All patients that were included into this study were admitted to our clinic due to acute respiratory infectious disease. COV^pos had to be SARS-CoV2 positive confirmed by polymerase chain reaction (PCR) testing. COV^neg suffered from pneumonia or infect-triggered acute exacerbation of COPD and had to be SARS-CoV2 negative confirmed by PCR. Individuals had to be at least 18 years old and did not suffer from a known hematological or hemostatic disease, coagulopathy, or acute bleeding event. Patients form ICU or normal floor were eligible. Dual antiplatelet therapy was prohibited. Patients were recruited between May 2020 and May 2021. Routinely clinically assessed blood values were determined by the hospital laboratory (Labor Berlin, Berlin, Germany). Mean platelet volume is assessed by the routine diagnostics lab by impedance-based particle counting. Within the same measurement, counts and size of platelets, as well as erythrocytes and leukocytes, are assessed in diluted samples and without lysis. Leukocyte counts are assessed in a separate measurement after erythrocyte lysis. The study was approved by the local ethics committee (EA2/066/20, EA4/147/15). The study was conducted in compliance with the 1964 Declaration of Helsinki and its amendments and the Principles of Good Clinical Practice by the International Council for Harmonization 1996.

Blood was drawn from the cubital veins using ethylenediaminetetraacetic acid (EDTA) (3 ml, Vacurette^®, Greiner Bio-One, Kremsmünster, Austria), citrate (3 ml, 3.2% sodium citrate, Vacurette^®, Greiner Bio-One, Kremsmünster, Austria), and hirudin tubes (Sarstedt-Monovette^®, Sarstedt, Nümbrecht, Germany). Whole blood was separated for the experiments requiring plasma by centrifugation (1,200×g, 10 min, room temperature) and stored at −80° for further analysis.

In hirudinized whole blood which was previously diluted with 0.9% sodium chloride, platelet’s reactivity to TRAP (32 μmol/l), ADP (6.4 μmol/l), and AA (0.5 mmol/l) was measured by multiple electrode aggregometry (MEA; Multiplate^® Analyzer; Roche, Germany; regents also by Roche) not later than 3 h after sampling. According to the manufacturers’ instructions, a measurement time of 6 min was set, and the area under the curve was calculated and translated to Multiplate^®-specific units. The exact method was reported previously (23, 24).

Flow cytometric characterization of leukocytes was performed as previously established (25). Within 1 h of collection, 100 µl of EDTA-anticoagulated whole blood was added to a master mix consisting of 100 µl fluorescence-activated cell scanning (FACS) staining buffer (BioLegend, San Diego, CA, USA) and 2 µl of each of the following antibodies: anti-human CD14-Pacific Blue™, CD16-Brilliant Violet 510™, CD4-Brilliant Violet 605™, CD45-Brilliant Violet 711™, CD3-Alexa Fluor 488™, CD26-PE, CD19-PE/Dazzle 594™, CD8-PE/Cyanine7, and CD41-Alexa Fluor 647™ (all BioLegend, San Diego, CA, USA). The mixture was incubated in the dark at room temperature for 30 min. Stained samples were fixed by adding 800 µl of 0.5% paraformaldehyde in phosphate-buffered saline (Sigma-Aldrich, St. Louis, MO, USA) and kept at +4°C in the dark. Samples were acquired on an Attune NxT Acoustic Focusing Cytometer (ThermoFisherScientific, Waltham, MA, USA) within 3 days. Stability of signal detection was verified and documented by daily measurements of Attune Performance tracking beads. Kaluza version 2.1 software (Beckman Coulter, Brea, CA, USA) was used for gating. Scatter parameters were interpreted as measures of morphological features, i.e., side scatter as a measure of granularity and forward scatter as a measure of cell size.

The cytokines IL-1RA, IL-2, IL-6, IL-7, IL-10, monocyte chemotactic protein 1 (MCP-1), chemokine (C-C motif) ligand (CCL) 3, chemokine (C-X-C Motif) ligand (CXCL) 8, CXCL10, interferon (IFN)-α2, IFN-γ, granulocyte colony-stimulating factor (GCSF), and tumor necrosis factor alpha (TNF-α) were determined in platelet-depleted plasma samples using the “COVID-19 Cytokine Storm Panel 1” 13-plex bead array (BioLegend, San Diego, CA, USA) according to the manufacturer’s instructions. Data were acquired on the flow cytometer mentioned above and analyzed using Kaluza version 2.1 software.

Thrombin-antithrombin complex (TAT) ELISA (AssayPro, St. Charles, MO, USA) was performed according to manufacturer’s instructions and measured using the plate reader Tecan Infinite 200Pro (Tecan Group, Maennedorf, Switzerland).

All reported probability values are two sided, and a value of p < 0.05 was considered statistically significant. Median values and quartiles are reported, and nonparametric tests were used if not stated otherwise. If not stated differently, Mann–Whitney U test or Chi-squared test were used to evaluate differences between two groups. Correlations were calculated with Spearman’s test. For network building p < 0.01 and r > 0.3 or r < −0.3 were used as cutoffs for showing connections. Statistics were calculated using SPSS Statistics, version 27 for Windows and macOS (IBM, Armonk, NY, USA) and R version 4.1.0 (2021-05-18). Graphs and networks were plotted using R within R Studio version 1.3.1093 (RStudio PBC, Boston, MA, USA).

More patients in the COV^neg group had chronic obstructive lung disease (Table 1). More patients had received prophylactic and less received intermediate-dose anticoagulation in the COV^neg than COV^pos cohort (Table 2). There was no difference in the frequency of patients receiving therapeutic dose anticoagulation (Table 2). Administration of oral glucocorticoids and inhalative bronchodilators was more common in COV^pos (Table 2). In COV^pos, 10 out of 50 individuals died compared with none in COV^neg patients. The sequential organ failure assessment (SOFA) score was calculated for all patients with sufficient data (COV^pos n = 33, COV^neg n = 21). Laboratory parameters did not differ significantly between COV^neg and COV^pos patients (Table 3). Four COV^pos patients and two COV^neg patients had thrombocytopenia as defined by <150 thrombocytes/nL. No patient participating on our study had received vaccination against SARS-CoV2 prior to study participation.

Platelets from COV^pos patients exhibited higher TRAP-induced aggregability than platelets from COV^neg patients (Figure 1A). As TRAP is a strong coagulation stimulus, this observation points to the hyperaggregability of platelets under stimulation with a thrombin-substitute in COVID-19. ADP-induced platelet aggregability was also increased COV^pos as compared with COV^neg patients (Figure 1B), although ADP is a much weaker platelet agonist than TRAP. Additionally, also AA-induced platelet aggregation was higher in COV^pos than COV^neg patients (Figure 1C). No significant differences between COV^pos and COV^neg patients were observed for platelet counts or MPV (Supplemental Figure 1).

Since TAT is a surrogate parameter for thrombin generation and thrombin is the strongest known platelet activator (26), TAT was quantified in both patient groups. Higher TAT levels were measured in COV^pos than COV^neg patients, reflecting the linkage between higher platelet reactivity and activated coagulation system (Figure 2).

In our study, SOFA score values differed significantly between COV^pos and COV^neg (Figure 3A) and between COV^surv and COV^non-surv patients (Figure 3B). MPV was higher in COV^non-surv compared with COV^surv patients (Figure 3C). MPV- (Figure 3D) and TRAP-6-induced platelet aggregation (Figure 3E) correlated positively with the SOFA score values only in COV^pos but not in COV^neg patients. These findings point to an important role of platelet function for the clinical prognosis of patients suffering from COVID-19. Instead, levels of individual leukocyte subtype aggregates with platelets did not correlate with SOFA score in COV^pos or in COV^neg patients.

We observed more than 2-fold higher plasma levels of IL-6, IL-7, IL-10, IL-1RA, MCP1, and CXCL10 in COV^pos as compared with COV^neg patients, pointing to the severe cytokine burst (Figure 4A). The cytokines IL-2, CXCL8, IFN-α2, IFN-γ, GCSF, and TNF-α were less than 2-fold higher in COV^pos in comparison with COV^neg patients (Figure 4A). Moreover, a higher percentage of CD4^pos T_H lymphocytes, CD14^hiCD16^neg classical monocytes, and CD14^hiCD16^pos intermediate monocytes formed aggregates with platelets in COV^pos as compared with COV^neg patients (Figure 4A and Supplemental Figure 2). The granularity of CD14^loCD16^pos nonclassical monocytes was higher and the granularity of CD19^pos B lymphocytes was lower in COV^pos in comparison with COV^neg (Figure 4A and Supplemental Figure 2), potentially indicating a differential activation state of these cell types.

Within the COV^pos group, we also compared our panel of inflammatory parameters between COV^surv and COV^non-surv patients (Figure 4B). Values of IL-6, IL-1RA, CXCL8, MCP1, and CXCL10 were over 2-fold higher in COV^non-surv than in COV^surv. Moreover, lower relative lymphocyte abundance and higher absolute leukocyte count, IL-2, and CCL3 levels were observed in the COV^non-surv versus COV^surv patients. In COV^non-surv patients, the granularity of CD8^pos cytotoxic T cells as well as the percentage of T_H cells forming aggregates with platelets were lower than in COV^surv patients.

Correlations between the SOFA score, parameters of platelet function, and inflammation are visualized by networks for patients with COVID-19 (Figure 5A), with other respiratory disease (Figure 5B), and for parameters correlated in both cohorts to elucidate common effects (Supplemental Figure 3).

In COV^pos patients, the SOFA score correlated with CRP at the time of blood sampling, peak CRP level during hospital stay, leukocyte count, and MPV- and TRAP-induced platelet aggregation. Peak CRP level during hospital stay was strongly correlated with platelet aggregation after stimulation with TRAP-6 and AA (Figure 5A). Moreover, IL-6 was correlated with MPV, platelet aggregation after stimulation with TRAP-6 and AA in COV^pos but not in COV^neg patients (Figures 5A, B). In contrast to COV^pos, only peak CRP but no platelet function marker correlated with the SOFA score in COV^neg patients (Figure 5B). Correlation between SOFA and peak CRP was much stronger in COV^pos than in the COV^neg group (Supplemental Figure 3). These findings support a link between characteristic markers of inflammation and platelet reactivity and SOFA score for patients with COVID-19 but not for those with other acute respiratory diseases. This panel of characteristic markers is associated with the clinical outcome in COVID-19.

In both patient groups with respiratory symptoms, platelet aggregation upon stimulation with ADP or with AA correlates with the absolute number of platelet aggregates formed with B lymphocytes and with NK cells, respectively (Supplemental Figure 3).

We demonstrated higher platelet reactivity measured by MEA in COVID-19 compared with patients with other acute respiratory disease. Less ADP- and TRAP-induced aggregation (20) and no significant differences for TRAP- and AA-induced aggregation compared with healthy volunteers or reference ranges (19) have previously been reported in patients hospitalized or receiving ICU-level care. Another study demonstrated a predictive value of ADP- and TRAP-induced platelet activation for the duration of the hospital stay (20). However, in these studies, COVID-19 patients were compared only with healthy controls but not with diseased individuals. In our study, platelets of patients with COVID-19 reacted stronger to TRAP, ADP, and AA compared with those patients suffering from other pulmonary infections. In linkage between coagulation system and platelet hyperreactivity, we found TAT to be higher in COV^pos than in COV^neg patients. Elevated thrombin level may explain the increased platelet reactivity as well as the higher numbers of platelet-leukocyte complexes despite antiplatelet therapy in patients with COVID-19.

Comparing COV^surv and COV^non-surv groups, we observed a higher MPV in COV^non-surv, pointing to a hyperreactivity of platelets in vivo (27). A retrospective analysis including patients hospitalized for COVID-19 also suggested that MPV relates to the clinical outcome (28). The SOFA score values predict disease severity and mortality in patients with COVID-19 (29, 30). Our data additionally link TRAP-induced platelet aggregation and MPV as markers of platelet reactivity with SOFA score. This once again stresses the important relation between hyperreactive platelets and disease severity.

In patients with COVID-19, thrombocytopenia has been shown to be associated with a worse clinical outcome (31). In our patients, we found four COV^pos and two COV^neg individuals presenting thrombocytopenia at the time point of platelet function measurement, reflecting the disease severity of our patient cohort.

Cytokines typically associated with COVID-19-induced “cytokine storm” were mostly higher in COV^pos compared with COV^neg, verifying earlier reports and also in comparison with patients with non-COVID-19 respiratory diseases (32–34). Regarding disease severity, we observed higher levels of IL-6, IL-1RA, MCP1, CXCL8, and CXCL10 comparing COVID survivors and nonsurvivors. MCP1 and CXCL10 as chemotactic agents for monocytes and CXCL8 and CXCL10 as monocyte-derived cytokines point to the central role of monocytes in this setting (35). IL-6 as cytokine amplifier is known to play a prognostic role in COVID-19 (35). IL-1RA antagonizes IL-1 and can also be derived from monocytes (36). Plasma values of IL-1RA correlate with a worse clinical prognosis in COVID-19 (36). Within our COVID cohort, nonsurvivors were characterized by higher cytokine levels, especially IL-1RA, IL-6, MCP-1, CXCL8, and CXCL10 than patients who survived COVID-19.

In COVID-19, monocytes have been shown to release procoagulant proteins in a platelet-dependent manner (22). In line with other studies, we here demonstrate differences in monocyte-platelet conjugates in comparison with patients with acute respiratory diseases other than COVID-19. We found a higher proportion of classical and intermediate monocyte-platelet aggregates in COV^pos compared with COV^neg. These findings are in line with data of other studies, which also demonstrated more platelet aggregates with monocytes in patients with COVID-19 compared with healthy individuals (15, 37) or not further specified patients as controls (22). This points to the importance of the innate immune system interacting with platelets in COVID-19.

In this study, more T_H lymphocytes formed aggregates with platelets in COV^pos than in COV^neg patients. These conjugates seem to have proinflammatory effects in autoimmune neuroinflammation (38). Further research is needed to elucidate the interplay between T_H cells and platelets, explaining its relevance for COVID-19.

Granularity of B cells was lower in COV^pos than COV^neg, potentially indicating release of synthesized proteins—presumably immunoglobulins—from B cells surpasses synthesis within the cell. Similarly, granularity of CD8^pos T_cyt lymphocytes was lower in COV^non-surv than COV^surv, also potentially reflecting a predominant degranulation in these patients, who died from COVID-19 (39). The lower relative lymphocyte abundance and higher leukocyte count in those patients who died confirm previous findings (40).

A positive correlation was observed between the SOFA score as marker for disease severity and CRP, leukocytes, and markers of platelet hyperreactivity in patients with COVID-19 only but not in those with acute respiratory disease of other reason. The peak CRP level correlated with AA- and TRAP-induced platelet aggregation and IL-6 with MPV, reflecting an interplay between platelet hyperreactivity and inflammation. Importantly, none of these relations could be demonstrated in COV^neg patients.

Since platelet hyperreactivity contributes to worse clinical outcome in COVID-19, it is tempting to speculate that antiplatelet-directed therapies would improve the clinical prognosis. In a pandemic retrospective analysis, aspirin administration in patients with COVID-19 reduced the risk for mechanical ventilation, ICU admission, and in-hospital mortality (41). Within the RECOVERY trial, the aspirin group did show a significant reduction in hospital stay duration, thromboembolic events, and percentage of patients who had been discharged alive compared with best medical care in COVID-19 (42). These clinical data are in line with our experimental findings and highlight the clinical impact of platelet hyperreactivity on the clinical outcome, pointing to the importance of antithrombotic therapy in COVID-19.