Role of endothelial dysfunction in the thrombotic complications of COVID-19 patients

Dear Editor,

We note with interest the review of Kunutsor SK. et al.¹ on cardiovascular implications of coronavirus disease 2019 (COVID-19). Indeed, Sars-Cov-2 infection begins in lungs but moves rapidly to the vascular system with platelet alterations and blood clotting abnormalities, and associates with a high incidence of cardiovascular events and venous thromboembolism (VTE), especially in critically ill patients (10–34%).²

Based on autopsy findings, endothelial injury has been hypothesized to play a crucial role in the Sars-Cov-2 associated pro-coagulant condition.³ Very few studies, however, have assessed circulating biomarkers of endothelial damage in COVID-19 patients. Among these particularly interesting are circulating endothelial cells (CECs), circulating endothelial progenitor cells (EPCs), endothelial extracellular vesicles (EEVs) and soluble forms of endothelial adhesive proteins (CAM) which are known to be altered in conditions associated with enhanced cardiovascular risk and to be predictive of vascular complications in various conditions, including infectious diseases.⁴ For these parameters no data are available, to our knowledge, in Sars-Cov-2 infection. Aim of our study was to assess the role of cellular and soluble circulating endothelial derangement parameters as markers of endothelial damage in COVID-19 patients and to unravel if they may identify patients developing VTE or adverse outcome.

Fifty-six COVID-19 patients and 36 healthy, age- and sex-matched controls were enrolled in a multicenter study in the Umbria Region, Italy. Peripheral blood was collected for EEVs, CECs and EPCs by flow cytometry and for sVCAM and sICAM by ELISA.⁵ Four of the enrolled patients were reassessed after disease recovery confirmed by 2 negative nasopharyngeal swabs (mean 72.7 days, 95% CI 32.8–112.6, after the second).

Statistical analyses were performed using GraphPad Prism 8.4 for Windows software. Data not normally distributed were analyzed with the Mann Whitney test; otherwise with the two-tailed unpaired Student's t-test. P<0.05 was considered statistically significant.

The study was approved by the local Ethics Committees (CEAS Umbria n. 3656/20, University of Perugia Bioethics Committee n. 2020–36,346).

Demographics, clinical and main laboratory features of the study population are summarized in Table 1 . All patients were hospitalized (median hospitalization 26 days) and were studied on average 4.2 ± 0.5 days (95%CI 3.2–5.2) from the last positive nasopharyngeal swab. Sixteen patients (28.5%) were admitted into the intensive care unit (ICU) while the others into non-ICU COVID-19 wards. Five ICU and 3 non-ICU patients died during hospitalization. Of the 56 patients, 19 had a partial pressure of oxygen/fraction of inspiration oxygen ratio (PaO₂/FiO₂) lower than 300 and 31 required mechanical ventilation (15 invasive and 16 non invasive). Ten (17.8%) developed 11 thrombotic events (one suffered two thrombotic events) during or immediately after hospitalization (median 9.5 days) confirmed by computed tomography pulmonary angiography or compression ultrasonography (6 pulmonary embolism, 4 deep vein thrombosis, 1 cava vein thrombosis): of these, six were under prophylactic low molecular weight heparin (LMWH) (n = 3 standard-dose, n = 3 intermediate-dose) and four under therapeutic-dose LMWH (one for atrial fibrillation and one for a previous pulmonary embolism).

Table 1.

Demographic and clinical characteristics of the study population.

	COVID-19 patients n = 56	Healthy subjects n = 36	p value
Age (years)	72.1 ± 1.8	68.0 ± 3.0		ns
Sex (% M)	57.1%	40.1%		ns
Leukocytes (x 103/μL)	7.0 ± 0.7	5.7 ± 0.8		ns
Neutrophil-to-lymphocyte ratio (NLR)	8.0 ± 0.9	2.0 ± 0.2		<0.005
Platelets (x 103/μL)	211.1 ± 17.0	208 ± 17.2		ns
D-dimer (ng/ml)	1663 ± 299.0	180.6 ± 21.7		<0.0001
Fibrinogen (mg/dL)	403.1 ± 27.0	323.2 ± 26		ns
VWF: Ag (%)	273.8 ± 26.4	104.0 ± 7.0		<0.0001
VWF: RCo (%)	298.0 ± 28.0	90.0 ± 4.9		<0.0001
Procalcitonin (ng/ml)	1.2 ± 0.5	N.A.
CRP (mg/dL)	4.8 ± 1.5	N.A.
LDH (U/L)	247.2 ± 33.3	N.A.
PaO2/FiO2	241.3 ± 27.1	N.A.
SOFA score (total)	6.0 ± 0.4	N.A.
Days from positive swab	4.2 ± 0.5	N.A.
Thrombotic events (n)	11	N.A.
Comorbidities
Hypertension (n)	32	3		<0.01
Type 2 Diabetes Mellitus (n)	11	1		<0.05
Obesity (n)	12	1		<0.05
Smoker (n)	6	3		ns
Atrial Fibrillation (n)	7	0		ns
Cirrhosis (n)	1	0		ns
Kidney failure (n)	7	0		ns
Stroke (n)	4	0		ns
Peripheral artery disease (n)	7	0		ns
COPD (n)	4	0		ns
Drugs
Antihypertensive agents (n)	11	1		<0.05
Statins (n)	11	0		<0.05
Antiplatelet treatments:
Aspirin (n)	9	0		<0.05
Anti P2Y12 (n)	3	0		ns
Anticoagulant treatments:
LMWH (n)	45	0		<0.0001
-standard	32
-incremented	8
-therapeutic	5
Apixaban (n)	6	0		ns
COVID-19 Treatments
Hydroxycloroquine (n)	5	N.A.
Darunavir/Cobicistat (n)	2	N.A.
Tolicizumab (n)	1	N.A.

Results are reported as mean±SEM if not differently indicated. N.A. not applicable; SOFA: sequential organ failure assessment.

COVID-19 patients had significantly higher CECs and EEVs in comparison with healthy subjects (21.5 ± 2.2 vs 8.1 ± 1.4/μl, p<0.01 and 286.5 ± 38 vs 127.6 ± 21/μl, p<0.05 respectively). CECs correlated with C-reactive protein levels (r = 0.49, p<0.05), neutrophil-to-lymphocyte ratio (r = 0.40, p<0.01) and d-Dimer (r = 0.45, p<0.05), biomarkers of inflammation and hypercoagulability, but did not differ between patients who developed a thrombotic event and those who did not.

Three distinct populations of circulating EPCs (CD34+ and CD309+) were detected based on their CD45 expression. CD45 negative (CD45^neg), which express the regenerative potential of EPCs against vascular damage, were significantly lower in COVID-19 patients compared to controls (Fig. 1 A), while a significant increase of CD45 positive intermediate (CD45^+int) (Fig. 1B) and CD45 positive high (CD45^+high) was observed, suggesting that these EPCs with high phagocytic capability may represent a reactive mechanism to limit viral proliferation.⁶

Fig. 1.

Cellular and soluble markers of endothelial dysfunction in COVID-19 samples. Subpopulations of circulating EPCs based on CD45 expression: angioblast EPCs CD45^neg (A) and hematopoietic EPCs CD45^+int (B) in COVID-19 patients and in healthy controls. Results are expressed as absolute number/μl. *=p<0.05 vs healthy controls. sICAM-1 (C) levels are increased in COVID-19 patients compared to healthy controls (p<0.05) and in plasma of COVID-19 patients admitted into intensive care unit (ICU) or in non-ICU COVID-19 wards (p<0.05) (D), with normal or abnormal PaO₂/FIO₂ ratio (p<0.05) (E) and patients with or without thromboembolic event (p<0.05) (F). (*=p<0.05).

COVID-19 patients also had higher plasma levels of soluble markers of EC disturbance, sVCAM-1 (3122 ± 324 vs 1135 ± 82 ng/ml, p<0.001) and sICAM-1 (Fig. 1C) and VWF:Ag and VWF:RCo (Table 1), as compared with controls. Notably, sICAM-1 was significantly more elevated in COVID-19 patients admitted into ICU compared to those not in ICU (Fig. 1D) and in patients with reduced PaO₂/FiO₂ ratio compared to those with normal PaO₂/FiO₂ (Fig. 1E), suggesting that severe respiratory syndrome and hypoxemia are associated with endothelial damage. A significant correlation was also found between sICAM-1 and the SOFA score (r = 0.65, p<0.01), suggesting that elevated sICAM-1 may represent a marker of severe disease evolution in Sars-Cov-2 infection.

D-dimer, VWF:Ag (not shown) and sICAM-1 (Fig. 1F) were significantly higher in patients who developed VTE than in patients who did not. ROC curve analysis showed that sICAM-1 >519.06 ng/ml discriminates COVID-19 patients with VTE from those without with moderate accuracy (AUC= 0.83, p<0.01) (Suppl. Fig. 1).

Most patients were under standard- (n = 32) or incremented-dose (n = 8) prophylactic LMWH (40/56, 71%) but no differences between treated and untreated patients were found for any of the circulating endothelial dysfunction markers assessed.

In patients who had recovered from COVID-19, CECs, EMPs, EPCs, VWF:Ag, VWF:RCo, sICAM-1 and sVCAM-1 returned to levels close to those of healthy controls, suggesting that endothelial damage is strictly dependent on active COVID-19 infection (Suppl. Fig. 2).

Our results show that COVID-19 patients have increased circulating CECs, EMPs and phagocytic EPCs and increased plasma levels of sICAM-1, sVCAM-1, VWF:Ag and VWF:RCo, with concomitant decrease of angiogenic EPCs, proving that circulating parameters of endothelial derangement are strongly altered in COVID-19 patients. In particular, plasma levels of sICAM-1 and of sVCAM-1 were more than threefold increased probably reflecting the enhanced adhesiveness of microvascular endothelium mediating the strong leukocyte extravasation in tissue, in particular in lungs. Moreover, the endothelial activation triggered by SARS-CoV-2 probably contributes to the strong in vivo platelet activation found in COVID-19 patients and to platelet adhesion to lung endothelium leading to lung injury.

Elevated sICAM-1 predicts cardiovascular events in apparently healthy men and in patients with cardiovascular disease and is associated with recurrent VTE,⁷ and in our study strongly associated with VTE incidence and disease severity, therefore this marker warrants more extensive investigation for prognostic prediction in COVID-19 patients.

In our study prophylactic-dose LMWH did not affect biomarkers of endothelial dysfunction, in agreement with low clinical efficacy in preventing VTE in COVID-19 patients.⁸ A recent study evaluated the impact of therapeutic-dose anticoagulation given to COVID-19 patients prior to hospitalization on endothelial damage, measured by CECs, suggesting that early treatment may prevent COVID-19-associated endothelial lesion.⁹ Thus sICAM-1 might be used as an indicator to switch to therapeutic dose heparin in high-risk patients.

Finally, our data, strongly confirming that COVID-19 is an endothelial disease, provide the rationale for the search of novel therapeutic strategies targeting inflammatory mediators and/or promoting endothelial protection/repair to prevent the thrombotic and systemic complications of COVID-19.

Declaration of Competing Interest

The authors declare no conflict of interest.