Neutrophils predominate the immune signature of cerebral thrombi in COVID-19 stroke patients

Coronavirus disease 2019 (COVID-19) is associated with an increased risk of thrombotic events. Ischemic stroke in COVID-19 patients entails high severity and mortality rates. Here we aimed to analyze cerebral thrombi of COVID-19 patients with large vessel occlusion (LVO) acute ischemic stroke to expose molecular evidence for SARS-CoV-2 in the thrombus and to unravel any peculiar immune-thrombotic features. We conducted a systematic pathological analysis of cerebral thrombi retrieved by endovascular thrombectomy in patients with LVO stroke infected with COVID-19 (n = 7 patients) and non-covid LVO controls (n = 23). In thrombi of COVID-19 patients, the SARS-CoV-2 docking receptor ACE2 was mainly expressed in monocytes/macrophages and showed higher expression levels compared to controls. Using polymerase chain reaction and sequencing, we detected SARS-CoV-2 Clade20A, in the thrombus of one COVID-19 patient. Comparing thrombus composition of COVID-19 and control patients, we noted no overt differences in terms of red blood cells, fibrin, neutrophil extracellular traps (NETs), von Willebrand Factor (vWF), platelets and complement complex C5b-9. However, thrombi of COVID-19 patients showed increased neutrophil density (MPO+ cells) and a three-fold higher Neutrophil-to-Lymphocyte Ratio (tNLR). In the ROC analysis both neutrophils and tNLR had a good discriminative ability to differentiate thrombi of COVID-19 patients from controls. In summary, cerebral thrombi of COVID-19 patients can harbor SARS-CoV2 and are characterized by an increased neutrophil number and tNLR and higher ACE2 expression. These findings suggest neutrophils as the possible culprit in COVID-19-related thrombosis.


Introduction
Coronavirus disease 2019 (COVID-19) is primarily characterized by pulmonary involvement but neurological manifestations [10,11,14,30,39] and thrombotic complications [17] are also frequent. Previous observations and pathological descriptions reported both large vessel and microvascular thrombosis, suggesting peculiar prothrombotic pathophysiology of COVID-19 [9,40]. Diverse pathophysiological mechanisms have been hypothesized to explain the prothrombotic state in COVID-19, namely direct viral invasion of vascular structures (such as endothelial cells) [37] or of blood cells, or as a consequence of the organ's immune response. An excessive inflammatory response to SARS-CoV-2 associated with the development of coagulopathy are thought to be the most significant features of poor prognosis in COVID-19 patients [19,36]. High D-dimer levels and coagulation abnormalities have been frequently reported in COVID-19 patients suggesting that the activation of the coagulation system may be involved as an effector pathway of the immune response to the virus. Angiotensin-converting enzyme 2 (ACE2)-the putative functional receptor for SARS-CoV-2 entry into host cells [15]-is mainly expressed in endothelial cells, macrophages and perivascular pericytes. Dysregulation of the angiotensin 2 (AngII)/angiotensin receptor type 1 (AT1R) pathway downstream of ACE2 could lead to cytokine release syndrome and severe endothelial dysfunction with consequent increased vascular permeability and uncontrolled inflammation, which could be implied in virus-specific thrombo-inflammatory mechanisms. Also, hypoxia, commonly observed in COVID-19 pneumonia, may induce a prothrombotic state by affecting coagulation and fibrinolysis pathways as well as endothelial and neutrophil functioning [29]. Furthermore, a hyper-inflammatory state with macrophage activation, hyperactivation of the myeloid compartment and cytokine storm has been observed in severe COVID-19 patients [33].
Ischemic stroke is not uncommon in patients with COVID-19, especially those with severe infection and pre-existing vascular risk factors [25]. A meta-analysis showed that acute cerebrovascular disease occurs in about 1.4% of the hospitalized COVID-19 population, with a prevalence of acute ischemic stroke over intracerebral hemorrhage [25]. Although the true relationship between COVID-19 and stroke incidence remains to be clarified, multicenter and meta-data suggest that ischemic stroke in COVID-19 patients is more severe with a worse functional outcome and higher mortality [21,27].
The analysis of cerebral thrombi retrieved by endovascular procedures in patients with acute ischemic stroke from large vessel occlusion (LVO-AIS) has emerged as a tool for investigating the diverse pathophysiological mechanisms that contribute to thrombus formation [4,34]. This may also apply to patients with LVO-AIS and concomitant COVID-19, whereby evaluating the composition of cerebral thrombi in terms of immune cells, endothelium and coagulation cascade components could provide new insights into the pathogenesis of SARS-CoV-2-related thrombosis and the association between ischemic stroke and SARS-CoV-2 infection.
In the present study, we analyzed cerebral thrombi retrieved after acute mechanical thrombectomy (MT) from AIS patients with COVID-19 and controls to provide a comprehensive description. We investigated, i) the expression of the SARS-CoV-2-docking receptor (ACE2) by cells within the thrombi, ii) the presence of SARS-CoV-2 within cerebral thrombi and iii) thrombus composition in terms of structural components (red blood cells, fibrin, von Willebrand Factor, platelets and complement) and the immune profile (neutrophils, macrophages, T and B-lymphocytes, and neutrophil extracellular traps [NETs] density).

Study population
The study was conducted on patients admitted to three comprehensive stroke centers in Italy and Switzerland (San Raffaele Hospital, Milan, Italy; San Gerardo Hospital, Monza, Italy; Lausanne University Hospital, Lausanne, Switzerland). From February 2020 to March 2021, we included prospective consecutive adult patients with COVID-19 and concomitant LVO-AIS treated with MT, and with a cerebral thrombus available for histological analysis. We defined a confirmed case of COVID-19 by the following criteria (± 15 days from the index event): 1.) clinical [8] and/or radiological [32,38] features suggestive of COVID-19 infection (i.e. fever, dry cough, myalgias, dyspnea, hypo/anosmia, hypo/ageusia; peripheral ground glass lung opacities on chest x-rays and/or on chest CT scan); 2.) positive result for SARS-CoV-2 on real-time reverse-transcription polymerase chain reaction analysis of nasopharyngeal swab specimens and/or SARS-CoV-2 positive serology. Clinical and radiological features were respectively assessed as part of the standard clinical practice from treating physicians and expert radiologists.
In the primary analysis we selected a control group of COVID-19-negative (control) patients from consecutive LVO-AIS patients admitted to the San Raffaele Hospital between July 2016 and November 2019, treated by MT and with a thrombus available for analysis. Patients in the control group were matched to COVID-19 patients for stroke etiology, administration of intravenous thrombolysis (IVT) and anti-thrombotic drug at the index event.
In a secondary analysis, a distinct set of thrombi of LVO-AIS patients with recent pre-stroke infections (present at stroke symptom onset) and not related to SARS-CoV-2 was analysed. Recent pre-stroke infections were defined as suggestive symptoms (i.e., cough, dyspnea, pleuritic pain, urinary tract symptoms, etc.), history of fever within the previous 7 days of stroke symptom onset and/or determination of body temperature ≥ 37.5 • C at admission and white blood cell count ≥ 11, 000/mL or ≤ 4, 000/mL, pulmonary infiltrate on chest X-rays or cultures positive for a pathogen at admission [31].
The local Ethics Committees (Lausanne Hospital, San Raffaele Hospital, San Gerardo Hospital) approved the study and informed consent was collected. The research was conducted in compliance with international and community guidelines including the Convention on Human Rights and Biomedicine, the Council Recommendation of Europe on the protection of health data and the Helsinki declaration of the World Medical Association on principles for research involving human subjects.

Clinical variables
For each included patient we collected: demographic data, vascular risk factors, prior anti-thrombotic therapy on admission, stroke severity (assessed by the National Institutes of Health Stroke Scale, NIHSS), administration of intravenous thrombolysis, early ischemic changes on non-contrast brain CT scan (assessed by the Alberta Stroke Program Early Computed Tomography Score, ASPECTS), vascular occlusion site, collateral grading score [35], details of the MT procedure (timings, number of maneuvers and type of device), degree of achieved reperfusion (by modified Treatment in Cerebral Infarction, mTICI scale), stroke etiology (according to the Trial of Org 10,172 in Acute Stroke Treatment (TOAST) [1] classification) and 3-month functional outcome (by modified Rankin Scale, mRS). Data from laboratory tests obtained up to 48 h from stroke symptoms onset were collected. Complete blood counts was assessed with an automated hemocytometer. We calculated the ratio between the absolute circulating neutrophil and lymphocyte count (blood neutrophil-to-lymphocyte ratio, NLR).

Thrombi collection
Each participating center immediately fixed cerebral thrombi retrieved during the MT procedure in 10% formalin and stored them at + 4 °C until processing. Formalin-fixed specimens were then embedded in paraffin and cut into 5 μm serial sections. Thrombus analysis was performed centrally at the San Raffaele Scientific Institute, Milan, Italy, and included molecular biology, histology and electron microscopy analyses (details in the dedicated sections below). Thrombus analysis was performed in accordance with the guidelines of the institutional biosafety committee, with the use of proper personal protective equipment (PPE).

Electron microscopy
For transmission electron microscopy (TEM), we fixed small representative parts of the thrombus by immersion in 2.5% glutaraldehyde in 0.12 M sodium phosphate buffer overnight, then rinsing in the same buffer and immersing in 1% osmium tetroxide in 0.12 M sodium phosphate buffer for 2 h. We dehydrated tissues by a graded alcohol series and then infiltrated them with EPON Resin. Ultrathin sections of 60-70 nm were cut at using a microtome. We assessed sections with a Talos 120C (Fei) electron microscope.

Molecular biology
To detect SARS-CoV-2 in fragments of formalin-fixed paraffin-embedded thrombi (n = 6), we extracted total RNA using the QIAamp DNA FFPE Tissue Kit (Qiagen) according to the manufacturer's instructions. For one thrombus (n = 1) it was possible to search for SARS-CoV-2 RNA using a not paraffin-embedded thrombus fragment; in this case, we extracted total RNA with the RNeasy Lipid Tissue Mini kit (Qiagen) using an elution volume of 50 μL. RNA concentration and quality were measured. We tested RNA samples for presence of SARS-CoV-2 virus using the forward primer (5ʹCAA GTG GGG TAA GGC TAG ACTTT-3ʹ) and reverse primer (5ʹ-ACT TAG GAT AAT CCC AAC CCAT-3ʹ) recognizing a 344 bp sequence of the RNA-dependent RNA polymerase (RdRp) gene present in all severe acute respiratory syndrome (SARS)-related coronaviruses [7]. We performed reverse transcription and subsequent amplification using the SuperScript ™ III One-Step RT-PCR System with Platinum ™ Taq DNA Polymerase (ThermoFisher Scientific). We analyzed PCR products by electrophoresis on 1.5% agarose gels and confirmed specificity for SARS-CoV-2 in a sample by Sanger sequencing. For this, we purified the amplicon using HT ExoSAP-IT (Thermo Fischer Scientific) according to the manufacturer's instructions and then performed sequencing using the BigDye Terminator kit v. 3.1 and cleaning with the BigDye XTerminator Purification Kit (Applied Biosystems Foster City, CA, USA). We analyzed purified products of the sequencing cycle on the ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems) and generated nucleotide sequences with SeqScape ® Software (ThermoFisher Scientific, Waltham, MA, USA).

Statistical analysis
We described categorical variables by frequencies and percentages, and continuous variables using mean and standard deviation (SD) or median with interquartile range (IQR

Clinical, radiological and laboratory characteristics
During the study period, we included seven COVID-19 patients with LVO ischemic stroke treated by MT (mean age 70.9 years [± 12.4]; 42.9% females). The median delay between the beginning of COVID-19 symptoms and stroke onset was 5 days (IQR 3-10). The most frequent COVID-19 symptoms were fever (85.7%), cough (42.9%) and dyspnea (28.6%). All seven patients had lung opacities typical of COVID-19 on pulmonary imaging (chest X-ray or CT scan). We included a control group of 23 LVO stroke patients without SARS-CoV-2 34.8% for controls; p = 0.018). No significant differences were found between the two groups in terms of blood leukocyte and blood neutrophil counts or the blood neutrophil-to-lymphocyte ratio ( Table 1).
In summary, we found that the thrombi of COVID-19 patients showed a higher number of ACE2 + cells compared to controls. In both groups, the ACE2 positive cells were predominantly monocytes/macrophages.

SARS-CoV-2 within the retrieved thrombi of COVID-19 stroke patients
Next, we investigated molecular evidence for SARS-CoV-2 in cerebral thrombi of COVID-19 patients using a multimodal approach. In transmission electron microscopy (TEM), there was no evidence of obvious particles resembling SARS-CoV-2 in the analyzed thrombi (n = 7 thrombi of COVID-19 patients and n = 4 thrombi of control group) ( Fig. 2a-b). However, we were able to successfully identify SARS-CoV-2 RNA by polymerasechain-reaction (PCR) in one thrombus of a COVID-19 patient, while in the other six thrombi of COVID-19 patients and thrombi of controls, no SARS-CoV-2 RNA could be detected by PCR (Fig. 2C). Importantly, we further confirmed the viral PCR detection by sequencing, which revealed presence of the SARS-CoV-2 clade 20A, identified by NextClade v 1.6.0 (clades.nextstrain.org) (data not shown). Overall, we found that we could detect SARS-CoV-2, although rarely, within retrieved thrombi of COVID-19 patients.
A secondary analysis comparing thrombi of COVID-19 patients with thrombi deriving from patients with non-SARS-CoV2 pre-existing infections at stroke onset, confirmed the increased neutrophil density and the higher tNLR in the COVID-19 group. We found no other significant differences regarding thrombus immune phenotype and composition between thrombi of patients with COVID-19 and with pre-existing infections (Additional file 1: Fig. 1 and Additional file 2: Table 1).
Altogether, thrombi of COVID-19 patients differ from those of controls in terms of an increased neutrophil content, particularly evident after calculating the thrombus neutrophil-to-lymphocyte ratio.

Discussion
In this prospective multicenter study, cerebral thrombi of COVID-19 stroke patients featured an increased content of neutrophils and a higher neutrophil-to-lymphocyte ratio. We could detect SARS-CoV-2 directly in one thrombus of a COVID-19 patient, and ACE2 levels were higher in cerebral thrombi of COVID-19 patients compared to controls.
Our findings suggest that the endothelial cells within our cohort of cerebral thrombi have a limited significance as a direct target for SARS-CoV-2. Previous studies described local endotheliopathy as an important element in COVID-19-associated coagulopathy [13], however in our analyses we could not study the arterial wall where we extracted the thrombus from. Still, we observed that the increased expression of ACE2 in COVID-19 thrombi was mainly driven by a subset of CD68 + monocyte/macrophages within the thrombus. Recent reports have stated that inflammatory signals can trigger ACE2 expression, such as type I interferon [41]. Therefore, CD68 + monocytes/macrophages present in the thrombus might potentially represent a target for SARS-CoV-2 and a trigger for immune-induced thrombosis. Of note, despite a recent report showing ACE2 expression in a rare CD4 + T-cell subset, [18] in our study we almost never found lymphocytes expressing ACE2.
In this work, we report for the first-time proof that SARS-CoV-2 can be detected in the cerebral thrombotic material. Our finding was also confirmed by RNA sequencing, ruling out the possibility of a false-positive result. The limited detection rate of SARS-CoV-2 in the thrombi of COVID-19 patients may be due to technical issues, such as the formalin fixation (required in our institution for safety reasons), paraffin embedding and the scarcity of the virus in thrombi, possibly hampering the detection of SARS-CoV-2 in many cases. In addition, pathophysiological reasons may also play a role as COVID-19 may cause or trigger the stroke in some patients, but be an incidental concomitant comorbidity in others. A previous single-case report analyzing the cerebral thrombus of a COVID-19 patient could not detect the virus [5] in the thrombus nor in the endothelial cells of coronary heart vessels [28]. In another study on coronary thrombi however, thrombus viral load was found to be a possible determinant of the thrombus dimension independently of risk factors, and of poorer myocardial blush grade [20]. The composition of a thrombus has been described to be influenced by the underlying etiology of the stroke, the site of thrombus origin and the age of the thrombus [4]. While COVID-19 may trigger alterations in the coagulation cascade, in systemic inflammation and in endotheliopathy [2] [13,[22][23][24]26], we did not find any significant difference in the content of red blood cells, platelets, fibrin and von Willebrand factor when comparing thrombi of the COVID-19 patients and controls. Contrary to a recent post-mortem study on COVID-19 patients that found a significant increase in fibrin and terminal complement C5b-C9 in heart microthrombi [28], we observed almost complete absence of C5b-C9, similar to the level the authors saw in larger coronary artery thrombus aspirates from COVID-19 STEMI cases [28].
On the other hand, analyzing the immune signature of the thrombus of COVID-19 stroke patients we found that thrombi of COVID-19 patients contain an increased density of neutrophils and a reduced level of lymphocytes compared to non-COVID-19 stroke patients. This difference becomes particularly remarkable on calculating the thrombus neutrophil-to-lymphocyte ratio. Indeed, in the ROC analysis, we found that the tNLR was the best predictor for discriminating the thrombi of the two groups of patients. We also confirmed the higher neutrophil density and tNLR of COVID-19 LVO patient thrombi compared to thrombi of stroke patients with pre-existing non-SARS-CoV-2 infections. Thrombus neutrophils and tNLR were not correlated with blood neutrophils or NLR, respectively; suggesting the thrombus is a site of active neutrophil recruitment and not a mere reflection of the blood cell content [12]. Similarly, a recent study on myocardial thrombi found increased markers of neutrophil activation in patients with COVID-19, including neutrophil-platelet aggregates and neutrophil-rich clusters in the macrothrombi [16]. Beyond a quantitative change in neutrophils, reports also describe a deranged phenotype and functionality in COVID-19 [6,33]: the neutrophil activation signature shows prominent features of immature neutrophils in severe COVID-19 cases, representing a clear indication of emergency myelopoiesis [3]. Transcriptional and functional analyses of the neutrophil compartment in the blood of COVID-19 patients have shown an increased capacity for NET formation and enhanced cytokine production and calprotectin release [33]. However, in our study we did not find a significant change in NET density between thrombi of COVID-19 patients and controls, despite the increased neutrophil content. The limited number of thrombi of COVID-19 patients as well as the recently described heterogeneity of NETs in thrombi of diverse etiology [12] might have reduced the possibility of finding differences in NET density between our patient groups.
Our study has some limitations including the small sample size of COVID-19 thrombi analyzed, the limited immunophenotyping of the immune cell subpopulations and the difficulty of dissecting possible thrombus composition peculiarities from SARS-CoV2 compared to other viral infections despite the control group. The strength of this study is it brings the first thorough investigation of the microbiological, structural and inflammatory features of cerebral thrombi retrieved from patients with COVID-19 and large vessel occlusion stroke.
In conclusion, cerebral thrombi of COVID-19 patients can carry the SARS-CoV2 and have an increased neutrophil number, tNLR and ACE2 expression. These findings suggest that neutrophils are the possible culprit in COVID-19-related thrombosis.