Causal Relationship between PECAM-1 Level and Cardiovascular Diseases: A Mendelian Randomization Study

Abstract

Background: Platelet endothelial cell adhesion molecule (PECAM-1) is present in the vascular endothelium and plays important roles in various biological processes. Several recent studies have reported associations between PECAM-1 and certain subtypes of cardiovascular diseases (CVDs). However, further research is necessary to clarify the causal effects of PECAM-1 on CVDs. To determine whether PECAM-1 and CVDs are causally associated, we conducted a two-sample Mendelian randomization (TSMR) study.

Methods: Single nucleotide polymorphisms (SNPs) associated with PECAM-1 were used as instrumental variants (IVs) to estimate the causal effects of PECAM-1 on CVDs. Six SNPs were included in our TSMR study. The inverse-variance weighted (IVW) method was applied in the primary analysis. To confirm the initial results, we conducted several complementary analyses and pleiotropy analyses.

Results: In the IVW analysis, higher genetically predicted PECAM-1 levels were associated with lower risk of coronary artery disease (CAD) (OR, 0.835; CI, 0.757–0.92; P = 3 × 10⁻⁴) and myocardial infarction (MI) (OR, 0.79; CI, 0.709–0.881; P = 2.03 × 10⁻⁵).

Conclusions: The findings confirmed that elevated PECAM-1 levels may decrease the risk of CAD and MI. These results confirm the causal effect of PECAM-1 on CVDs and may facilitate further investigation of the mechanism of PECAM-1 in CVD pathogenesis.

Main article text

Introduction

Cardiovascular diseases (CVDs) are a class of diseases involving the heart and blood vessels. CVDs include a range of conditions such as coronary artery disease, myocardial infarction, heart failure, atrial fibrillation, hypertension, stroke, and peripheral artery diseases [1]. CVDs and their resultant complications are among the leading causes of death worldwide. CVDs pose a substantial challenge to global public health. According to the World Health Organization, CVDs cause approximately 17.9 million deaths annually, thus resulting in a heavy global disease burden [2]. The etiology of CVDs is complex. A variety of risk factors can induce CVDs; these risk factors include high levels of triglycerides (TG) [3], high levels of low-density-lipoprotein cholesterol (LDL-C) [4], low levels of high-density lipoprotein cholesterol (HDL-C) [5], alcohol intake [6], type 2 diabetes (T₂D) [7], smoking [8], body mass index (BMI) [9], and depression [10]. Exploring more effective therapeutic strategies is crucial in light of the substantial mortality rate and financial burden associated with CVDs.

PECAM-1, also known as CD31 [11], is a cell-cell adhesion molecule present on the surfaces of endothelial cells in blood vessels and certain immune cell subsets, such as granulocytes, macrophages, and lymphocytes [12, 13]. Extensive research has investigated the functions of PECAM-1, and revealed its involvement in various cellular pathways and its important roles in many biological processes. For instance, PECAM-1 facilitates leukocyte transmigration through the intercellular junctions of endothelial vascular cells during the inflammatory process [13]. PECAM-1 is critical in the pathogenesis of CVDs such as myocardial infarction (MI) [14, 15]. PECAM-1 also has a major role in platelet signal transduction pathways [16, 17].

Several epidemiological studies and basic medical research studies have reported associations between PECAM-1 and several types of CVD, such as CAD, MI, and stroke [18–22]. One study has indicated that inhibition of PECAM-1 expression alleviates leucocyte migration after stroke [23]. Another clinical trial has found that a co-therapy comprising docosahexaenoic acid nanoencapsulated with anti-PECAM-1 prevents atherosclerosis progression [24]. Moreover, in another randomized clinical trial, PECAM-1 levels have been found to increase 24 hours after stroke [25]. However, those studies have not clarified the causal effects of PECAM-1 levels on CVDs. Previous studies have focused only on certain types of CVD, whereas studies revealing the association between PECAM-1 and other common CVDs, such as AF and hypertension, are lacking. Thus, the causal effect of PECAM-1 levels on CVD risk required further investigation.

Mendelian randomization (MR) is an increasingly applied method using single nucleotide polymorphisms (SNPs) as genetic variants to evaluate the causal effects of risk factors on outcomes [26]. This method was applied to provide new insights into the role of PECAM-1 as a risk factor that prevents or triggers the development of CVDs. The principle of MR analysis involves using genetic variants as instrumental variants (IVs). Genetic variants are randomly allocated at conception, and thus are not influenced by confounding factors and reverse causation [27]. By leveraging naturally occurring genetic variants as instrumental variables, which are associated with the exposure of interest but are independent of confounding factors, MR approaches emulate randomized controlled trials, thereby enabling causal inferences to be drawn. MR can also mitigate confounding issues frequently encountered in observational studies, by exploiting the random allocation of genotypes at conception, thus minimizing the influence of unmeasured or uncontrolled factors, such as lifestyle or environmental variables. We conducted two-sample MR (TSMR) analysis to examine the causal relationship between PECAM-1 levels and CVDs, including CAD, MI, AF, HF, ischemic stroke (IS), cardioembolic stroke (CS), large artery stroke (LAS), and small vessel stroke (SVS). The workflow of our study is shown in Figure 1.

Figure 1

Study Design for TSMR Analysis.

Schematic shows the application of Mendelian randomization to assess the causal relationships between PECAM-1 and cardiovascular diseases (CVDs).

Methods

Study Design

TSMR analysis was performed to evaluate the causal relationships between PECAM-1 and CVDs. SNPs were used as IVs in this study. The IVs were required to meet the following three key assumptions for inclusion in subsequent analysis: 1) the IVs were strongly associated with the exposure, i.e., PECAM-1 (each SNP for PECAM-1 reached genome-wide significance, P < 5 × 10⁻⁸); 2) the IVs were not associated with any confounders or outcomes; and 3) the IVs affected the outcomes only via the exposure [28]. The workflow for the overall study design is shown in Figure 1.

Data Sources for CVDs

Summary genetic data on CVDs were obtained from several published genome-wide association studies (GWAS), including the following: 1) patients with CAD vs. the general population from the CardiogramplusC4D consortium (60,801 cases vs. 123,504 controls) [29]; 2) patients with MI vs. the general population from the CardiogramplusC4D consortium (43,676 cases vs. 128,197 controls) [29]; 3) patients with HF vs. the general population from the HERMES Consortium (47,309 cases vs. 930,014 controls) [30]; 4) patients with AF vs. the general population from HUNT, deCODE, MGI, DiscovEHR, UK Biobank, and AFGen Consortium (65,446 cases vs. 522,744 controls) [31]; 5) Liu et al., for hypertension (146,562 individuals) [32]; and 6) patients with IS vs. the general population from the MEGASTROKE consortium (34,217 cases vs. 404,630 controls) [33]. In reference to the Trial of Org 10172 in Acute Stroke Treatment criteria, in our study, IS was further categorized as LAS, SVS, or CS [34]. We downloaded the genetic data and conducted a quality control step to ensure data quality. We identified metadata errors through a comparison of the data used against external reference data sets (the National Human Genome Research Institute-European Bioinformatics Institute GWAS catalog and 1000 Genomes super populations), and identified other analytical issues through a comparison of reported vs. expected genetic effect sizes. Because our study involved analysis of published genetic data, no ethics approval was required. Details regarding the datasets included in the analysis are provided in Table 1.

Table 1

Detailed Information on GWAS and Datasets Included in the Mendelian Randomization Analysis.

GWAS	Phenotype	Participants	Ancestry	Use in this MR study
Folkersen et al., 2020 [35]	PECAM-1	21,758 individuals	European	Exposure
CARDIoGRAMplusC4D [29]	Coronary artery disease	60,801 cases	Multi-ancestry (77% European)	Outcome
		123,504 controls
	Myocardial infarction	43,676 cases
		128,197 controls
HERMES Consortium [30]	Heart failure	47,309 cases	European	Outcome
		930,014 controls
Nielsen et al., 2018 [31]	Atrial fibrillation	60,620 cases	European	Outcome
		970,216 controls
MEGASTROKE [33]	Any ischemic stroke	34,217 cases	European	Outcome
		404,630 controls
	Large artery stroke	4373 cases
		146,392 controls
	Small vessel stroke	5386 cases
		192,662 controls
	Cardioembolic stroke	7193 cases
		204,570 controls
Liu et al., 2016 [32]	Hypertension	146,562 individuals	Multi-ancestry (82% European)	Outcome

GWAS, genome-wide association study.

Data Sources for PECAM-1 and IV Selection

The summary genetic data for PECAM-1 were derived from a published GWAS involving 21,758 participants of European ancestry [35]. We first used the PLINK clumping algorithm to select SNPs that were associated with PECAM-1 and met the criteria for MR analysis at a genome-wide significance level (P < 5 × 10⁻⁸, window size = 10 Mb). The SNPs meeting the criteria for MR analysis for subsequent analysis were retained. We also performed linkage disequilibrium analysis to identify whether SNPs were in high LD with each other (r² < 0.001). We used LD clumping methods to select independent SNPs not in LD with one another, to help avoid bias arising from use of correlated SNPs. F statistics were calculated to assess the strength of each selected SNP with the following formula:

$$F=\frac{R^2(N\text{–}2)}{(1\text{–}{R}^2)}$$

where R ² represents the percentage variability in PECAM-1 explained by each selected SNP, and N represents the sample size of the GWAS used for selecting the SNPs. An F statistic greater than 10 was considered sufficient to counteract the risk of weak instrument bias [36]. The exclusion criteria were as follows: 1) P value greater than 5 × 10⁻⁸; 2) r² of the LD analysis greater than 0.001; and 3) F statistic smaller than 10.

Statistical Analysis

To perform TSMR analysis, we first extracted the data from the published GWASs, after harmonization of the effect alleles across GWASs, and use of a Wald estimate to estimate the effect of PECAM-1 on CVDs [37]. We then used the Delta method to account for possible measurement errors in the estimation of the causal association between PECAM-1 and CVDs [37, 38].

Fixed-effects inverse variance-weighted (IVW) analysis was the main method used to estimate the final effect. This method is sensitive to SNP pleiotropy, because it is based on the assumption that no directional pleiotropy exists, and it ignores potential pleiotropy of other risk factors. These aspects may affect the results of causal estimates. Therefore, we applied Cochran’s Q test to assess the horizontal pleiotropy in the IVW analysis. A P-value < 0.05 indicated the presence of horizontal pleiotropy. The random-effects IVW method was used if the P-value was less than 0.05 in Cochran’s Q test [39]. The MR-Egger intercept test was also conducted to detect potential directional pleiotropy in the IVW analysis. An intercept P-value < 0.05 indicated the presence of significant pleiotropic bias [39].

In addition to the IVW method, several sensitivity analyses and complementary analyses were applied to ensure the robustness of our results and to decrease the bias due to horizontal pleiotropy. These methods included the weighted median method, the simple median method, the MR-Egger regression method [40], the MR pleiotropy residual sum, and the IVW outlier (MR-PRESSO) method [41]. The advantage of the MR-Egger regression method is that it can assess the directional pleiotropy of IVs and provide a valid causal estimate even if all IVs are invalid [42]. However, this method has a lower efficiency power than the IVW method [39]. Additionally, I² _GX was calculated to test the potentially weak IVs bias in the MR-Egger regression method. An I² _GX >95% indicates a low risk of bias [40]. MR-PRESSO can identify and adjust for outlier IVs in an IVW analysis, thus enabling more accurate estimates after removal of IV outliers [41]. To remove the IVs associated with any confounders that might have affected PECAM-1, we also searched each selected SNP in Phenoscanner [43] and the GWAS catalog [44] for identified associations (P < 5 × 10⁻⁸) with relevant confounders or CVDs found in previously published research. Additionally, to ensure that the conclusions of MR estimates were unaffected by the presence of pleiotropic pathways acting through CVD risk factors, we conducted multivariable MR to confirm the initial results of our IVW analyses [45]. Total cholesterol (TC), TG, LDL-C, HDL-C, alcohol intake, T₂D, smoking, BMI, and depression were all considered confounders. Scatter plots of the MR effects estimated with each method are also presented.

A two-sided P-value < 0.05 was set as the significance threshold, and the Bonferroni corrected threshold for statistical significance (0.05/2 × 9 = 0.0028) was further used to correct for multiple comparisons. All MR analyses described above were implemented with the R packages TwosampleMR, MRPRESSO, and Mendelian Randomization in R software (version 3.5.4; www.r-project.org).

Results

SNPs Selected in this Study

In total, six SNPs were obtained as IVs in this study. Each SNP associated with PECAM-1 reached genome-wide significance (P < 5 × 10⁻⁸). For all chosen SNPs, the F statistics consistently exceeded 10, and ranged between 26 and 279 (Supplementary Table 1), thus indicating that all SNPs had sufficient strength to counteract the risk of weak instrument bias for the subsequent MR analysis. Detailed characteristics of all selected SNPs associated with PECAM-1 are shown in Supplementary Tables 1, 2, and 5.

Causal Effects of PECAM-1 on CVDs

Cochran’s Q test was conducted to detect the directional heterogeneity in each IVW analysis. The findings suggested the absence of significant heterogeneity, with the exception of the IVW analysis for SVS (P = 0.0449) (Supplementary Table 3). Consequently, a random-effects IVW method was applied in the analysis for SVS. Because the P values of Cochran’s Q test in other IVW analyses all exceeded 0.05 (Supplementary Table 3), fixed-effects IVW methods were used for the remaining.

MR analyses in this study. According to the results of the IVW analyses, higher genetically predicted PECAM-1 levels were associated with lower risk of CAD (OR, 0.835; CI, 0.757–0.92; P = 3 × 10⁻⁴) and MI (OR, 0.79; CI, 0.709–0.881, P = 2.03 × 10⁻⁵). Figure 2 displays scatter plots illustrating the MR effect assessed with each method.

Figure 2

Scatter plots for Assessment of Causal Associations Between PECAM-1 and CVDs.

(A) CAD, coronary artery disease. (B) MI, myocardial infarction. (C) AF, atrial fibrillation. (D) HF, heart failure. (E) Hypertension. (F) IS, ischemic stroke. (G) CS, cardioembolic stroke. (H) LAS, large artery stroke. (I) SVS, small vessel stroke. The slope of each line corresponds to the estimated Mendelian randomization (MR) effect for each method; circles indicate marginal genetic associations of each variant with PECAM-1 and the risk of outcomes. Error bars indicate 95% CIs. The x axis depicts the strength of the genetic instrument for the exposure variable.

A higher PECAM-1 level was also causally associated with a higher risk of AF (OR, 1.072; CI, 1.004–1.144; P = 0.0367), and a lower risk of HF (OR, 0.914; CI, 0.847–0.986; P = 0.0197); however, with a Bonferroni corrected threshold, these correlations disappeared (Table 2). PECAM-1 did not show any associations with hypertension, IS, LAS, CS, or SVS (Table 2). Scatter plots depicting the MR effect evaluated with each method are shown in Figure 2.

Table 2

Mendelian Randomization Estimates of the Causal Relationships between PECAM-1 and Cardiovascular Diseases.

Outcomes	No. SNPs	IVW		Weighted median		Simple median		MR-egger		MR-PRESSO
		OR	P-value	OR	P-value	OR	P-value	OR	P-value	OR	P-value
		(95% CI)		(95% CI)		(95% CI)		(95% CI)		(95% CI)
CAD	6	0.835	3.00E-04*	0.786	4.40E-05*	0.893	0.155	0.741	0.0009*	0.845	0.0123 †
		(0.757–0.92)		(0.7–0.882)		(0.764–1.044)		(0.621–0.885)		(0.716–0.954)
MI	6	0.79	2.03E-05*	0.767	5.97E-05*	0.867	0.1177	0.685	0.0002*	0.793	0.0060*
		(0.709–0.881)		(0.673–0.873)		(0.725–1.037)		(0.562–0.834)		(0.714–0.866)
AF	6	1.072	0.0367 †	1.07	0.0721 †	1.073	0.1573	1.023	0.7092	1.064	0.0375 †
		(1.004–1.144)		(0.994–1.151)		(0.973–1.182)		(0.908–1.152)		(1.023–1.121)
HF	6	0.914	0.01967 †	0.893	0.0110 †	0.962	0.5545	0.854	0.2747	0.926	0.0389 †
		(0.847–0.986)		(0.818–0.974)		(0.847–1.093)		(0.743–0.983)		(0.873–0.955)
Hypertension	6	1.004	0.9687	0.98	0.8431	1.007	0.9572	1.111	0.5143	1.017	0.9539
		(0.841–1.197)		(0.799–1.201)		(0.774–1.311)		(0.809–1.526)		(0.865–1.143)
IS	6	0.916	0.1187	0.901	0.0778	0.953	0.5603	0.796	0.015	0.946	0.1795
		(0.82–1.023)		(0.803–1.012)		(0.811–1.12)		(0.662–0.957)		(0.827–0.985)
CS	6	1.003	0.9774	0.904	0.375	1.192	0.3277	0.784	0.1859	1.043	0.9785
		(0.813–1.238)		(0.724–1.129)		(0.838–1.695)		(0.546–1.124)		(0.872–1.134)
LAS	6	0.831	0.1441	0.862	0.3111	0.949	0.7926	0.691	0.1116	0.863	0.1391
		(0.648–1.065)		(0.647–1.149)		(0.644–1.399)		(0.438–1.089)		(0.645–1.017)
SVS	6	0.941	0.7353	0.943	0.6721	0.82	0.3075	0.833	0.6101	0.922	0.749
		(0.662–1.338)		(0.717–1.24)		(0.559–1.201)		(0.414–1.679)		(0.688–1.194)

CAD, coronary artery disease; MI, myocardial infarction; AF, atrial fibrillation; HF, heart failure; IS, ischemic stroke; CS, cardioembolic stroke; LAS, large artery stroke; SVS, small vessel stroke; OR, odds ratio; CI, confidence interval; IVW, inverse-variance-weighted method; MR-PRESSO, MR pleiotropy residual sum and outlier method; SNPs, single-nucleotide polymorphisms.

*P < 0.0028; ^†P < 0.05.

Results of Complementary Analyses

In the sensitivity analyses, weighted median, simple median, MR-PRESSO, and MR-Egger regression were conducted to confirm the causal effects of PECAM-1 on CVDs. The results of the sensitivity analyses were consistent with the findings from the IVW analysis (Table 2). Additionally, we calculated I² _GX to quantify the potential bias arising in the MR-Egger analysis. In our study, I² _GX was higher than 0.95 in MR-Egger regression, thereby indicating a low risk of weak IV bias (Supplementary Table 4).

A summary of all results from the TSMR analyses indicated that the PECAM-1 level was causally associated with CAD and MI. We conducted additional investigations to determine whether the MR association between genetically determined PECAM-1 levels and these specific types of CVD could be attributed to pleiotropic pathways associated with confounders identified in this study. We conducted multivariable MR analysis adjusting for TC, TG, LDL-C, HDL-C, alcohol intake, T₂D, smoking, BMI, and depression. The results remained stable regardless of the model for CAD and MI (adjusted for TC, TG, LDL-C, HDL-C, alcohol, T₂D, smoking, BMI, and depression), thus supporting an independent association of PECAM-1 levels with CAD and MI. However, the association between PECAM-1 and AF did not persist after adjustment for depression (CI, 1–1.14; P = 0.056). Regarding HF, the results became nonsignificant after adjustment for BMI (CI, 0.87–1.04; P = 0.296). Details of the results of the multivariable MR analyses are shown in Table 3.

Table 3

Multivariable MR Analysis Adjusted for the Confounders Identified in this Study.

Model	CAD			MI			AF			HF
	OR	CI	P-value	OR	CI	P-value	OR	CI	P-value	OR	CI	P-value
Adjusted for TC	0.84	0.73–0.91	0.023	0.82	0.7–0.95	0.010	1.13	1.02–1.25	0.018	0.91	0.83–0.99	0.012
Adjusted for TG	0.79	0.71–0.89	<0.001	0.76	0.67–0.86	<0.001	1.08	1.01–1.16	0.049	0.89	0.82–0.97	0.007
Adjusted for LDL-C	0.85	0.73–0.98	0.029	0.82	0.7–0.96	0.014	1.13	1.02–1.25	0.021	0.91	0.83–0.98	0.023
Adjusted for HDL-C	0.81	0.72–0.91	0.001	0.78	0.68–0.89	<0.001	1.11	1.02–1.2	0.017	0.9	0.82–0.99	0.035
Adjusted for alcohol consumption	0.81	0.73–0.91	<0.001	0.79	0.68–0.89	<0.001	1.08	1.02–1.16	0.037	0.91	0.84–0.99	0.024
Adjusted for T2D	0.83	0.75–0.92	0.001	0.79	0.71–0.89	<0.001	1.09	1.02–1.17	0.016	0.91	0.84–0.99	0.025
Adjusted for smoking	0.84	0.75–0.93	0.001	0.79	0.7–0.89	<0.001	1.08	1.01–1.15	0.023	0.91	0.84–0.98	0.014
Adjusted for BMI	0.88	0.79–0.98	0.023	0.83	0.73–0.94	0.004	1.09	1.01–1.17	0.021	0.95	0.87–1.04	0.296
Adjusted for depression	0.83	0.75–0.93	0.001	0.79	0.7–0.89	<0.00	1.08	1–1.14	0.056	0.91	0.84–0.98	0.015

CAD, coronary artery disease; MI, myocardial infarction; AF, atrial fibrillation; HF, heart failure; TC, total cholesterol; TG, triglyceride; LDL-C, low-density-lipoprotein cholesterol; HDL-C, high-density-lipoprotein cholesterol; T2D, type 2 diabetes; BMI, body mass index.

Discussion

Prior observational studies have highlighted associations between PECAM-1 and CVDs [18, 19]. The discoveries from those studies motivated our attempts to better understand the causal effects of PECAM-1 on CVDs. To our knowledge, no conclusive study has fully clarified the causal effects of PECAM-1 levels on any particular subtypes of CVD. To answer this unresolved question, we performed TSMR analysis to estimate the causal associations between PECAM-1 and various subtypes of CVD.

Our findings indicated that a higher PECAM-1 level was causally associated with a lower risk of CAD, in agreement with findings from a prior observational study. In a previous study reporting that PECAM-1 levels are associated with CAD, the researchers compared serum PECAM-1 levels between patients with CAD (n = 137) and controls (n = 110), and found that higher PECAM-1 levels were associated with lower risk of serious coronary artery stenosis [19]. Our results remained significant after adjustment for the confounders identified in this study. These findings support a protective function in CAD. Prior investigations have indicated that the gene polymorphisms of PECAM-1 are genetic risk factors for MI; however, the researchers were unable to clarify the gene polymorphisms’ causal effects on MI and thus could not identify how PECAM-1 levels affected MI risk [18]. Through TSMR analysis, we able to answer this unanswered question, and the results remained significant after adjustment for the confounders. Additionally, the results of the sensitivity analyses were consistent with those of the IVW analysis. The potential relationship of PECAM-1 with CAD and MI may be associated with PECAM-1’s important role in vascular endothelial cells, and its involvement in regulating the physiological processes of vascular inflammation, thrombosis, and vascular permeability [46]. CAD and MI are diseases in which the heart muscle becomes ischemic and necrotic, because of blockage of the coronary arteries, usually caused by atherosclerosis. In the pathological process of atherosclerosis, vascular endothelial injury and inflammation are important factors leading to increased expression of adhesion molecules in vascular endothelial cells, thereby promoting inflammation and the formation of atherosclerotic plaques [47], PECAM-1 may play important roles in these processes, thus providing a potential mechanism explaining the correlation of PECAM-1 with CAD and MI. Our results provide new insights into the role of PECAM-1 as a protective factor that can decrease CAD and MI risk.

To our knowledge, studies exploring the causal effect of PECAM-1 on HF or AF are lacking. Although a higher PECAM-1 level was causally associated with a lower risk of HF (OR, 0.914; CI, 0.847–0.986; P = 0.0197), and a higher PECAM-1 level was associated with a higher risk of AF (OR, 1.072; CI, 1.004–1.144; P = 0.0367), with the Bonferroni corrected threshold, no correlation was observed between PECAM-1 and HF or AF. Moreover, the relationship between PECAM-1 and HF became nonsignificant when BMI was considered in a multivariable model. A similar phenomenon occurred when we considered the results of AF, in which the results became nonsignificant after adjustment for depression in the multivariable model. These findings indicated that the discrepancy might have been due to interference by confounding factors. Further studies should be conducted to further investigate the causal relationship of PECAM-1 in HF and AF.

Some studies have reported an association between PECAM-1 and IS. An observational study in 23 IS patients has found that PECAM-1 levels increase in the serum in the 24 hours after IS [20]. Another study has demonstrated significantly elevated PECAM-1 expression in a mouse model of IS [21]. One study has reported that PECAM-1 controls leukocyte transmigration through the endothelium in a mouse model of IS [22]. These studies together demonstrated that the occurrence of IS may affect PECAM-1 expression levels; however, the causal effect of PECAM-1 on IS remained unclear. In our study, we conducted a TSMR analysis to evaluate the causal effect of PECAM-1 on IS and its subtypes. Unexpectedly, PECAM-1 was not found to be causally associated with IS and its subtypes, and sensitivity analyses indicated the same conclusion as the IVW analyses. These results might potentially have been because when IS occurred, the leukocytes infiltrated into injured areas. PECAM-1 has been identified as a key molecule in leucocyte migration through the endothelium during inflammation; thus, IS may induce increases in PECAM-1 levels [48, 49]. Further research is necessary to investigate the causal effect of PECAM-1 on IS.

This study has several strengths. First, to our knowledge, this study is the first to systematically evaluate the causal association between PECAM-1 and multiple types of CVD, including CAD, MI, AF, HF, hypertension, IS, CS, LAS, and SVS. Second, we conducted various complementary analyses, including the weighted median method, simple median method, and MR-Egger regression method (MR-PRESSO). We also conducted a Cochran’s Q test and a MR-Egger intercept test to address the pleiotropic bias in our study. These complementary analyses and pleiotropy analyses ensured the robustness of the results of our study. Third, to ensure that the results were not affected by confounding factors, we conducted a multivariable MR analysis to further confirm the results of the IVW analyses.

However, this MR study has several limitations that should be acknowledged. First, SNPs associated with PECAM-1 were derived from a published GWAS in 21,718 participants of European ancestry. Although our study confirmed the causal relationship between PECAM-1 and CVDs in the population of European ancestry, the limitation of the study population to individuals of European ancestry restricts the generalizability of our results. Further studies including other populations are warranted to confirm our findings. Second, although no apparent pleiotropy was detected for the IVs in our study, other undiscovered causal pathways between PECAM-1 and CVDs might exist. Third, although we conducted multivariable MR, other potential confounding factors might have gone unnoticed.

Conclusions

Overall, the TSMR analysis conducted in this study provides evidence substantiating a causal association between PECAM-1 levels and both CAD, MI, at the genetic level. PECAM-1 levels is also causally associated with HF and AF. Further investigation should be conducted, because the results did not remain stable after consideration of several confounding factors of AF and HF. Our study does not provide sufficient evidence to support causal effects of PECAM-1 on stroke or hypertension. Our discoveries contribute to a deeper exploration of the etiology of CVDs and encourage researchers to further confirm the role of PECAM-1 as a clinical biomarker.

Data Availability Statement

The data used in this study are from public online datasets. The names of the datasets and accession numbers can be found in the articles in the reference list.

Ethics Statement

This study is an analysis of previously published GWAS data; therefore, ethical review and approval were not required for this study, in accordance with the local legislation. An informed consent statement is not applicable in this study.

Author Contributions

Conceptualization, M.Z.S., J.P., and A.F.C.; methodology, M.Z.S., G.X.L., and Y.C.Z.; software, J.P. and A.F.C.; validation, J.P., G.X.L., and X.X.Y.; formal analysis, M.Z.S., G.X.L., and Y.Z.C.; investigation, M.Z.S.; resources, J.P.; data curation, M.Z.S., Y.M.Z. and X.Q.Y.; writing – original draft preparation, M.Z.S. and G.X.L.; writing – review and editing, J.P. and A.F.C.; visualization, M.Z.S. and A.F.C.; supervision, J.P.; project administration, J.P. All authors have read and agreed to the published version of the manuscript.

Acknowledgments

We thank all researchers who shared their statistics on GWAS for PECAM-1 and cardiovascular diseases.

Conflict of Interest

Jun Pu is Co-Editor-in-Chief of Cardiovascular Innovations and Applications. Jun Pu was not involved in the peer review or decision-making process of the manuscript. The other authors have no competing interests to disclose.

Supplementary Materials

Supplementary materials can be accessed at the following link: https://cvia-journal.org/wp-content/uploads/2024/04/Supplementary_Material.pdf.

Citation Information

Download Citation.

References

1. M Kivimaki, A Steptoe. Effects of stress on the development and progression of cardiovascular disease. Nat Rev Cardiol 2018;15:215–29.

2. GA Roth, GA Mensah, CO Johnson, G Addolorato, E Ammirati, et al. Global Burden of Cardiovascular Diseases and Risk Factors,1990-2019: Update From the GBD 2019 Study. J Am Coll Cardiol 2020;76(25):2982–3021.

3. CE Kosmas, S Rodriguez Polanco, MD Bousvarou, EJ Papakonstantinou, E Pena Genao, E Guzman, et al. The triglyceride/high-density lipoprotein cholesterol (TG/HDL-C) ratio as a risk marker for metabolic syndrome and cardiovascular disease. Diagnostics (Basel) 2023;13:929.

4. E Tasdighi, R Adhikari, O Almaadawy, TM Leucker, MJ Blaha. LP(a): structure, genetics, associated cardiovascular risk, and emerging therapeutics. Annu Rev Pharmacol Toxicol 2023;64:135–57.

5. F Kronenberg. HDL in CKD-the devil is in the detail. J Am Soc Nephrol 2018;29:1356–71.

6. Y Zhang, J Ren. ALDH2 in alcoholic heart diseases: molecular mechanism and clinical implications. Pharmacol Ther 2011;132:86–95.

7. ND Wong, N Sattar. Cardiovascular risk in diabetes mellitus: epidemiology, assessment and prevention. Nat Rev Cardiol 2023;20:685–95.

8. US Preventive Services Task Force; CM Mangione, MJ Barry, WK Nicholson, M Cabana, D Chelmow, et al. Statin use for the primary prevention of cardiovascular disease in adults: US Preventive Services Task Force Recommendation Statement. J Am Med Assoc 2022;328:746–53.

9. FB Ortega, CJ Lavie, SN Blair. Obesity and cardiovascular disease. Circ Res 2016;118:1752–70.

10. DL Hare, SR Toukhsati, P Johansson, T Jaarsma. Depression and cardiovascular disease: a clinical review. Eur Heart J 2014;35:1365–72.

11. PJ Newman, MC Berndt, J Gorski, GC White 2nd, S Lyman, C Paddock, et al. PECAM-1 (CD31) cloning and relation to adhesion molecules of the immunoglobulin gene superfamily. Science 1990;247:1219–22.

12. ME Berman, Y Xie, WA Muller. Roles of platelet/endothelial cell adhesion molecule-1 (PECAM-1, CD31) in natural killer cell transendothelial migration and beta 2 integrin activation. J Immunol 1996;156:1515–24.

13. C Golias, E Tsoutsi, A Matziridis, P Makridis, A Batistatou, K Charalabopoulos. Leukocyte and endothelial cell adhesion molecules in inflammation focusing on inflammatory heart disease. In Vivo 2007;21:757–69.

14. L Gao, Y Zhao, H Wu, X Lin, F Guo, J Li, et al. Polycystic ovary syndrome fuels cardiovascular inflammation and aggravates ischemic cardiac injury. Circulation 2023;148:1958–73.

15. Y Zhao, X Lu, F Wan, L Gao, N Lin, J He, et al. Disruption of circadian rhythms by shift work exacerbates reperfusion injury in myocardial infarction. J Am Coll Cardiol 2022;79:2097–115.

16. JR Privratsky, PJ Newman. PECAM-1: regulator of endothelial junctional integrity. Cell Tissue Res 2014;355:607–19.

17. Z Chen, E Tzima. PECAM-1 is necessary for flow-induced vascular remodeling. Arterioscler Thromb Vasc Biol 2009;29:1067–73.

18. T Sasaoka, A Kimura, S Hohta, N Fukuda, T Kurosawa, T Izumi. Polymorphisms in the platelet-endothelial cell adhesion molecule-1 (PECAM-1) gene, Asn563Ser and Gly670Arg, associated with myocardial infarction in the Japanese. Ann N Y Acad Sci 2001;947:259–70.

19. L Fang, H Wei, SH Chowdhury, N Gong, J Song, CK Heng, et al. Association of Leu125Val polymorphism of platelet endothelial cell adhesion molecule-1 (PECAM-1) gene & soluble level of PECAM-1 with coronary artery disease in Asian Indians. Indian J Med Res 2005;121:92–9.

20. J Zaremba, J Losy. sPECAM-1 in serum and CSF of acute ischaemic stroke patients. Acta Neurol Scand 2002;106:292–8.

21. LH Deddens, GA van Tilborg, A van der Toorn, HE de Vries, RM Dijkhuizen. PECAM-1-targeted micron-sized particles of iron oxide as MRI contrast agent for detection of vascular remodeling after cerebral ischemia. Contrast Media Mol Imaging 2013;8:393–401.

22. J Winneberger, S Schöls, K Lessmann, J Rández-Garbayo, AT Bauer, A Mohamud Yusuf, et al. Platelet endothelial cell adhesion molecule-1 is a gatekeeper of neutrophil transendothelial migration in ischemic stroke. Brain Behav Immun 2021;93:277–87.

23. E Arias, N Nadkarni, R Fang, M Haynes, A Batra, W Muller, et al. Inhibition of PECAM-1 significantly delays leukocyte extravasation into the subcortex post-stroke. FASEB J 2022;36 Suppl 1.

24. M de Castro Leao, A Raffin Pohlmann, A de Cristo Soares Alves, S Helena Poliselli Farsky, M Klimuk Uchiyama, K Araki, et al. Docosahexaenoic acid nanoencapsulated with anti-PECAM-1 as co-therapy for atherosclerosis regression. Eur J Pharm Biopharm 2021;159:99–107.

25. RG Kowalski, A Ledreux, JE Violette, RT Neumann, D Ornelas, X Yu, et al. Rapid activation of neuroinflammation in stroke: plasma and extracellular vesicles obtained on a mobile stroke unit. Stroke 2023;54:e52–7.

26. G Davey Smith, G Hemani. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet 2014;23:R89–98.

27. DA Lawlor, RM Harbord, JA Sterne, N Timpson, G Davey Smith. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med 2008;27:1133–63.

28. SN Islam, T Ahammed, A Anjum, O Albalawi, MJ Uddin. Reporting methodological issues of the Mendelian randomization studies in health and medical research: a systematic review. BMC Med Res Methodol 2022;22:21.

29. M Nikpay, A Goel, HH Won, LM Hall, C Willenborg, S Kanoni, et al. A comprehensive 1,000 genomes-based genome-wide association meta-analysis of coronary artery disease. Nat Genet 2015;47:1121–30.

30. S Shah, A Henry, C Roselli, H Lin, G Sveinbjornsson, G Fatemifar, et al. Genome-wide association and Mendelian randomisation analysis provide insights into the pathogenesis of heart failure. Nat Commun 2020;11:163.

31. JB Nielsen, RB Thorolfsdottir, LG Fritsche, W Zhou, MW Skov, SE Graham, et al. Biobank-driven genomic discovery yields new insight into atrial fibrillation biology. Nat Genet 2018;50:1234–9.

32. C Liu, AT Kraja, JA Smith, JA Brody, N Franceschini, JC Bis, et al. Meta-analysis identifies common and rare variants influencing blood pressure and overlapping with metabolic trait loci. Nat Genet 2016;48:1162–70.

33. R Malik, G Chauhan, M Traylor, M Sargurupremraj, Y Okada, A Mishra, et al. Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes. Nat Genet 2018;50:524–37.

34. HP Adams Jr, BH Bendixen, LJ Kappelle, J Biller, BB Love, DL Gordon, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993;24:35–41.

35. L Folkersen, S Gustafsson, Q Wang, DH Hansen, AK Hedman, A Schork, et al. Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals. Nat Metab 2020;2:1135–48.

36. G Davey Smith, MV Holmes, NM Davies, S Ebrahim. Mendel’s laws, Mendelian randomization and causal inference in observational data: substantive and nomenclatural issues. Eur J Epidemiol 2020;35:99–111.

37. MJ Brion, K Shakhbazov, PM Visscher. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol 2013;42:1497–501.

38. JR Thompson, C Minelli, MF Del Greco. Mendelian randomization using public data from genetic consortia. Int J Biostat 2016;12:2.

39. J Bowden, G Davey Smith, S Burgess. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015;44:512–25.

40. J Bowden, MF Del Greco, C Minelli, G Davey Smith, NA Sheehan, JR Thompson. Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. Int J Epidemiol 2016;45:1961–74.

41. M Verbanck, CY Chen, B Neale, R Do. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018;50:693–8.

42. S Burgess, SG Thompson. Interpreting findings from Mendelian randomization using the MR-Egger method. Eur J Epidemiol 2017;32:377–89.

43. MA Kamat, JA Blackshaw, R Young, P Surendran, S Burgess, J Danesh, et al. PhenoScanner V2: an expanded tool for searching human genotype–phenotype associations. Bioinformatics 2019;35:4851–3.

44. A Buniello, JAL MacArthur, M Cerezo, LW Harris, J Hayhurst, C Malangone, et al. The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics 2019. Nucleic Acids Res 2019;47:D1005–12.

45. S Burgess, SG Thompson. Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects. Am J Epidemiol 2015;181:251–60.

46. A Woodfin, MB Voisin, S Nourshargh. PECAM-1: a multi-functional molecule in inflammation and vascular biology. Arterioscler Thromb Vasc Biol 2007;27:2514–23.

47. P Kong, ZY Cui, XF Huang, DD Zhang, RJ Guo, M Han. Inflammation and atherosclerosis: signaling pathways and therapeutic intervention. Signal Transduct Target Ther 2022;7:131.

48. NA Nadkarni, E Arias, R Fang, ME Haynes, HF Zhang, WA Muller, et al. Platelet endothelial cell adhesion molecule (PECAM/CD31) blockade modulates neutrophil recruitment patterns and reduces infarct size in experimental ischemic stroke. Am J Pathol 2022;192:1619–32.

49. A Kalinowska, J Losy. PECAM-1, a key player in neuroinflammation. Eur J Neurol 2006;13:1284–90.