Clinicopathologic Analysis of COVID‐19 Associated Thrombi in the Setting of Large Vessel Occlusion: A Prospective Case–Control Study

Abstract

Background

Acute ischemic stroke secondary to large vessel occlusion is among the most serious complications associated with COVID‐19 infection resulting in worse morbidity and mortality. We sought to study the association between COVID‐19 infection and large vessel occlusion thrombus pathology to better define the etiopathogenesis of this atypical cause of stroke.

Methods

Thrombi were collected during mechanical thrombectomy and stained using hematoxylin and eosin. Blinded analysis of pathology was prospectively performed by a board‐certified neuropathologist. Red blood cell, fibrin, and white blood cell predominance was ascertained. Concomitant peripheral blood counts and clinical and imaging data were collected and analyzed. All samples underwent performance of reverse transcription polymerase chain reaction for SARS‐CoV2.

Results

Between January 2020 and February 2022, a total of 952 acute ischemic stroke admissions were seen at the University Medical Center of El Paso, TX. Of these, 195 patients (20.5%) had large vessel occlusions and underwent mechanical thrombectomy and 53 patients had thrombus collected and analyzed. Seven patients (3.6%) tested positive for SARS‐CoV2. COVID‐19 positive patients were more likely to be younger (mean 57.4 years; P=0.07), male (85.7%; P=0.03), and have red blood cell predominant thrombi (85.7%; P=0.03). There was a statistically significant association between peripheral neutrophil count and white blood cell lysis in the overall cohort (P=0.015), who did not differ according to COVID‐19 status.

Conclusion

Thrombi retrieved from patients who were COVID‐19 positive and had stroke demonstrated red blood cell predominance. This finding requires further investigation using appropriate immunohistochemical techniques in a larger cohort of patients.

Nonstandard Abbreviations and Acronyms

AIS

acute ischemic stroke

LVO

large vessel occlusion

MCA

middle cerebral artery

MT

mechanical thrombectomy

NLR

neutrophil‐to‐lymphocyte ratio

Clinical Perspective

• COVID‐19 associated large vessel occlusion is a devastating complication of SARS‐CoV2 infection with associated high morbidity and mortality. Clot pathology is an important and underused tool in understanding the pathophysiology of emerging stroke subtypes. There are only a handful of studies looking at COVID‐19 associated stroke and thrombus pathophysiology.

• We found an association between red blood cell‐rich thrombi and COVID‐19 infection. The prevalence of red blood cell‐rich thrombi in patients who were COVID‐19 positive was 85.7% compared with patients who were COVID‐19 negative of whom only 41.3% were red blood cell rich (P=0.03). There was no association of COVID‐19 status with the hyperdense middle cerebral artery sign or number of passes during thrombectomy. There was, however, an association between peripheral neutrophil count and thrombus white blood cell lysis in the overall cohort. This did not differ by COVID‐19 status.

Acute ischemic stroke (AIS) can be a devastating complication for patients infected with SARS‐CoV2. Although the incidence of AIS in patients with COVID‐19 is relatively low (1.3%), the risk of a poor outcome as represented by death or a discharge destination other than home is twice as high. ¹ , ² Furthermore, severe COVID‐19 has been associated with a higher incidence of AIS compared with nonsevere disease (5.7% compared with 0.8%, respectively). ³ , ⁴ In addition, older age and pulmonary symptoms at presentation are associated with higher mortality in the setting of large vessel occlusion (LVO) and COVID‐19. ⁵ Nonetheless, AIS can be the only presenting symptom in patients with positive serology for SARS‐CoV2. ⁶

The significantly higher prevalence of RBC rich thrombi in the cohort who were COVID‐19 positive raises the possibility of a previously underrecognized stroke mechanism in this group. Further elucidating this mechanism using refined immunohistochemical staining techniques for thrombi may shed light on potential new therapeutic targets. The absence of an association with the hyperdense MCA sign and RBC‐rich thrombi in the COVID‐19 cohort is also intriguing and may allude to inherently different pathophysiologic and imaging properties of thrombi in this newly emerging cause of stroke.

The pathophysiology of AIS in COVID‐19 is complex with multiple proposed pathways involved. ³ Considering the high morbidity and mortality associated with stroke in patients with COVID‐19, ⁷ understanding the pathological and physiological properties of stroke thrombi may be critical in identifying potential new therapeutic targets.

Due to the inflammatory nature of the disease, proposed mechanisms include the development of a cytokine storm and activation of the innate immune system resulting in endothelial cell damage and increased clotting factors leading to hypercoagulability. ³ , ⁶ Other mechanisms include embolic events related to arrythmias, hypoxia‐induced ischemia related to respiratory failure, thrombotic microangiopathy, and endotheliopathy. ³ Neutrophil extracellular trap formation has been increasingly linked with the initiation of intravascular thrombotic events. ⁸ It has been previously demonstrated that COVID‐19 thrombi in patients with stroke had higher neutrophil density and increased neutrophil‐to‐lymphocyte ratio (NLR); no difference in red blood cell (RBC), white blood cell (WBC), fibrin, or neutrophil extracellular trap formation was found, however. ⁹ We therefore sought to evaluate thrombus pathology in LVO strokes in patients who were COVID‐19 positive and compare to patients who were COVID‐19 negative who underwent mechanical thrombectomy (MT). We also studied associated demographic, clinical, laboratory, and imaging characteristics.

Methods

This study was a prospective analysis of a clot bank as well as a retrospective analysis of prospectively maintained database in a dedicated neurointerventional service at the University Medical Center of El Paso. All stroke patients received a SARS‐CoV2 antigen and polymerase chain reaction test during the index admission. Patients were defined as COVID‐19 positive when the SARS‐CoV2 reverse transcription polymerase chain reaction test was positive on admission; repeat testing for symptomatic patients initially testing negative did not yield additional cases in the LVO cohort. Inclusion criteria included patients with LVOs undergoing MT with successful clot retrieval and with available SARS‐CoV2 reverse transcription polymerase chain reaction test results. Exclusion criteria included unavailable SARS‐CoV2 reverse transcription polymerase chain reaction test results or thrombus sample/pathology.

We included consecutive patients presenting with LVO admitted to the University Medical Center of El Paso between January 2020 and February 2022. This is the major academic center and 1 of 3 centers subserved by the Interventional Neurology team at Texas Tech University Health Sciences Center El Paso.

Thrombi were collected during MT using sterile specimen containers. The samples were sent to the pathology laboratory where the thrombus was immediately fixed in 10% neutral buffered formaldehyde and then sectioned and stained using hematoxylin and eosin. Blinded analysis of pathology of the thrombi was performed with each slide being reviewed and analyzed by J.L. (board‐certified neuropathologist). The percentages of RBC, fibrin, and WBC were manually estimated from the hematoxylin and eosin‐stained section under a high‐power microscope (Olympus BX43, Tokyo, Japan). The percentage of neutrophils of total WBC in the thrombus was also manually estimated. Based on previous studies looking at thrombus pathology, RBC‐rich thrombus was variably defined as 40%–60% RBC composition. ¹⁰ , ¹¹ We arbitrarily defined “RBC‐rich” thrombus when the presence of RBC was ≥ than 50% and “fibrin‐rich” thrombus when the presence of fibrin was ≥50%. Of note, there are currently no standardized definitions accepted by the scientific community.

Patient images were reviewed by F.S., A.M., G.C., and R.F. The following data were collected: admission and 24‐hour Alberta Stroke Program Early CT Score, symptomatic hemorrhagic conversion, presence of hyperdense middle cerebral artery (MCA) sign, and final Thrombolysis in Cerebral Ischemia scores. Symptomatic hemorrhagic conversion was defined as parenchymal hematoma type 2 as defined by European Cooperative Acute Stroke Study investigators, ¹² with associated worsening neurologic exam as defined by National Institutes of Health Stroke Scale decline of 4 points or higher.

Permission was granted by the ethics committee and institutional review board at Texas Tech University Health Sciences Center, El Paso. Informed consent was waived due to minimal perceived patient risk and because thrombi would otherwise be considered biological waste.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Statistical Analysis

All statistical analyses were performed with Statistical Package for Social Sciences Version 28.0 statistic software package. Data are presented as mean and SD for continuous variables and as frequency (percentages) for categorical variables. A value of P≤0.05 was considered statistically significant. In analysis of variance model, Eta squared value was calculated to determine the effect size of the association between peripheral neutrophil counts on WBC lysis in clot. Bivariate correlation procedure was performed to determine the correlation between the aforementioned variables. Simple error bars depicting peripheral neutrophil and lymphocyte counts versus WBC lysis in the entire cohort were created.

Results

A total of 952 admissions for AIS were seen at the University Medical Center of El Paso. Of these, 195 patients (20.5%) had LVOs and underwent MT. Of these, 53 patients had sufficient thrombus specimen available for analysis and were included in the study, of whom 7 patients had a positive SARS‐CoV2 reverse transcription polymerase chain reaction. This accounted for 3.6% of all patients undergoing MT. In some patients undergoing MT, the sample collected was not suitable for analysis due to the small amount of thrombus retrieved and the associated difficulties in the fixing and staining processes.

The mean age of patients who were COVID‐19 positive was 57.4 (±25.8) years, and 6 patients (85.7%) were male. All but 2 of our patients were admitted during the first wave of COVID‐19 infections; all our patients were either not vaccinated or had an unknown vaccination status. Post admission, they all received dexamethasone, oxygen therapy, and prophylactic low molecular weight heparin in the absence of hemorrhagic conversion. Remdesivir and other antiviral therapies were not routinely available early in the pandemic. Similarly, subtyping of COVID‐19 variants was not routinely performed in the early phases of the pandemic.

Five patients (71%) presented within 8 hours of symptom onset. All patients had stroke as the presenting symptom. The mean Alberta Stroke Program Early CT Score on admission was 7.7±2.4 and 6.1±3.2 at 24 hours. The hyperdense MCA sign was seen in 3 patients (42.9%) and intravenous thrombolysis was administered to 2 patients (28.5%). All patients had clots in the anterior circulation; mainly in the MCA distribution (4 M1 segment and 1 M2 segment) and the remaining 2 were internal carotid artery‐terminus occlusions. Three patients (42.9%) were treated with aspiration technique and 4 patients (57.1%) were treated with a combination technique (aspiration and stent retrieval). The mean number of passes was 2.3±1.1 and the final Thrombolysis in Cerebral Ischemia scores were as follows: 2 patients (0), 3 patients (2b), and 2 patients (3). One patient had symptomatic hemorrhagic conversion. Only 2 patients were discharged home and 2 died during their hospital stay (Table 1).

Table 1.

Demographics of Patients Who Were COVID‐19 Positive.

Age, y	Gender	Symptomatic	LSW, hours	NIHSS adm	CT ASPECTS adm	Hyperdense MCA sign	IV TPA	CT ASPECTS 24 hours	Symptomatic hemorrhagic conversion	Midline shift at 24 hours, mm	Angiographic LVO site	ADAPT	Solumbra	TICI	Total number of passes	Location at discharge
59	M	Y	5	15	7	Y	N	7	N	0	M1	N	Y	2b	1	Home
81	F	Y	0.5	21	10	Y	Y	10	N	0	M1	N	Y	3	2	Home
48	M	Y	3.0	20	10	N	Y	8	Y	0	M2	Y	N	2b	2	Deceased
8	M	Y	1.5	Unable to assess	3	Y	N	0	N	6	ICA‐T	Y	N	0	3	In patient rehab
67	M	N	19	7	9	N	N	7	N	0	ICA‐T	Y	N	3	1	SNF
53	M	Y	8	16	8	N	N	4	N	2	M1	N	Y	0	4	Deceased
86	M	N	Unknown	14	7	N	N	7	N	0	M1	N	Y	2b	3	SNF

LVO site: M1, sphenoidal segment of MCA; M2, insular segment of MCA. ADAPT indicates direct aspiration; ASPECTS, Alberta Stroke Program Early CT Score; F, female; ICA‐T, internal carotid artery terminus; LSW, last seen well; LVO, large vessel occlusion; M, male; MCA, middle cerebral artery; NIHSS, National Institute of Health Stroke Scale score; SNF, skilled nursing facility; Solumbra, local aspiration combined with stent retriever; and TICI, Thrombolysis in Cerebral Ischemia score.

In contrast, patients who were COVID‐19 negative were older (mean 67.8±15.1 years; P=0.07) and less likely to be male (41.3%; P=0.03). The mean Alberta Stroke Program Early CT Score on admission was 7.8±1.4 and 5.1±2.7 at 24 hours; this was not statistically different from the cohort who were COVID‐19 positive. The incidence of hyperdense MCA sign was 21/46 (45.6%), also not statistically different from the cohort who were COVID‐19 positive. Thrombi in the anterior circulation were seen in 42/46 patients (91.3%); 24/46 patients (52.2%) were treated with aspiration technique only during 1 or more passes and 37/46 (80.4%) were treated with combination technique (aspiration and stent retriever). The mean number of passes was 1.9±1.3; this was lower but not statistically different from the cohort who were COVID‐19 positive (Table 2).

Table 2.

Demographics and Clinicopathologic Characteristics of Patients Who Were COVID‐19 Positive Versus Negative.

Variable	COVID positive, n=7	COVID negative, n=46	P value
Age (mean±SD)	57.4±25.8	67.8±15.1	0.07
Male (n, %)	6 (85.7%)	19 (41.3%)	0.03*
Clot in anterior circulation (n, %)	7 (100%)	42 (91.3%)	–
Aspiration only (n, %)	3 (42.9%)	24 (52.2%)	0.64
Aspiration and retriever combined (n, %)	4 (57.1%)	37 (80.4%)	0.17
Hyperdense MCA sign	3 (42.9%)	21 (45.7%)	0.61
Number of passes (mean±SD)	2.29±1.11	1.87±1.29	0.42
Thrombus analysis
RBC rich (≥50%)	6 (85.7)	19 (41.3)	0.03*
Fibrin rich (≥50%)	1 (14.3)	17 (36.9)	0.24
WBC (lysis)	3 (42.9)	24 (52.1)	0.65
WBC (aggregated)	5 (71.4)	33 (71.7)	0.98
Neutrophil percentage (mean±SD)	55.0±25.3	55.2±25.3	0.98
Peripheral blood counts
Neutrophil to lymphocytes ratio (mean±SD)	6.25±4.21	7.98±7.86	0.57
Peripheral lymphocytes (mean±SD)	1.39±0.77	1.40±0.66	0.97
Peripheral neutrophils (mean±SD)	6.38±3.10	8.11±4.70	0.35

MCA, middle cerebral artery; RBC, red blood cell; and WBC, white blood cell.

^* Statistically significant difference.

Thrombus pathology analysis revealed the following: the incidence of RBC‐rich clot was 6/7 (85.7%) in the cohort who were COVID‐19 positive compared with 19/46 (41.3%) in the cohort who were COVID‐19 negative; P=0.03 (Table 2). See Figure 1A for an example of RBC‐rich clot in a patient who was COVID‐19 positive from our cohort. Fibrin‐rich clot was less prevalent in the cohort who were COVID‐19 positive compared with the cohort who were COVID‐19 negative (14.3% versus 36.9%, respectively; P=0.24). See Figure 1B for examples of fibrin‐rich clot in a patient who was COVID‐19 negative.

Figure 1.

Thrombus histopathology in patients with LVO according to COVID-19 status.

A, Thrombus in a patient who was COVID‐19 positive, stained with H&E demonstrating predominantly RBCs (solid yellow arrow) with interspersed fibrin (solid black arrow) and platelets. Original magnification X100. B, Thrombus in a patient who was COVID‐19 negative, stained with H&E; demonstrating predominantly fibrin. Original magnification X100. H&E indicates hematoxylin and eosin; LVO, large vessel occlusion; and RBC, red blood cell.

There was no statistical difference in the peripheral neutrophil percentage, peripheral lymphocyte percentage, or the percentage of thrombus demonstrating WBC lysis and aggregation between the 2 cohorts. Similarly, the peripheral NLR was not different between the 2 cohorts (Table 2).

Importantly there was an association between peripheral neutrophil count and thrombus pathology demonstrating WBC lysis in the entire cohort irrespective of COVID‐19 status; a higher peripheral neutrophil count was associated with the presence of WBC lysis on thrombus analysis (P=0.015); and Eta squared value was 0.110. This is also depicted by the simple error bar in Figure 2.

Figure 2.

Simple error bars depicting peripheral neutrophil count versus WBC lysis in the entire cohort.

^*Represents a statistically significant difference between the simple error bars. ABS indicates absolute; and WBC, white blood cell.

Discussion

In our study, patients who were COVID‐19 positive with LVO were more likely to be younger, male, and have an RBC‐rich clot, which was statistically significant. Patients who were COVID‐19 positive also required a higher number of passes during MT; however, this finding did not reach statistical significance. There was also a significant association between peripheral neutrophil count and presence of cytolysis on thrombus pathology; this did not differ by COVID‐19 status.

Analysis of thrombus pathology in LVO stroke has become of great interest among the scientific community with the attempt of understanding the underlying pathophysiology and cause of different stroke subtypes. ¹³ , ¹⁴ However, there are relatively few dedicated thrombus pathology studies in the setting of COVID‐19 infection. ⁹

Our cohort is the largest thus far to specifically compare thrombus pathology in patients who were COVID‐19 positive and negative. ⁹ We found that COVID‐19 associated thrombi were predominantly RBC rich. This finding could have implications for understanding the pathogenesis and therefore identify potential therapeutic targets for patients with COVID‐19 with LVO. Given the differences in clot pathology between our study and that of Genchi et al., it highlights the fact that there could be multiple possible mechanisms involved in the pathogenesis of AIS in COVID‐19 infection. ⁹ It also gives credence to the finding that there are morphologic changes in the RBC structure and function induced by SARS‐CoV2 infection. In 1 Italian study that analyzed peripheral blood smears from patients with COVID‐19 related anemia, huge rouleaux formation, and autoagglutination was seen in 85% of patients. ¹⁵ The same group found RBC‐bound antibodies detectable by direct antiglobulin test in almost half their patients with severe COVID‐19 infection. ¹⁶ They hypothesized therefore that the loss of RBC biconcavity and complement activation would promote RBC “stacking,” spontaneous agglutination, and therefore microthrombosis characteristic of SARS‐CoV2 infection. ¹⁵ , ¹⁶ In another study, RBCs from mild COVID‐19 demonstrated highly altered hematologic parameters such as mean corpuscular hemoglobin, mean corpuscular volume with concomitant reduction of RBC deformability at baseline and under an osmotic gradient. ¹⁷

These RBC integrity changes in COVID‐19 may be associated with deficiency of Band‐3 protein, which is an ion exchange protein responsible for bicarbonate and chloride anion exchange as well as for stability of the cytoskeleton multiprotein complex, which regulates RBC mechanical and elastic properties regulating blood viscosity. ¹⁸ , ¹⁹ Although in general there are lower serum nitric oxide (NO) levels in COVID‐19, ²⁰ intraerythrocyte NO levels are actually higher and may be an adaptive response related to silent hypoxia. ²¹ Given our findings of RBC‐rich clot in patients with COVID‐19, it is possible that hemolysis and local release of NO at the site of clot formation would allow binding of free heme and hemoglobin to NO forming methemoglobin and nitrates resulting in local vasoconstriction. ¹⁹ Taken together with the systemic imbalance of NO production, disseminated intravascular coagulopathy and cytokine storm, which is a hallmark of COVID‐19 infection, this may account for the local prothrombotic milieu in the cerebrovascular system and hence the poor outcomes despite successful recanalization. ⁷

This finding also raises the possibility of alternative therapeutic targets specifically for stroke and LVO associated with COVID‐19 infection. One possibility is pentoxifylline, which is a phosphodiesterase‐4 inhibitor and has immunomodulatory, antiinflammatory, as well as rheolytic properties with the potential to improve RBC deformability. ²² , ²³ Although a randomized study performed in patients with COVID‐19 showed lower levels of interleukin‐6 in the pentoxifylline group, there was no difference in intensive care unit admission and intubation; however AIS and other prothrombotic events were not specifically evaluated. ²⁴ The effectiveness of therapeutic anticoagulation with unfractionated or low‐molecular heparin in select cases of COVID‐19 associated stroke also alludes to a local prothrombotic milieu, ³ , ²⁵ compounded by systemic hypercoagulability associated with antiphospholipid antibodies. ²⁶ , ²⁷

Although RBC‐rich clots are typically known to be associated with more favorable outcomes such as having a higher chance of recanalization and increased tPA sensitivity, ²⁸ we did not find a statistically significant difference in number of passes or technique used for revascularization (aspiration only versus combination of aspiration and stent retriever). Of note, there was a higher mean number of passes in the group who were COVID‐19 positive compared with the group who were COVID‐19 negative (2.3 versus 1.9), which was not statistically significant and slightly higher than rates reflected in the literature. ²⁹ Similarly, there was no statistically significant difference in the incidence of hyperdense MCA sign, which has been previously associated with RBC‐rich thrombi ³⁰ In addition, the hyperdense MCA sign has been traditionally associated with success of recanalization. ³¹ The lack of association between HCMAS with RBC‐rich thrombi in our cohort together with the increased number of passes, albeit not statistically significant, possibly alludes to the inherently different pathophysiological and imaging properties of RBC‐rich thrombi in patients who were COVID‐19 positive versus COVID‐19 negative.

Given the concept of “immunothrombosis” as a mechanism of stroke in COVID‐19, we had hypothesized a higher thrombus WBC percentage with concomitant neutrophil predominance as found by another COVID‐19 clot analysis study. ⁹ This study also demonstrated a higher NLR in the cohort who were COVID‐19 positive. In our study, we did not specifically evaluate for NLR within thrombi. Although WBC aggregation and lysis were not statistically different between the cohorts who were COVID positive and negative possibly due to the relatively small number of patients who were COVID‐19 positive, we did find a significant relationship between peripheral neutrophil count and thrombus lysis in the overall cohort. This finding is intriguing, and the role of neutrophils and neutrophil extracellular traps has been inferred in stroke thrombus pathophysiology. ³² A similar association with extracranial carotid disease and associated atherothrombosis has been reported and attributed to COVID‐19 endotheliitis with a concomitant procoaguable state in the setting of COVID‐19. ³³

Similarly, there was no significant difference in NLR between the groups who were COVID‐19 positive and negative in our cohort possibly reflective of less severe COVID‐19 disease in our cohort with stroke. Previously, higher NLR has been associated with poorer functional outcomes in those with AIS in general. ³⁴ Specifically, in patients with COVID‐19, higher NLR has been similarly associated with poorer outcomes. ³⁵ Furthermore, in patients with COVID‐19 and stroke who were undergoing MT, higher NLR was also associated with worse outcomes regardless of recanalization status. ³⁶

COVID‐19 associated AISs are associated with a higher mortality rate than those without an active COVID‐19 infection ³⁷ ; conversely AIS in the context of SARS‐CoV2 infection is associated with twice the morbidity and mortality. Indeed, the outcomes are worse despite MT. ¹ , ⁷ Our cohort had an inpatient mortality of 28.6% similar to a large series of patients with COVID‐19 and stroke who were undergoing MT, which reported a mortality at discharge of 30.6%. ⁷

Limitations of our study include its relatively small subset of patients who were COVID‐19 positive in the overall sample leading to loss of statistical power for certain subgroup analyses. There is also the possibility of selection bias given the proportion of patients with LVO undergoing MT who did not have thrombus available for analysis, most of which were due to insufficient amount of thrombus retrieved for staining and analysis. Of note, only 27.2% of patients with LVO in our study had sufficient thrombus for analysis. Of note, there may be a higher incidence of intracranial atherosclerotic disease in our patient population, which is usually associated with lower thrombus yield. Lack of advanced immunohistochemical stains may also have limited further granular analysis of thrombus pathology for example for neutrophil extracellular traps.

Given the high morbidity and mortality associated with the COVID‐19 pandemic in general and in patients with stroke and COVID‐19 in particular despite timely revascularization; the role of thrombus pathology in understanding the etiopathogenesis and devising appropriate therapeutic strategies for this newly emerging cause of stroke was possibly a missed opportunity, particularly early in the pandemic. ¹ , ⁷ , ¹³ Newer techniques such as diffuse reflectance spectroscopy and optimization of existing immunohistochemical and mass spectrometry‐based proteomic techniques for timely analysis of thrombus particularly for newly emerging stroke causes may be a necessary component of a future multifaceted strategy to improve outcomes. ¹³ , ³⁸ , ³⁹

Disclosures

The authors declare there are no relevant or material financial interests or disclosures that relate to the research.