Increased circulating platelet-derived extracellular vesicles in severe COVID-19 disease

Abstract

Coagulation disturbances are major contributors to COVID-19 pathogenicity, but limited data exist on the involvement of extracellular vesicles (EVs) and residual cells (RCs). Fifty hospitalized COVID-19 patients stratified by their D-dimer levels into high (>1.5 mg/L, n = 15) or low (≤1.5 mg/l, n = 35) and 10 healthy controls were assessed for medium-sized EVs (mEVs; 200–1000 nm) and large EVs/RCs (1000–4000 nm) by high sensitivity flow cytometry. EVs were analyzed for CD61, CD235a, CD45, and CD31, commonly used to detect platelets, red blood cells, leukocytes or endothelial cells, respectively, whilst phosphatidyl serine EVs/RCs were detected by lactadherin-binding implicating procoagulant catalytic surface. Small EV detection (sEVs; 50–200 nm) and CD41a (platelet integrin) colocalization with general EV markers CD9, CD63, and CD81 were performed by single particle interferometric reflectance imaging sensor. Patients with increased D-dimer exhibited the highest number of RCs and sEVs irrespective of cell origin (p < .05). Platelet activation, reflected by increased CD61+ and lactadherin+ mEV and RC levels, associated with coagulation disturbances. Patients with low D-dimer could be discriminated from controls by tetraspanin signatures of the CD41a+ sEVs, suggesting the changes in the circulating platelet sEV subpopulations may offer added prognostic value during COVID progression.

Plain Language Summary

What is the context?

• Coronavirus disease 19 (COVID-19) frequently leads to blood clotting disturbances, including thromboses.

• Particles smaller than cells, extracellular vesicles (EVs), and residual cells (RCs) affect blood clotting, but data on their role and diagnostic utility in COVID-19 are sparse.

What is new?

• In this study, we assessed 50 hospitalized COVID-19 patients and 10 healthy controls for their different EV subpopulations and residual cells (50–4000 nm).

• Blood clotting marker D-dimer, which is elevated in severe COVID-19 infection, was used to characterize disease severity and stratify the patient subgroups. Fifteen patients (30%) with high D-dimer (>1.5 mg/L) were compared to controls, and 35 patients with lower D-dimer (≤1.5 mg/mL).

• The most topical state-of-the-art methods for detection of EV subpopulations, that is, high sensitivity flow cytometry (hsFCM) and single particle interferometric reflectance imaging sensor (SP-IRIS), were used with markers indicative of platelet, red blood cell, leukocyte or endothelial cells. The subpopulations differentiated by platelet and tetraspanin signatures by hsFCM and SP-IRIS, respectively.

• The main findings are

  • Patients with high D-dimer systematically exhibited the highest number of platelet EVs in all subpopulations (p < .05).

  • Small EVs subpopulations (differentiated by the tetraspanin signatures) could discriminate patients with low D-dimer (p < .001) from healthy controls.

  • Differences between the two D-dimer groups were seen in the platelet-derived (large and medium EVs and RCs), RBC-derived mEVs and l EVs and RCs, and lactadherin-positive large EVs and RCs (p < .05).

What is the impact?

• Platelet activation, reflected by increased EVs was associated with blood clotting disturbances. Small EVs signatures revealed changes in the EV subpopulations in association with blood clotting during COVID-19. Such signatures may enable identification of severely ill patients before the increase in coagulation is evident by coagulation parameters, for example, by high D-dimer.

GRAPHICAL ABSTRACT

Keywords:

• CD63
• CD81
• coagulation; COVID-19; CD41/61; CD9
• extracellular vesicles; subpopulation
• high sensitivity flow cytometry
• lactadherin
• SP-IRIS

Introduction

Severe COVID-19 infection causes frequent coagulation and hemostasis disturbances.¹ Common laboratory findings in severe COVID-19 infection include high plasma D-dimer and fibrinogen levels, leukocytosis with lymphocytopenia, thrombocytosis, and anemia.^2,3 There is limited data on the impacts of COVID-19 response on cell-mediated hemostasis. Platelets also participate in immune reactions by interacting directly with pathogens through membrane receptors, such as toll-like receptors and CLEC-2.⁴ Activated platelets release a heterogeneous population of extracellular vesicles (EVs), which are thought to mediate both inflammation and coagulation. Platelet-derived EVs are increased in a wide variety of inflammatory conditions, including autoimmune diseases, as well as cancer and viral infections.^5–7 Platelet-derived EVs have also been shown to increase in hospitalized COVID-19 patients, compared to COVID-negative patients.⁸

Despite the evidence of circulating EVs being promising biomarkers of variety of diseases, there is a lack of standardized analytical methodology. The small size and heterogeneity of EVs make them undetectable with traditional methods, including conventional cell flow cytometers.⁹ Development of the high sensitivity flow cytometry (hsFCM) has enabled the detection of single EVs and their accurate sizing. However, due to the weak refractive index of EVs, the most numerous population of EVs (<200 nm) remains poorly detectable.¹⁰ Single particle interferometric reflectance imaging sensor (SP-IRIS) offers a novel method for the detection of single small EVs (sEVs) in the range of 50–200 nm.¹¹ In addition to hsFCM and SP-IRIS, there are other emerging technologies available for single EV analysis, but they are unsuitable for clinical use due to their laborious and time-consuming nature.⁹

In this study, we aimed to explore the associations between coagulation markers and EV in COVID-19 inflammation. We used the current state-of-the-art single EV methods, SP-IRIS and hsFCM, and correlated these results with the commonly measured coagulation parameters. The platelet EV and residual cell (RCs) profiles in plasma were examined in the following size ranges: 50–200 nm for sEVs (SP-IRIS), 200–1000 nm for medium EVs (mEVs), and 1000–4000 nm for large EVs (lEVs) and RCs (hsFCM). Cellular source of EVs was also determined by hsFCM in the latter two subpopulations for leukocyte, erythrocyte, and endothelial markers.

Materials and methods

Patients and plasma samples

The study included patients (n = 50) hospitalized for COVID-19 infection during April–May 2020, the first wave of the COVID-19 pandemic in Finland. Surplus plasma samples (3.2%, 109 nM Na-citrate) from cohorted wards were collected for analysis with a storage time (including transportation) not more than 8 hours at room temperature before centrifugation. For the EV and thrombin generation (TG) studies, the plasma was centrifuged twice: after initial centrifugation at 2500g, 10 min, the supernatant was transferred to another tube and centrifuged again at 2500g, 10 min, after which the supernatant was stored at −80°C until further analysis. All analyses were done from thawed plasma and the plasma was not refreezed before analyses.

Coagulation data had been previously described in a wider cohort.² Patient data were anonymized, with informed consent waived by the institutional committee. No additional samples were taken and patients were not contacted during the course of the study. This study was conducted in line with the ethical principles of the Declaration of Helsinki and had received institutional study permit (HUS Diagnostics Centre, HUS/211/2020 §23, 1.7.2020).

EV analysis by high sensitivity flow cytometry (hsFCM)

The process of EV analyses using flow cytometry and SP-IRIS is presented in Figure 1. Details of the experiments including antibody clones, representative scatter plots, and gating strategy can be found in Supplementary File (S1). Twenty microliters of plasma was diluted (in 10 mM Hepes containing 140 mM NaCl, pH 7.2) and then incubated for 2 h at room temperature in the dark with 2–2.5 µl of the following antibodies: phycoerythrin (PE)-conjugated CD31, fluorescein isothiocyanate (FITC)-conjugated CD45, PE-conjugated CD61, and FITC-conjugated CD235a, or with 3.15 µl FITC-conjugated lactadherin. For full data on used clones, see Supplement Table S1. Protein aggregates were removed from lactadherin-FITC by centrifugation at 18 090 × g for 5 min at room temperature prior to use. Isotype-matched negative controls at the same concentration as the primary antibody were used to assess unspecific binding. After the incubation, the sample was diluted with 200 µl of HEPES buffer. All samples were measured on an A50-Micro (Apogee Flow Systems Ltd., UK) keeping the count rate between 1000 and 4000 to avoid swarm detection. The flow rate was calibrated using Apogee bead mix (Apogee Flow Systems Ltd.). Each sample was measured for 120 s, triggering on 405 nm side scatter. A Rosetta Calibration beads and software (Exometry, The Netherlands) were used to calibrate light scatter signals and to determine the diameter gates for the used flow cytometer as previously described.¹² A detectable EV diameter gate of 200–1000 nm was set for mEVs, in which the lower limit was set to exclude noise and the upper limit was to exclude lEVs and RCs (1000–4000 nm). Next, fluorescent gates were determined for every marker (CD61 for platelets, CD31 for endothelial cells, CD45 for leukocytes, CD235a for red blood cells, and lactadherin for phosphatidyl serine) using unlabeled EVs and isotype controls. CD31, commonly used to detect endothelial cells, is also expressed in platelets. Positive events (+) were defined as events with fluorescent signal exceeding the gate. The data analyses were performed with FlowJo^TM (v10.7.1; FlowJo, LLC), Prism 9.0 (GraphPad, USA), and R (R Core Team 2022. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/). The data are presented as concentrations, that is, the number of detected events corrected for the total sample dilution, flow rate, and measurement time. Free antibody/lactadherin-FITC in buffer, and buffer used for EV dilutions were used as additional controls according to the guidelines of MIFlowCyt-EV.¹³

Figure 1. High sensitivity flow cytometry (hsFCM) and single-particle interferometric reflectance imaging sensor (SP-IRIS) were used to isolate and characterize medium and large EVs (200–4000 nm) and small EVs (50–200 nm), respectively. EV, extracellular vesicle; SEC, size-exclusion chromatography; RC, residual cell.

EV characterization by SP-IRIS

According to the manufacturer’s recommendation, the EVs were isolated from the plasma with size exclusion chromatography (SEC). Briefly, 100 µl of plasma was loaded on to the SEC column (qEVsingle/70nm, IZON Science, USA) and eluted with dPBS. Four fractions of 200 µl (fractions 1–4) were collected after the void volume (1 mL) with automated fraction collector (IZON Science, USA). The fractions containing EVs were combined, and the counts of EVs were measured by nanoparticle tracking analysis (NTA) using NanoSight LM14 instrument equipped with blue laser (405 nm, 70 mW) and a sCMOS camera (Malvern Instruments Ltd., UK). The samples were diluted in dPBS filtered with 0.1 µm filter to obtain 40–100 particles/frame, and five 30 s videos were recorded with camera level 14. The data were analyzed using NTA 3.0 software (Amesbury, UK) with detection threshold 4 and screen gain at 10.

The isolated EV samples were then analyzed with SP-IRIS using ExoView^TM Plasma Tetraspanin kit and an ExoView^TM R100 scanner (NanoView Biosciences, USA) according to the manufacturer’s instructions. The samples were diluted at desired, optimized particle count (10⁹ particles/mL) based on NTA measurement using incubation buffer provided in the kit. All samples (35 µL of each) were added directly to the chip and incubated at room temperature for 16 h. The samples were then subjected to immunofluorescence staining using fluorophore-conjugated antibodies (CD9/CD63/CD81, provided in the kit), washed, dried, and scanned. The data obtained were analyzed using the NanoViewer analysis software (NanoView Biosciences) version 3.0 with sizing thresholds set from 50 to 200 nm diameter.

Coagulation analyses

Coagulation tests D-Dimer (HaemosIL D dimer HS 500, FEU units, Instrumentation Laboratory), von Willebrand factor activity (VWF:Act) and antigen (VWF:Ag), as well as blood counts and C reactive protein (CRP) were performed using routine methods in clinical laboratory of HUS Diagnostic Center (Helsinki, Finland). The instruments were as follows: coagulation tests with ACL TOP® 500 and 750 analyzers (Instrumentation Laboratory, Naples, Italy); blood count with Sysmex® XN-9000 (Kobe, Japan); and CRP with Siemens Atellica® (Siemens Healthineers, München, Germany). These have been previously described in a larger cohort.² In addition, a chemiluminescent ADAMTS13 assay and VWF collagen binding assay (both Acustar®, Werfen, Barcelona, Spain) were performed. TG was analyzed using the calibrated automated thrombogram (CAT® Thrombinoscope, Diagnostica Stago, Asnieres, France). TG was done with 5 pM tissue factor activator, without thrombomodulin addition.

High D-dimer of over 1.5 mg/L is considered a significant elevation in COVID.¹⁴ This threshold was used to discriminate patients into high D-Dimer (over 1.5 mg/L), severe COVID-19 patients to lower D-dimer (up to 1.5 mg/L), less severe patients. For coagulation tests, local reference intervals are provided. In the EV analyses, 10 plasma samples from non-matched healthy controls were also analyzed to provide a reference. In TG assay, normal pooled plasma was run in parallel to patient samples to provide a reference of normal TG profile.

Statistical analyses

The comparisons between D-dimer groups were done in pairwise comparisons, using the Mann–Whitney U test, with Bonferroni correction. Analyses were done with R software (R Core Team (2022). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/).

Results

Summary of the coagulation and basic laboratory test results is shown in Table I and Supplement Table S2. Coagulation markers D-dimer, fibrinogen, FVIII, and VWF:Act were elevated in most patients with medians already above the reference intervals. Since D-dimer has been identified as one of the clinical biomarkers relating to the COVID-19 severity, we used it to divide patients in the groups of severe (D-dimer over 1.5 mg/L) and less severe (D-dimer up to 1.5 mg/L) COVID-19, for the subsequent EV comparisons. For D-dimer, a limit three times above reference interval was used, that is, 1.5 mg/L. D-dimer exceeded 1.5 mg/L in 15/50 (30%) of the patients. Most, 43/46 (96%), patients who had inflammatory marker CRP measured, had elevated levels (median 60, interquartile range 21–114 mg/L), while CRP was very high, >100 mg/L in 15/50 (30%) patients. There was severe anemia with hemoglobin <100 g/L in 9/50 (18%) patients.

Table I. Summary of the laboratory test results of the 50 patients with COVID-19 infection. Results between high and low D-dimer groups are presented in Supplemental Table S2 and statistically significant differences in supplemental figure S3.

	Median	Range	Interquartile range	Reference range	NAs
PT (%)	99	23–133	90–106	70–130	1
APTT (s)	35	27–138	32–39	28–37	4
Fibrinogen (g/L)	5.8	0.4–11.5	4.3–7.4	2.0–4.0	0
FVIII (IU/mL)	162	35–489	113–239	60–160	2
Thrombin time (s)	22	16–37	19–24	17–25	7
VWF:Act (IU/dL)	256	10–563	129–305	50–190	0
VWF:Ag (IU/dL)	386	135–874	254–486	50–190	0
VWF:CB (IU/dL)	186	4–530	64–248	55–180	8
Antithrombin (IU/dL)	86	7–137	72–105	85–125	3
D-dimer (mg/L)	0.7	0.2–11.3	0.4–1.8	<0.5	0
Plt (×10^9/L)	294	93–694	215–434	150–360	1
Anti-Xa (IU/mL)	0.12	0–0.59	0.04–0.22	-	6
Hgb (g/L)	118	73–147	102–129	^1	2
Leuk (×10^9/L)	6.5	2.0–20.6	4.7–9.8	3.4–8.2	1
CRP (mg/L)	61	4–398	21–115	<4	2
ADAMTS13%)	72	19–140	42–92	60–130	4
CAT Peak (nM)	64.5	1.6–324.6	27–141	142–178^2	7
CAT ETP (nM*min)	1069.3	109–2173	573–1450	1140–1391^2	15
CAT LT (min)	5.6	2.7–41.5	3.9–7.2	3.00–3.67^2	7

¹Hgb ref interval 134–167 g/L for men, 117–155 g/L for women. ²range of CAT TG from normal plasma.

PT, prothrombin time; APTT, activated partial thromboplastin time; Fibr, fibrinogen; VWF:Act, von Willebrand factor GPIb activity; VWF:Ag, VWF antigen; Plt, platelet count; Hgb, hemoglobin; Leuk, leukocytes; CRP, C-reactive protein; VWF:CB, von Willebrand factor, collagen binding assay; CAT ETP, thrombin generation with CAT®, endogenous thrombin potential; CAT Peak, peak thrombin generation, CAT LT, Lag Time of thrombin generation; SD, standard deviation; NAs, samples with no results

Large and medium EVs

In hsFCM, the lEVs/RCs positive for CD61, CD235, CD31, CD45, and lactadherin were all elevated in the high D-dimer group compared to the low D-dimer group and controls (Figure 2). When mEVs were examined, the number of CD61+ and CD45+ EVs was significantly elevated in the high D-dimer group in comparison to the low D-dimer group, but not in comparison with controls (Figure 3).

Figure 2. Residual cells/large extracellular vesicles (1000–4000 nm) and medium-sized extracellular vesicles (200–1000 nm) measured by high sensitivity flow cytometry. Markers for red blood cells (CD235a), endothelial cells (CD31), leukocytes (CD45), platelets (CD61), and phosphatidyl serine exposure (lactadherin) were measured. Significant differences between low D-dimer and control groups compared with the high D-dimer >1.5 mg/L are marked with asterisks: *p < .05, **p < .01, ***p < .001. One outlier sample, with particle counts exceeding 6 × 10⁹/L is not shown. Pairwise comparisons were determined using Mann–Whitney U test, with Bonferroni correction. Box plots show median, 25th and 75th percentiles and range.

Figure 3. Particle counts and diameters of small EVs 50–200 nm by SP-IRIS (exoview R100). CD41a+, CD63+, CD81+, and CD9+ EV counts were all significantly higher in the high D-dimer group >1.5 mg/L compared to low D-dimer (≤1.5 mg/L) or controls. In terms of particle diameter, CD9+ particles were significantly larger in high D-dimer patients in comparison to low D-dimer patients. Significant differences between low D-dimer and control groups compared with the high D-dimer >1.5 mg/L are marked with asterisks: *p < .05, **p < .01, ***p < .001. Three outliers, with particle count exceeding 3 × 10⁵ are not shown. Pairwise comparisons were done using Mann–Whitney U test, with Bonferroni correction. Box plots, show median, 25th and 75th percentiles and range.

Lactadherin+ EVs, indicating PS+ procoagulant surfaces, were also increased in the high D-dimer group compared to the low D-dimer group, but again, not when compared to controls.

Small EVs

When samples were analyzed for sEVs with SP-IRIS, all capture spots (CD41a, CD9, CD63, and CD81) showed increased particle counts in the high D-dimer patient samples, when compared to the low D-dimer group and controls. In SP-IRIS, the mean sizes of the antibody-captured particles ranged between 50 nm and 65 nm (Figure 3). Only the size of the CD9+ sEV differed between the low D-dimer and high D-dimer groups, otherwise, no statistically significant differences were observed in the sEVs in terms of particle diameters. The difference in the platelet-derived EV subpopulations (s-, m-, and l) between the high and low D-dimer groups could not be explained by increased platelet counts since there were no difference in the EV counts or diameters between samples where platelet count exceeded 360 × 10⁹/L (upper limit of reference interval) and those where the platelet count was below 360 × 10⁹/L. Neither did platelet counts correlate with EV counts (R² = 0.07).

Small EV colocalization profiles

SP-IRIS also allowed the analysis of the common EV marker profiles in the platelet EVs based on the colocalization of the tetraspanins CD9, CD63, and CD81 on the CD41a+ sEVs. In the colocalization data, only CD41a/CD81+ and CD41a/CD63/CD9+ tetraspanins were different between high and low D-dimer groups (Figure 4). Instead, elevated subpopulations of CD41a/CD63+, CD41a/CD81+, and CD41a/CD63/CD9+ were detected in patients with low D-dimer when compared to the healthy controls (Figure 4). When observing CD63+ and CD9+ colocalization, there were no differences between low and high D-dimer groups, but similar finding between controls and low D-dimer levels was found as in CD41a+ colocalization. With CD81+ colocalization, no groups differed from one another.

Figure 4. SP-IRIS tetraspanin colocalization analysis with platelet marker CD41a used as the capture antibody, as well as colocalizations with CD63, CD81, and CD9. Results are shown in stacked boxplots as proportions of bound antibodies. Significant differences between controls and low ≤1.5 mg/L and high >1.5 mg/L D-dimer groups are marked with white asterisks: *p < .05, **p < .01, ***p < .001. Significant differences between the high and low D-dimer groups are marked with red asterisks: *p < .05, **p < .01 three outliers, with particle count exceeding 3 × 10⁵ are not shown.

Effects of anemia and von Willebrand factor

In SP-IRIS and hsFCM, no differences were seen in the EV particle counts or diameters when patients were grouped by moderate anemia with no anemia (hemoglobin concentration below 100 g/L) or severe inflammation (CRP over 100 mg/L) with normal CRP-levels (data not shown).

Another way to group the patients is to inspect the VWF involved in primary hemostasis. VWF-ADAMTS13 axis is known to be involved in primary hemostasis, and disturbances in the ADAMTS13 levels may lead to thrombosis (e.g., thrombotic thrombocytopenic purpura). In our data, VWF activity correlated modestly with VWF antigen (R² = 0.52) and with VWF collagen binding assay (R² = 0.38), as expected, but there was no correlation with ADAMTS13 and vWF activity (R² = 0.01) or antigen (R² = 0.11; Supplemental Figure S2). The EV profile was similar when using VWF:Act 200 IU/dL as a cutoff as it was with D-dimer 1.5 mg/L (data not shown).

Expectedly, in samples where D-dimer exceeded 1.5 mg/L, VWF activity and platelet counts were higher and ADAMTS13 lower than in the other patient samples (Supplemental figure S3). A clearly elevated TG peak (over 200 nM, while maximum of normal plasma was 178 nM) was observed in only 6/50 (12%) in functional TG assay measured by CAT®, while many patients exhibited diminished TG (Supplemental Figure S4).

Discussion

Early on, during the COVID-19 pandemic, high D-dimer (cutoff 1.5 mg/L) was recognized as a risk factor for intensive care unit admission and mortality.³ In our study, 30% of COVID-19 patients exceeded the 1.5 mg/L cutoff, and they were considered as severely ill in the following comparative EV analyses. While the pathological coagulation disturbances in the context of COVID-19 are well known,¹⁵ the role of cell activation, and the blood cell-derived EVs are hitherto not well described.

Both sEVs (and lEVs) were increased in the severely ill COVID-19 patients compared to the healthy controls, while in mEVs only the severely ill COVID-19 patients were different from the mEVs from COVID-19 patients with milder disease, but not from controls. We studied mEVs derived from platelets, endothelial cells, erythrocytes, and leukocytes. All EV subpopulations, but especially the platelet-associated (CD61+) and procoagulant (lactadherin+) residual platelets and EVs, were increased in the COVID patients with high D-dimer values.

Platelet sEVs captured with platelet-specific CD41a antibody and the prevalent EV tetraspanins CD63, CD81, and CD9 were all more abundant in severely ill COVID-patients than in controls. Flow cytometric studies have shown elevated numbers of platelet-associated EVs expressing CD41 in hospitalized COVID-19 patients compared to other hospitalized patients and healthy controls^8,16 Notably, our data demonstrate that COVID-19 patients with D-dimer less than 1.5 mg/L most often expressed similar EV profile than controls. Indeed, it seems that alterations in EV profiles are most pronounced in the severe cases. This is somewhat contradictory to some earlier studies, where hospitalized patients with milder COVID-19 infection had higher CD41+ EV levels than those with severe illness.^16,17 Interestingly, platelet count was not associated with the EV counts, neither did hemoglobin concentration, despite the more severe patients being frequently anemic.² Platelet-derived EVs are the most numerous EV type in healthy individuals, and they have been described to be significantly upregulated in both bacterial sepsis, as well as in Dengue virus infection.^18,19 As the EV counts did not correlate with platelets, platelet-derived lEVs/residual platelets may be a population of activated platelets with altered shape, that is, they are not detected in the blood count analyzer. In these sEV colocalization with CD41a, CD63, and CD9, the most remarkable differences in particle colocalization patterns were seen between patients with low D-dimer and controls, while controls and patients with high D-dimer were closer to one another. This is in concordance with a previous study, where colocalization with SARS-CoV-2-S peptide was more abundant with CD41a+ and CD9+ sEVs in mild COVID-19 patients than those with severe infection.¹⁷ Platelet contribution seems important and smaller EVs might be more prevalent in the earlier stages of the disease, offering a putative marker for disease severity. The distribution of EVs in the vasculature might vary as well, for example, COVID-19 has direct effects on the pulmonary endothelium, with increased VWF staining in histological samples²⁰ and platelet-derived EVs have been shown to become deposited into forming fibrin.²¹

A key mechanism of COVID-19 infection is SARS-CoV-2 binding to angiotensin-converting enzyme (ACE) receptors.²² Platelet activation may be a link between this mechanism and increased risk of thrombosis in severely ill COVID-19 patients. Interestingly, SARS-CoV-2 mRNA have been found in the platelets of some COVID-19 patients, but the clinical relevance of this finding is unclear.^16,23 There are some discrepancies between studies, but some groups have shown ACE2-receptors on platelets, which could be a mechanism for infection. It has also been postulated that platelet-derived EVs may interact with SARS-CoV-2, which may lead to platelet hyperreactivity, exacerbating risk of thrombosis.^24–26 In spiking experiments, SARS-CoV-2 spike protein can directly activate platelets .²⁷ In contrast, endothelial cells readily express ACE2 and are a major target for COVID-19 infection. Endothelial dysfunction, in turn, causes platelet activation, enhanced coagulation, and fibrin turnover, with a hallmark of elevated D-dimer concentration. It is a well-known component of many thromboinflammatory conditions, including COVID-19 and bacterial sepsis^28,29 The beneficial effects of heparin on the disease prognosis may be in part explained by its inhibitory effect on the ACE2-receptor binding.^28,30

Other primary hemostasis factors were also studied. VWF-ADAMTS13 axis is involved in the thrombotic complications of inflammation as shown in thrombotic microangiopathies and COVID-19.^31,32 Platelets and platelet-associated EVs have their effect in the endothelial surface. In patients with systemic inflammation, endothelial cells release the contents of Weibel Palade bodies, and they differ from classic thrombotic thrombocytopenic purpura in that ADAMTS13 level may decrease only mildly or remain in the normal range. Instead, ADAMTS13 activity is inhibited by thrombospondin-1, released from the alpha-granules of platelets.³³ ADAMTS13 may be decreased and ADAMTS13 levels tend to decrease with increasing disseminated intravascular coagulation risk.³⁴ In our study, concordantly with previous studies, both low ADAMTS13 activity and high VWF were seen in many patients, but ADAMTS13 levels remained mostly within the reference interval and did not correlate with VWF activity or antigen.³² A large series of hospitalized COVID-19 patients showed lower ADAMTS13 level and higher VWF antigen in non-survivors compared to survivors.³⁵ In this regard, COVID-19 behaves similarly as other inflammatory processes, and it may be possible that these findings might be generalized for other inflammatory diseases as well.

The main limitation of this study is the lack of clinical data on the patients, due to the anonymized study design, precluding prospective follow-up of the studied markers over the disease course. The samples were not taken at standardized time intervals after start of symptoms due to this design. This, in turn, mandated the use of D-dimer biomarker for disease severity characterization, which may not capture all severe cases. The limited time of the sample collection (April–May 2020), during which time there were no significant differences or changes in practice in the treatment of COVID-19 patients in Finland, strengthens our data. During this time, all hospitalized patients received low molecular weight heparin thromboprophylaxis. However, the unique clinical situation of the patients affects heparin dosing, which may influence our coagulation test results. This might, in turn, explain the paradoxically low thrombin generation in some patients. Combining hsFCM with appropriate state-of-the-art calibration and reporting for EV analytics (MIFlowCyt) and the use of the novel SP-IRIS analysis to be able to quantitate and characterize sEVs abled us to have a more profound view of the wide spectrum of the circulating vesicle populations. The elevated levels of platelet-derived EVs were seen with both methods, and included EVs of all sizes, warranting analysis of platelet-derived EVs in context of COVID-19 pathogenicity. Multiple groups have reported similar findings in COVID-19 patients.^8,36,37

Importantly, the ability to predict disease progression into a very severe form, would require new prognostic tests. Here, we observed a promising association of the elevated CD41a/CD81+ only in the sEV subpopulation with milder COVID-19 which discriminated the healthy controls from patients with lower level of coagulation activation evidenced by lower D-dimer level. As this subpopulation of platelet EVs is virtually unanalyzed in the clinical context, it warrants further study in cohorts where the clinical disease progression is known unlike in the current early patient cohort. So far, the sEV subpopulation has evaded inspection due to absence of accurate detection methods with a lower limit of detection in the sEV size range. In future, this population may reveal a novel information source due to the sensitivity of platelet activation in many diseases.⁷

As shown in previous studies by flow cytometry, lEVs/RCs positive for all markers were elevated in the severe COVID-patients, but the largest contributions were only seen in EVs and residual platelets with CD61 and lactadherin, marking platelets and phosphatidylserine activation. As COVID-19 is now recognized to be a vascular disease with major damage to microcirculation, association of also these EV/cell subpopulations to disease severity may offer interesting prognostic biomarker candidates in future studies provided that the flow cytometry, or better hsFCM will be calibrated for size determination to enable multicenter studies.

In conclusion, our data suggest that platelet-derived and phosphatidylserine-positive EVs are associated with more severe COVID-19 infection, differentiated by high D-dimer levels. Yet, small EVs exhibit most remarkably different subpopulations in patients with milder COVID-19 infection. These signature changes in EVs bring forth our understanding of the pathogenicity of COVID-19 and highlight the role of platelets and primary hemostasis during this disease.

Acknowledgement

We acknowledge the EV Core (University of Helsinki) for the flow cytometry and SP-IRIS EV analyses.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/09537104.2024.2313362.

Additional information

Funding

This work was supported by the HUS Diagnostic Center Research grant [Y78002220]; Academy of Finland grants [330486, 337641]; Business Finland [Extracellular Vesicle Ecosystem (EVE) for Theranostic Platforms grant [1842/31/2019]; Finnish Red Cross Blood Service Research Fund Magnus Ehrnrooth Foundation; Medicinska Understödsföreningen Liv och Hälsa rf (PRMS); Helsinki University Library.