Arterial and venous thromboembolism risk associated with blood eosinophils: A systematic review and meta‐analysis

Abstract

The association between blood eosinophil (EOS) counts and arterial/venous thrombosis is unclear. We aim to explore whether EOS count is a risk factor for thrombosis. We searched several databases and preprint platforms using core terms ‘eosinophil’, ‘myocardial infarction’, ‘ischemic stroke’, and ‘venous thromboembolism’ (VTE), among others. Studies comparing the odds ratios (ORs) or risk ratios (RRs) of EOSs with the abovementioned diseases were eligible. Overall, 22 studies were included. A high EOS count was associated with acute coronary artery thrombosis events (OR: 1.23, 95% CI: 1.15–1.32), short‐term cerebral infarction and mortality (RR: 2.87, 95% CI: 1.49–5.51). The short‐term risk of VTE was more common in patients with EOS‐related diseases (RR: 6.52, 95% CI: 2.42–17.54). For coronary artery disease, a high EOS count was a protective factor against 6‐month to 1‐year mortality (RR: 0.56, 95% CI: 0.45–0.69) but was associated with long‐term mortality (RR: 1.64, 95% CI: 1.25–2.14). Therefore, we conclude that for coronary artery thrombosis, EOS count is not associated with AMI events in general population. It may be associated with NSTEMI and STEMI in CAD patients, but more studies are needed to confirm this. In addition, EOS count is associated with an increased risk of both short‐ and long‐term mortality but is not predictive of the composite endpoints. For cerebral artery thrombosis, EOS count may be associated with cerebral infarction and could lead to an increased risk of poor short‐term prognosis. For VTEs, EOS count was a risk factor for some patients, especially those with acute‐phase EOS‐related diseases.

We searched 22 related articles from the PubMed, Embase, Cochrane Library databases and preprint platforms. Then we conducted a systematic review and meta‐analysis to assess the association between blood EOS count and thrombotic events and their associated adverse outcomes in patients with various arterial or venous thrombosis disorders.

1. INTRODUCTION

Blood eosinophilia is reported in numerous conditions, including allergic, infectious, inflammatory and neoplastic disorders. ¹ In addition, recent evidence has shown that arterial/venous thrombosis disorders are associated with abnormal blood eosinophil (EOS) counts. ² , ³ These studies emphasize the need to better understand the role of EOS counts in predicting the occurrence and severity of arterial/venous thrombosis disorders.

Coronary artery thrombosis and cerebral artery thrombosis, also called acute myocardial infarction (AMI) and ischemic stroke, respectively, are the most common arterial thrombotic diseases affecting patient health. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ Recently, there has been increased clinical interest in the use of EOS counts as a predictor of coronary and cerebral arterial thrombosis disorders. However, the findings concerning EOS count and coronary or cerebral thrombosis disorders vary widely across studies. ³ , ⁵ , ⁹ , ¹¹ Patients with venous thromboembolism (VTE), including deep venous thrombosis (DVT) and pulmonary embolism (PE), ³ , ¹² , ¹³ had higher EOS counts than patients without thrombosis, ¹⁴ and persistent EOSs were the sole variable influencing the short‐term recurrence of venous thrombosis. ¹³ Therefore, the impact of EOS count on thrombotic events and the prognosis and severity of arterial/venous thrombosis diseases may depend on the type of arterial/venous thrombosis disorder.

Therefore, in view of the existing evidence and contradictory observations and taking into account the value of blood EOS counts in predicting the risks of thrombotic events and the poor prognosis related to such events, we considered that a systematic review and meta‐analysis assessing the association between blood EOS counts and thrombotic events and their associated adverse outcomes in patients with various arterial or venous thrombosis disorders is needed.

2. METHODS

The protocol of this study was preregistered in PROSPERO (CRD 42021284580) and adheres to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA guidelines [Table S1]). ¹⁵ Since published deidentified data were used, this study was exempt from institutional review board approval.

2.1. Systematic search

We systematically searched the MEDLINE/PubMed, Cochrane Library and Embase databases using the core terms ‘eosinophil’, ‘myocardial infarction’, ‘acute coronary syndrome’, ‘coronary heart disease’, ‘cerebral infarction’, ‘ischemic stroke’, ‘pulmonary embolism (PE)’, ‘venous thromboembolism (VTE)’ and ‘deep venous thrombosis (DVT)’ in the article titles, abstracts and keywords. We also searched articles from preprint platforms (Research Square and medRxiv) and scrutinized the reference lists of relevant articles to find additional pertinent studies. Two reviewers independently searched the databases and reviewed the full texts of the eligible articles.

2.2. Inclusion and exclusion criteria

Cohort, cross‐sectional and case–control studies comparing the odds ratios (ORs) or hazard ratios (HRs) of arterial/venous thrombosis events (AMI, ischemic stroke, cerebral infarction, VTE, PE and DVT) and clinical outcomes (including but not limited to mortality, severity classification, and major adverse cardiac events [MACEs]) related to arterial/venous disorders between patients with normal and abnormal EOS counts were eligible. The inclusion criteria were as follows: (1) studies that reported ORs, HRs, or RRs of the association between EOS count and coronary artery thrombosis and related outcomes among patients with AMI and CHD; (2) studies that reported ORs, HRs, or RRs of the association between EOS count and cerebral artery thrombosis and related outcomes among patients with cerebral infarction or ischemic stroke; (3) studies that reported ORs or HRs of the association between EOS count and VTE and related outcomes among VTE patients; and (4) studies published in English. Any studies that did not meet the abovementioned criteria and had data that were not extractable were excluded.

2.3. Data extraction

Two authors (W‐J.X. and S.W.) independently extracted information from the studies, including the authors, study design, sample size, inclusion/exclusion criteria, follow‐up terms, and results. Disagreements were resolved through consensus by discussion and inspection.

2.4. Outcomes

The primary outcomes of interest were the ORs or HRs of arterial/venous thrombosis events, severity and composite outcomes associated with EOS count. The predefined arterial/venous thrombosis events were AMI, ischemic stroke/cerebral infarction, VTE, PE and DVT. The secondary outcome was mortality.

2.5. Quality assessment

The risks of bias of the included studies were assessed with the Newcastle–Ottawa Quality Assessment Scale (NOS) for cohort and case–control studies. ¹⁶ This scale rates studies on three major domains: selection, comparability, and exposure/outcome. The NOS score represents the quality of the included study. High‐quality studies were defined as those with a score of 7 or higher, moderate‐quality studies had a score between 4 and 6, and low‐quality studies had a score less than 4. The Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Analytical Cross‐Sectional Studies was applied for cross‐sectional studies. ¹⁷ The overall scores for each study were calculated as percentages and categorized as a quality assessment of high (80%–100%), fair (50%–79%) or low (<50%). ¹⁸

2.6. Statistical analysis

Data management, the transformation of the effect size, and the calculation of pooled prevalence were performed using Review Manager (RevMan) (Version 5.4, The Cochrane Collaboration, 2020). The results were reported using relative risks (RRs) with 95% confidence intervals (CIs) for cohort studies and ORs with 95% CIs for case–control and cross‐sectional studies using the generic inverse variance method. We applied fixed and random effect models depending on heterogeneity. The I ² statistic was calculated to describe the heterogeneity, ranging from 0% to 100% (values of 25%, 50%, and 75% were considered to represent low, moderate, and high heterogeneity, respectively). Funnel plot asymmetry was used to detect publication bias if more than 10 studies were included. ¹⁹

3. RESULTS

3.1. Search results

The initial systematic search identified 219 articles, including 209 records from the databases and 10 records from the preprint platforms, and 22 articles were qualitatively identified. Of these, 19 articles were ultimately quantitatively analyzed (Figure 1). The study characteristics of all of the included studies are summarized in Table 1. Ultimately, 22 articles were included with 805 329 subjects. Of these 22 articles, 13 were cohort studies, ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹² , ¹³ , ²⁰ , ²¹ , ²² , ²³ 6 were cross‐sectional studies, ¹⁰ , ¹⁴ , ²⁴ , ²⁵ , ²⁶ , ²⁷ and 3 were case–control studies. ¹¹ , ²⁸ , ²⁹ Of the included studies, 14 studies were related to cardiovascular diseases, 3 were related to cerebral thrombosis/cerebral infarction, and 6 were related to venous thrombotic events, among which the study by Anoop et al involved both cardiovascular disease and cerebral artery thrombosis. ⁹

FIGURE 1.

PRISMA diagram of articles screened for inclusion by disease type.

TABLE 1.

Characteristics of the included studies.

Author and year	Design and sample size	Inclusion and exclusion criteria	Follow‐up	Endpoints	Association results
Cardiovascular diseases
Robert Olivares 1993 20	CoS n = 3779	Inclusion: Male civil servants	Mean: 66.5 months	Prevalence of CAD	—
		Exclusion: CAD history and abnormal electrocardiogram
Iqbal S. Toor 2012 21	CoS n = 909	Inclusion: Elective indication or ACS (UA/NSTEMI)	Median: 54 months (IQR: 47–65)	All‐cause mortality	High EOS count was initially associated with reduced mortality but was associated with increased mortality after 6 months
		Exclusion: STEMI; hemodynamic instability requiring angioplasty; failed PCI procedure
Mehmet Demir 2014 27	CSS n = 80	Inclusion: Isolated CSF without any stenotic lesions; normal subjects	Perioperative period	CSF	High EOS count was associated with high CSF
		Exclusion: CAD; chronic renal, liver, lung, hematological or valvular diseases; hypertension; DM; CHD; left ventricular systolic dysfunction; anemia; pregnancy; OSAS; malignancy; hyperthyroidism; infection; hypercholesterolemia, etc.
Erhan Tenekecioglu 2015 10	CSS n = 251	Inclusion: NSTEMI and UA	Perioperative period	Correlation between EOS count and coronary thrombus	—
		Exclusion: History of trauma, surgery, neoplasm, infectious disease, cardiogenic shock, etc.
Monica Verdoia 2014 11	CCS n = 1546	Inclusion: Elective indication or ACS (UA/NSTEMI)	Perioperative period	Periprocedural MI; TnI ≥3 ULN or an increase by 50%	—
		Exclusion: STEMI and hemodynamically unstable patients requiring urgent angioplasty
Monica Verdoia 2015 28	CCS n = 3742	Inclusion: Elective indication and ACS	Perioperative period	Prevalence of CAD	—
		Exclusion: STEMI or UA patients requiring urgent angiography
Jun Wang 2016 26	CSS n = 502	Inclusion: UA with coronary stenosis ≥80%	Perioperative period	CCC	High EOS count was associated with high‐grade CCC
		Exclusion: AMI
Korkmaz Özge 2016 8	CoS n = 251	Inclusion: CAD patients undergoing isolated on‐pump CABG	Median: 6.2 ± 0.8 months	Cardiovascular mortality	Low EOS count predicted mortality in CAD patients with CABG
		Exclusion: Infections; hepatic, renal or hematological disorders; inflammatory diseases with steroid therapy; malignancy, etc.
Anoop Dinesh Shah 2016 9	CoS n = 775 231	Inclusion: Age ≥30 years without CVD at baseline	Median: 3.8 years	Occurrence of ACS, sudden coronary or cardiac death, etc.	Low EOS count in the general population was associated with increased short‐term incidence of HF and coronary death
		Exclusion: A prior history of CVD, pregnancy record within 6 months, HIV, splenectomy or dialysis
Takao Konishi 2017 7	CoS n = 331	Inclusion: Patients who underwent PCI for STEMI within 24 h	1 year	MACEs	High EOS count independently predicted MACEs
		Exclusion: Infections, allergic diseases, or malignancy; evaluated ≥48 h after onset; or died within 24 h
Arthur Shiyovich 2017 23	CoS n = 2129	Inclusion: AMI	Median: 8.1 year	All‐cause death at 1, 5, and 10 years	High EOS count independently predicted death in the long term after AMI
		Exclusion: Cancer, chronic inflammatory diseases, systemic infection, autoimmune disease, and in‐hospital stay <3 days
Xueyan Zhao 2019 5	CoS n = 8943	Inclusion: Triple‐vessel CAD	Median: 7.5 year	All‐cause death MACCEs	High EOS count independently predicted MACCEs
		Exclusion: Lack of C‐reactive protein data
Shanshan Gao 2019 25	CSS n = 5287	Inclusion: CAD patients undergoing coronary angiography	Period	NSTEMI/ASTEMI	Low EOS count remained strongly associated with severe CAD and acute coronary arterial thrombotic events
		Exclusion: Autoimmune or allergic diseases, parasitic infections, malignancies, severe liver or kidney dysfunction, HF or shock, rheumatic or valvular heart diseases, CABG
Mohammad Alkhalil 2019 4	CoS n = 606	Inclusion: STEMI with PCI intervention	Median: 3.5 year	Composite AEs (death, MI, stroke)	Low EOS count was a negative marker for AMI patients
		Exclusion: Not residents of Northern Ireland
Cerebral artery thrombosis
Yusuke S. Hori 2016 22	CoS n = 405	Inclusion: Evaluated within 24 h after symptom onset	2 months	Mortality	Eosinopenia was strongly associated with 2‐m mortality in cerebral infarction patients
		Exclusion: A history of cerebral infarction, infection, hematologic or chronic inflammatory disease, and the use of corticosteroids
Anoop Dinesh Shah 2016 9	CoS n = 775 231	Inclusion: Age ≥30 years without CVD at baseline in the CALIBER program	Median: 3.8 years	Occurrence of TIA, stroke, subarachnoid hemorrhage	EOS count was associated with subarachnoid hemorrhage but was not associated with the short‐term incidence of intracerebral or ischemic stroke
		Exclusion: A prior history of CVD, pregnancy record within 6 months, HIV, splenectomy or dialysis
Huimin Zhao 2020 6	CoS n = 174	Inclusion: Evaluated within 72 h after symptom onset	90 days	Modified Rankin Scale score >3	Eosinopenia was significantly associated with mortality
		Exclusion: Any diseases and drugs interfering with the EOS count
Venous thromboembolism events
Serdar Keceog˘lu 2014 29	CCS n = 89	Inclusion: Persistent AF	Not mentioned	Thrombus in the LA or LA appendage	High EOS count was associated with LA thrombus in patients with nonvalvular AF
		Exclusion: CAD; chronic renal, liver, hematological and valvular diseases; CHD; HF; left ventricular systolic dysfunction; marked LA dilatation; hyperthyroidism; pregnancy; OSAS; COPD; stroke; malignancy, etc.
F.N.Ozdemir 2005 14	CSS n = 141	Inclusion: AVF	6 months prior to thrombosis events	AVF thrombosis	High EOS count was associated with AVF thrombosis
		Exclusion: DM, graft, fistulae, amyloidosis, and vasculitis
Asli Kurtar Mansiroglu 2021 24	CSS n = 243	Inclusion: Suspected DVT	At the time of diagnosis before any treatment	Acute DVT	Low EOS count may lead to a higher probability of acute DVT rather than indeterminate and chronic DVT
		Exclusion: Prior DVT; pregnancy; inflammatory, infectious, hematological or autoimmune diseases; neoplasm; trauma; renal or liver failure
Yecheng Liu 2020 3	CoS n = 63	Inclusion: Aged 18–60 years; acute VTE; onset VTE after a chronic course of HE; Autar DVT risk assessment scale score ≤5	3–6 months after discharge	Recurrent VTE	HE is a potential risk factor for VTE
		Exclusion: Thrombophilia
Valériane Réau 2021 13	CoS n = 54	Inclusion: Aged ≥15 years; VTE; EOS ≥1G/L at VTE occurrence	Median: 24 [IQR: 10–62] months	Recurrence of VTE, major bleeding events, vascular relapse and death.	Persistent eosinophilia was associated with VTE relapse
		Exclusion: History of VTE, thrombophilia, or major transient or reversible predisposing factor for VTE
Alessandra Bettiol 2021 12	CoS n = 573	Inclusion: EGPA patients	Median: 1677 (IQR: 663–3137) days	First AVTE events a	High EOS count was associated with AVTE

Abbreviations: ACS, acute coronary syndrome; AEC, absolute eosinophil count; AF, atrial fibrillation; AVF, arteriovenous fistulae; AVTE, acute arterial and venous thromboembolic event; CABG, coronary artery bypass graft; CAD, coronary artery disease; CALIBER, cardiovascular disease research using linked bespoke studies and electronic health records; CCC, coronary collateral circulation; CCS, case–control study; CHD, congenital heart disease; COPD, chronic obstructive pulmonary disease; CoS, cohort study; CSF, coronary slow flow; CSS, cross‐sectional study; CVD, cardiovascular disease; CVST, cerebral venous sinus thrombosis; DM, diabetes mellitus; DVT, deep venous thrombosis; eGPA, eosinophilic granulomatosis with polyangiitis; HE, hypereosinophilia; HF, heart failure; IQR, interquartile range; LA, left atrium; MACCE, major adverse cardiovascular and cerebrovascular event; MACE, major adverse cardiac event; MI, myocardial infarction; NSTEMI, acute myocardial infarction with non‐ST‐elevation; OSAS, obstructive sleep apnea; PCI, percutaneous coronary intervention; RR, relative risk; SVT, superficial venous thrombosis; TIA, transient ischemic attack; UA, unstable angina pectoris; VTE, venous thromboembolism.

^a Venous events included DVT; PE; SVT; ovarian, renal, splanchnic, jugular, and retinal vein occlusion; and CVST. Arterial events included AMI, stroke, TIA, acute ischemia of the limbs, and arterial retinal occlusion.

3.2. Methodological quality assessment

All articles were assessed for methodological quality. According to the NOS assessment, most studies showed no evidence of selection bias, with good comparability in each cohort and outcome assessment. All included cohort, case–control, and cross‐sectional studies were rated as high quality (Tables S2–S4). For the subgroup analysis, fewer than 10 studies were included, so no publication bias assessment was applied.

3.3. Main analyses

3.3.1. Cardiovascular events

In an analysis of included studies of cardiovascular events, it was found that a high EOS count was associated with acute coronary artery thrombosis events (OR: 1.23, 95% CI: 1.15–1.32, p < 0.00001, I ²  = 87%) in all populations. A pooled analysis of 2 general population studies ⁹ , ²⁰ showed that an increase in EOS count was not associated with AMI with moderate heterogeneity (OR: 0.96, 95% CI: 0.83–1.12, p = 0.62, I ² = 60%). The study by Gao et al ²⁵ revealed that an increased EOS count was associated with an increased risk of non‐ST‐elevated myocardial infarction (NSTEMI) and ST elevation myocardial infarction (STEMI) in the CAD population, with an OR of 1.37 (95% CI: 1.26–1.49, p < 0.00001). However, Verdoia et al ¹¹ did not find an association between an increased EOS count and NSTEMI in the CAD population (OR: 1.04, 95% CI: 0.86–1.26, p = 0.69) (Figure 2A). In the study by Tenekecioglua et al, ¹⁰ an increased EOS count was found to be an independent predictor of coronary thrombus in patients with non‐ST segment elevation acute coronary syndrome (NST‐ACS). However, this study was not included in the pooled analysis because of the converted data.

FIGURE 2.

The association of EOS count with the acute phase of coronary artery thrombosis (A), the risk ratio of EOS count for mortality associated with CAD (B) and the risk ratio of EOS count for the composite endpoints of CAD (C).

With regard to mortality associated with CAD, including but not limited to AMI and isolated on‐pump coronary artery bypass, the RR of mortality was 0.56‐fold higher for patients with low EOS counts than for those with high EOS counts within 6 months to 1 year of an AMI diagnosis (95% CI: 0.45–0.69, p < 0.00001, I ² = 1%). ⁸ , ⁹ , ²¹ , ²³ Interestingly, a pooled analysis of 3 studies ⁵ , ²¹ , ²³ showed that a high EOS count was associated with an increased risk of long‐term mortality in a median follow‐up period of 5–7.5 years with moderate heterogeneity (RR: 1.64, 95% CI: 1.25–2.14, p = 0.0003, I ² = 58%) (Figure 2B).

For noncoronary artery thrombus and nonfatal cardiovascular events, high EOS counts were not found to be associated with CAD severity (OR: 0.99, 95% CI: 0.89–1.1, p = 0.9). ²⁸ A high EOS count was a predictor of high‐grade coronary collateral circulation (CCC) (OR: 1.969, 95% CI: 1.210–3.321, p = 0.006), ²⁶ and coronary slow flow (CSF) patients were more likely to have a high EOS count (OR: 1.02, 95% CI: 1.006–1.018, p = 0.076). ²⁷ A cross‐sectional study revealed ⁵ that a low EOS count was associated with severe CAD (OR: 0.897, 95% CI: 0.850–0.946, p < 0.0001) and NSTEMI/anterior ST‐elevation myocardial infarction (ASTEMI) (OR: 0.728, 95% CI: 0.670–0.791, p < 0.0001). Nevertheless, pooled analysis of 3 cohort studies ⁴ , ⁵ , ⁷ revealed that EOS count was not predictive of the composite endpoints, such as MACEs and major adverse cardiovascular and cerebrovascular events (MACCEs) (Figure 2C).

3.3.2. Cerebral artery thrombosis

For acute ischemic stroke, only one study revealed that EOS count was independently associated with larger cerebral infarction risk, increased by 4.05‐fold (RR: 4.05, 95% CI: 1.31–12.51, p = 0.015). Only the study by Hori et al described the long‐term effect of EOS count on ischemic stroke ⁹ ; there was no strong association between a low EOS count and ischemic stroke in the general population, either in the first 6 months or over a median follow‐up of 3.8 years (RR: 1.17, 95% CI: 0.87–1.56; RR: 1.02, 95% CI: 0.88–1.19, both p > 0.05, respectively). For the secondary outcome, in 2 studies evaluating mortality, ⁶ , ²² an increased EOS count led to an increased risk for poor short‐term prognosis, with an RR of 2.87 without heterogeneity in the short‐term follow‐up (95% CI: 1.49–5.51, p = 0.002, I ² = 0) (Figure 3A).

FIGURE 3.

The association of EOS count with short‐term cerebral infarction and mortality associated with such events (A). The association of EOS count with venous thromboembolism events in patients with EOS‐related diseases (B) and in the general population (C).

3.3.3. VTEs

In patients with EOS‐related diseases, hypereosinophilia is a potential risk factor for VTE. ³ , ¹³ In pooled analysis, hypereosinophilia increased the short‐term risk of VTE over a period of 6 months without heterogeneity (HR: 6.52, 95% CI: 2.42–17.54, p = 0.0002, I ² = 0) (Figure 3B). Only the study by Bettiol et al ¹² described the long‐term VTE risk associated with high EOS counts in EOS‐related diseases and found that EOS count (≥500 U/mmc) was not associated with long‐term VTE during a median follow‐up of 1677 days (IQR: 663–3137 days) (HR: 0.70, 95% CI: 0.29–1.67, p > 0.05). In addition, Liu et al ³ found that PE in patients with IHE/HES and DVT is associated with a higher EOS count. In the general population, a high EOS count was also a risk factor for venous thrombosis in the acute phase of venous thrombosis with mild heterogeneity (OR: 1.01, 95% CI: 1.00–1.01, p < 0.00001, I ² = 36%) ¹⁴ , ²⁹ (Figure 3C). However, in the study by Mansiroglu et al, ²⁴ a low EOS count was associated with DVT in the acute phase of DVT (OR: 0.035, 95% CI: 0.007–0.189, P < 0.001).

4. DISCUSSION

Our study found that there were significant differences in the risk of arteriovenous thrombosis in different populations and diseases.

4.1. EOS count and coronary artery thrombosis

In the general population, we found that EOS count was not associated with AMI events. In patients with CAD, Gao et al found that the RR of AMI patients with an increased percentage of EOSs was higher than that of patients with decreased EOS levels. However, Monica et al found that EOS levels were not associated with the occurrence of periprocedural myocardial infarction (PMI) or myonecrosis in CAD patients undergoing nonurgent percutaneous coronary intervention. ¹¹ We did not perform a pooled analysis of Gao and Verdoia's studies because of the high heterogeneity, ¹¹ , ²⁵ which may be due to the different populations of NSTEMI/ASTEMI and unstable angina (UA)/NSTEMI patients. Moreover, Gao's study confirmed that there were differences in the EOSs and the percentages of eosinophils in leukocytes (PELs) between NSTEMI patients and ASTEMI patients, and they found that the PELs and leukocyte counts had a strong negative correlation in groups of PEL tertiles and that low PELs were closely related to AMI. ²⁵ Therefore, we believe that the change in the PELs may be attributed to the change in EOS count, which is related to thrombosis events.

The role of EOSs in coronary thrombosis events usually follows two mechanisms: atherosclerotic plaque and thrombosis. On the one hand, EOSs can interact with IL‐5, interferon, interferon‐α and collagen, thereby involving them in the production of monocytes and endothelial cells and, ultimately, the development of atherosclerosis. ¹¹ , ³⁰ , ³¹ On the other hand, studies have shown that in the acute phase of coronary artery thrombosis EOSs enter coronary atherosclerotic plaques due to a stress response and the influence of treatment with anticoagulants and chemokines, ²¹ , ³² which leads to a decrease in the EOS count. Meanwhile, EOSs entering the plaque can not only interact with tissue factors to activate the endogenous coagulation pathway but can also release platelet‐activating factors, major basic protein (MBP), eosinophil cationic proteins (ECPs) and inflammatory substances, ultimately leading to thrombosis. ³

Interestingly, we found that patients with lower EOS counts had a high risk of short‐term (6–12 months) mortality, while patients with higher EOS counts had a high risk of long‐term mortality. The causes of the former may be emergency and stress responses, a large number of peripheral EOSs entering the thrombus site, low immunity or immunosuppression. ⁹ , ³² The reason for the latter may be related to the roles of EOSs, including infiltrating the stent implantation site, releasing mediators that increase platelet aggregation after PCI and promoting the formation of atherosclerotic plaques and thrombosis. ³³ In the early stage of disease treatment, dual antiplatelet therapy can offset this risk. ²¹ As the follow‐up time is prolonged, patients with high EOS counts may release more sufficient EOSs into the lesion site and induce immunosuppression by inhibiting T‐cell activation and reducing their activity, resulting in an increase in long‐term mortality in these patients.

There are several causes of the heterogeneity in MACCEs. First, the proportion of patients who died varied among studies. Second, different follow‐up durations were noted among the studies. In three studies, the follow‐up durations were 1 year, a median of 3.5 years, and a median of 7.5 years. Third, different definitions of MACCEs were applied. The endpoints in Zhao's study ⁵ did not include revascularization. Alkhalil's study showed that there was a significant difference in unplanned revascularization among patients with lower or normal EOS counts, which led to a significant difference in the MACCE results. ⁴

Whether the change in EOS count is the cause or result of coronary thrombosis is still unknown. In Demir's study, an increase in EOS count was related to CSF, which can lead to coronary thrombotic events. ²⁷ Gyurko's study showed that the increase in EOS count is secondary to other diseases and represents the inflammatory state caused by vascular injury; thus, increases in the EOS count are the result of atherosclerosis. Wang's study found that AMI patients establish a high level of CCC through the autocrine and paracrine regulation of EOSs and the regulation of EOSs on lymphocytes, which is conducive to a decrease in the incidence of MACCEs²⁶. This also proves that the change in EOS count is the result of thrombotic events. We need to further study the impact of EOS counts on different MACCE endpoints in different periods.

4.2. EOS count and cerebral artery thrombosis

In this meta‐analysis, no studies on the relationship between EOS count and acute phase cerebral infarction were found, indicating the need for further attention in the future. Our meta‐analysis showed that a decreased EOS count (e.g. eosinopenia) predicted a poor short‐term risk. The causes of eosinopenia are as follows: first, the human body stimulates the release of adrenal glucocorticoids and epinephrine due to acute stress, resulting in a reduction in EOS count ³⁴ ; second, chemokines such as complex 5A and fibrin fragments induce EOSs to enter the target organ or tissue, that is, the interior of the thrombus, resulting in a decrease in EOS count; and third, EOS degranulation causes cytotoxic protein‐mediated thrombosis and endothelial injury, resulting in EOSs that cannot be recognized in the circulation. ²⁶ The decrease in EOSs in the peripheral blood circulation leads to a decline in immunity, called ‘stroke‐induced immunosuppression’, making patients more susceptible to infection and increasing mortality. ²² Inevitably, massive infarction and tissue ischemia will lead to poor prognosis.

4.3. EOS count and VTE

The eosinophilia‐related diseases in this study included clonal, reactive, overlapping and idiopathic eosinophilia, such as hypereosinophilic syndromes (HESs), eosinophilic granulomatosis with polyangiitis (EGPA) and parasitic diseases. In addition, EOSs can lead to venous, arterial and mixed thrombosis. ¹³ The RR of venous thrombosis in eosinophilia‐related disease patients with persistent eosinophilia was 6.52 times higher than that in patients with nonpersistent eosinophilia.

EOSs can play a role in the internal pathway of coagulation by producing tissue factors. ³⁵ In addition, EOSs entering the thrombus are activated by platelets and release ECP, MBP and platelet‐activating factor, which have a positive effect on the process of coagulation activation. EOSs can also initiate and regulate local inflammation ²¹ and damage endothelial cells, leading to thrombosis, which is consistent with the findings of Reau et al. ¹³

This study also showed that patients with EOS‐related diseases with continuous increases in EOS count are more likely to develop venous thrombosis. According to our pooling analysis, we believe that increasing EOS counts are associated with thrombotic events. We speculate that EOS‐related diseases may have other mechanisms that can lead to thrombosis. In any case, we believe that thrombotic events can be prevented by reducing EOS levels. In addition, we found that coronary artery thrombosis, cerebral artery thrombosis, and VTE had different correlations with EOSs, which we thought might be related to the composition of the thrombus in addition to the location of arterial or venous thrombosis. An arterial thrombus is generally a white thrombus, while a venous thrombus is generally a red thrombus. This difference in composition may be related to EOSs.

By studying the correlation between eosinophilia and thrombus, we are able to assess the risk of coronary artery thrombosis, cerebral artery thrombosis, and DVT in the population, which can help us diagnose high‐risk patients as early as possible and improve prognosis. However, some of our findings are different from those of other articles, which we consider to be mainly related to the nature of systematic reviews, which include the results of multiple studies, and the small number of studies included.

5. LIMITATIONS

The main limitation of this study was the small sample size included in the analysis because there are few articles on the relationship between EOS count and arteriovenous thrombosis risk, which also highlights the importance of this review and meta‐analysis in providing evidence. Second, due to the small number of studies included, we did not perform a subgroup analysis of the impact of EOS count on the risk of long‐term ischemic stroke and the long‐term VTE risk in EOS‐related diseases. Third, thrombus may occur in familial EOS patients, and EOS mechanistically promotes thrombus generation. However, in all the studies we included, we could not extract relevant data to explore the relationship between familial EOS and thrombosis, and we hope to conduct more basic studies on familial EOS and thrombosis in the future. Fourth, we did not perform a sensitivity analysis. These factors may produce high heterogeneity and affect the reliability of the statistical results. In the future, multicenter randomized controlled trials are needed to further verify the impact of EOS counts on arterial and venous thromboembolism risk. Fifth, selection bias cannot be completely excluded from our meta‐analysis, and publication bias could not be assessed because of the insufficient number of research articles.

6. Clinical implications and future directions

Our analysis suggests that the levels of EOSs in different diseases may be related to the risk of arterial and venous thrombosis. For the general population, disorders of EOS levels have little influence on the occurrence of coronary and cerebral artery thrombosis events. However, for patients with cardiovascular disease, a high EOS count is associated with opposite effects on short‐term and long‐term mortality, which needs to be verified by further studies to determine whether different treatments should be used in different periods of disease for these patients. For patients with cerebral artery thrombosis, a high level of EOSs may predict a higher short‐term mortality, suggesting that we should pay more attention to these patients in the early diagnosis period and determine the treatment therapy in a timely manner. At the same time, for patients with hypereosinophilia, higher EOS levels may be associated with a short‐term risk of VTE, and more attention should be given to these patients to prevent the occurrence of VTE.

7. CONCLUSION

For coronary artery thrombosis, EOS count is not associated with AMI events in the general population. This finding may be associated with NSTEMI and STEMI in the CAD population, which needs more studies to be confirmed. In addition, it is associated with an increased risk for mortality in both the short and long term but is not predictive of the composite endpoints. For cerebral artery thrombosis, EOS count may be associated with cerebral infarction and could lead to an increased risk of poor short‐term prognosis. In terms of VTEs, EOS count was also a risk factor for some people, especially those with EOS‐related diseases in the acute phase. These findings are important for individual treatment. However, more clinical trials are needed to assess the ability of thrombotic screening and explore the optimal strategy for patients with abnormal EOS counts and a variety of different arterial/venous diseases.

AUTHOR CONTRIBUTIONS

W‐J.X. and R.J. conceived and designed the study. W‐J.X. and S.W. analyzed the data. L.W., P.Y., W‐J.X. and R.J. wrote the paper. J‐X.H. and R.J. edited the paper. All authors read and approved the final manuscript.

CONFLICT OF INTEREST

The authors have no conflicts of interest to declare.

ETHICS APPROVAL STATEMENT

This article does not contain any studies with human or animal subjects.

Supporting information

Table S1

Table S2

Table S3

Table S4

Xu W‐J, Wang S, Yuan P, Wang L, Huang J‐X, Jiang R. Arterial and venous thromboembolism risk associated with blood eosinophils: A systematic review and meta‐analysis. Anim Models Exp Med. 2022;5:470‐481. doi: 10.1002/ame2.12277

DATA AVAILABILITY STATEMENT

The data and materials are available from the corresponding author upon request.