a study of ERIC, the European Research Initiative on CLL. J Hematol Oncol 2022; 15 (01) 116
• 9 Jiang SQ, Huang QF, Xie WM, Lv C, Quan XQ. The association between severe COVID-19 and low platelet count: evidence from 31 observational studies involving 7613 participants. Br J Haematol 2020; 190 (01) e29-e33
• 10 Wibowo A, Pranata R, Lim MA, Akbara MR, Martha JW. Endotheliopathy marked by high von Willebrand factor (vWF) antigen in COVID-19 is associated with poor outcome: a systematic review and meta-analysis. Int J Infect Dis 2022; 117: 267-273
• 11 Gorog DA, Storey RF, Gurbel PA. et al. Current and novel biomarkers of thrombotic risk in COVID-19: a Consensus Statement from the International COVID-19 Thrombosis Biomarkers Colloquium. Nat Rev Cardiol 2022; 19 (07) 475-495
• 12 Winckers K, ten Cate H, Hackeng TM. The role of tissue factor pathway inhibitor in atherosclerosis and arterial thrombosis. Blood Rev 2013; 27 (03) 119-132
• 13 Nayak L, Sweet DR, Thomas A. et al. A targetable pathway in neutrophils mitigates both arterial and venous thrombosis. Sci Transl Med 2022; 14 (660) eabj7465
• 14 Wang R. Genetic variation of interleukin-1 receptor type 1 is associated with severity of COVID-19 disease. J Infect 2022; 84 (02) e19-e21
• 15 Mazilu L, Katsiki N, Nikolouzakis TK. et al. Thrombosis and haemostasis challenges in COVID-19 - therapeutic perspectives of heparin and tissue-type plasminogen activator and potential toxicological reactions-a mini review. Food Chem Toxicol 2021; 148: 111974
• 16 Mir Seyed Nazari P, Marosi C, Moik F. et al. Low systemic levels of chemokine C-C motif ligand 3 (CCL3) are associated with a high risk of venous thromboembolism in patients with glioma. Cancers (Basel) 2019; 11 (12) 14
• 17 Cheung CL, Ho SC, Krishnamoorthy S, Li GH. COVID-19 and platelet traits: a bidirectional Mendelian randomization study. J Med Virol 2022; 94 (10) 4735-4743
• 18 Robbins AJ, Che Bakri NA, Toke-Bjolgerud E. et al. The effect of TRV027 on coagulation in COVID-19: a pilot randomized, placebo-controlled trial. Br J Clin Pharmacol 2023; 89 (04) 1495-1501
• 19 Emdin CA, Khera AV, Kathiresan S. Mendelian randomization. JAMA 2017; 318 (19) 1925-1926
• 20 Zhou Y, Qian X, Liu Z. et al. Coagulation factors and the incidence of COVID-19 severity: Mendelian randomization analyses and supporting evidence. Signal Transduct Target Ther 2021; 6 (01) 222
• 21 COVID-19 Host Genetics Initiative. The COVID-19 Host Genetics Initiative, a global initiative to elucidate the role of host genetic factors in susceptibility and severity of the SARS-CoV-2 virus pandemic. Eur J Hum Genet 2020; 28 (06) 715-718
• 22 Didelez V, Sheehan N. Mendelian randomization as an instrumental variable approach to causal inference. Stat Methods Med Res 2007; 16 (04) 309-330
• 23 Palmer TM, Lawlor DA, Harbord RM. et al. Using multiple genetic variants as instrumental variables for modifiable risk factors. Stat Methods Med Res 2012; 21 (03) 223-242
• 24 Burgess S, Thompson SG. CRP CHD Genetics Collaboration. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 2011; 40 (03) 755-764
• 25 Russell AE, Ford T, Gunnell D. et al. Investigating evidence for a causal association between inflammation and self-harm: a multivariable Mendelian Randomisation study. Brain Behav Immun 2020; 89: 43-50
• 26 Chumbley JR, Friston KJ. False discovery rate revisited: FDR and topological inference using Gaussian random fields. Neuroimage 2009; 44 (01) 62-70
• 27 Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50 (05) 693-698
• 28 Mast AE. Tissue factor pathway inhibitor: multiple anticoagulant activities for a single protein. Arterioscler Thromb Vasc Biol 2016; 36 (01) 9-14
• 29 Francischetti IMB, Toomer K, Zhang Y. et al. Upregulation of pulmonary tissue factor, loss of thrombomodulin and immunothrombosis in SARS-CoV-2 infection. EClinicalMedicine 2021; 39: 101069
• 30 Hottz ED, Azevedo-Quintanilha IG, Palhinha L. et al. Platelet activation and platelet-monocyte aggregate formation trigger tissue factor expression in patients with severe COVID-19. Blood 2020; 136 (11) 1330-1341
• 31 Hackeng TM, Seré KM, Tans G, Rosing J. Protein S stimulates inhibition of the tissue factor pathway by tissue factor pathway inhibitor. Proc Natl Acad Sci U S A 2006; 103 (09) 3106-3111
• 32 Kyriazopoulou E, Poulakou G, Milionis H. et al. Early treatment of COVID-19 with anakinra guided by soluble urokinase plasminogen receptor plasma levels: a double-blind, randomized controlled phase 3 trial. Nat Med 2021; 27 (10) 1752-1760
• 33 CORIMUNO-19 Collaborative group. Effect of anakinra versus usual care in adults in hospital with COVID-19 and mild-to-moderate pneumonia (CORIMUNO-ANA-1): a randomised controlled trial. Lancet Respir Med 2021; 9 (03) 295-304
• 34 Caricchio R, Abbate A, Gordeev I. et al; CAN-COVID Investigators. Effect of canakinumab vs placebo on survival without invasive mechanical ventilation in patients hospitalized with severe COVID-19: a randomized clinical trial. JAMA 2021; 326 (03) 230-239
• 35 Gabay C, Lamacchia C, Palmer G. IL-1 pathways in inflammation and human diseases. Nat Rev Rheumatol 2010; 6 (04) 232-241
• 36 Dinarello CA. The IL-1 family of cytokines and receptors in rheumatic diseases. Nat Rev Rheumatol 2019; 15 (10) 612-632
• 37 Colotta F, Re F, Muzio M. et al. Interleukin-1 type II receptor: a decoy target for IL-1 that is regulated by IL-4. Science 1993; 261 (5120) 472-475
• 38 Garlanda C, Riva F, Bonavita E, Mantovani A. Negative regulatory receptors of the IL-1 family. Semin Immunol 2013; 25 (06) 408-415
• 39 Myöhänen H, Vaheri A. Regulation and interactions in the activation of cell-associated plasminogen. Cell Mol Life Sci 2004; 61 (22) 2840-2858
• 40 Yatsenko T, Skrypnyk M, Troyanovska O. et al. The role of the plasminogen/plasmin system in inflammation of the oral cavity. Cells 2023; 12 (03) 445
• 41 He S, Waheed AA, Hetrick B. et al. PSGL-1 inhibits the virion incorporation of SARS-CoV and SARS-CoV-2 spike glycoproteins and impairs virus attachment and infectivity. bioRxiv. Jul 6 2020
• 42 Choudhury R, Barrett CD, Moore HB. et al. Salvage use of tissue plasminogen activator (tPA) in the setting of acute respiratory distress syndrome (ARDS) due to COVID-19 in the USA: a Markov decision analysis. World J Emerg Surg 2020; 15 (01) 29
• 43 Rashidi F, Barco S, Rezaeifar P. et al. Tissue plasminogen activator for the treatment of adults with critical COVID-19: a pilot randomized clinical trial. Thromb Res 2022; 216: 125-128
• 44 He S, Waheed AA, Hetrick B. et al. PSGL-1 inhibits the incorporation of SARS-CoV and SARS-CoV-2 spike glycoproteins into pseudovirions and impairs pseudovirus attachment and infectivity. Viruses 2020; 13 (01) 46

Publication History

Received: 19 July 2023

Accepted: 17 November 2023

Accepted Manuscript online:
07 February 2024

Article published online:
05 March 2024