SARS-CoV-2 Receptors are Expressed on Human Platelets and the Effect of Aspirin on Clinical Outcomes in COVID-19 Patients

Coronavirus disease-2019 (COVID-19) caused by SARS-CoV-2 is an ongoing viral pandemic marked by increased risk of thrombotic events. However, the role of platelets in the elevated observed thrombotic risk in COVID-19 and utility of anti-platelet agents in attenuating thrombosis is unknown. We aimed to determine if human platelets express the known SARS-CoV-2 receptor-protease axis on their cell surface and assess whether the anti-platelet effect of aspirin may mitigate risk of myocardial infarction (MI), cerebrovascular accident (CVA), and venous thromboembolism (VTE) in COVID-19. Expression of ACE2 and TMPRSS2 on human platelets were detected by immunoblotting and confirmed by confocal microscopy. We evaluated 22,072 symptomatic patients tested for COVID-19. Propensity-matched analyses were performed to determine if treatment with aspirin or non-steroidal anti-inflammatory drugs (NSAIDs) affected thrombotic outcomes in COVID-19. Neither aspirin nor NSAIDs affected mortality in COVID-19. However, both aspirin and NSAID therapies were associated with increased risk of the combined thrombotic endpoint of (MI), (CVA), and (VTE). Thus, while platelets clearly express ACE2-TMPRSS2 receptor-protease axis for SARS-CoV-2 infection, aspirin does not prevent thrombosis and death in COVID-19. The mechanisms of thrombosis in COVID-19, therefore, appears distinct and the role of platelets as direct mediators of SARS-CoV-2-mediated thrombosis warrants further investigation.


Introduction 57
COVID-19 is caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) and curiously 58 displays a propensity for thrombosis in multiple vascular beds. COVID-19-related thrombosis may 59 contribute to severe organ injury and death. The incidence of thrombotic events was as high as 31% in 60 one cohort 1 . Clinical and autopsy studies of COVID-19 patients suggest an increased risk of 61 microthrombi, venous thromboembolism (VTE), and ischemic stroke 2,3 . Activated platelets are 62 circulating mediators of thrombosis and, therefore, may serve as a logical therapeutic target in COVID-63 19. Two registered clinical trials (NCT04363840 and NCT04365309) will prospectively evaluate patient 64 outcomes following low dose aspirin in the context of SARS-CoV-2 infection.

66
SARS-CoV-2 utilizes an spike glycoprotein to bind to the host transmembrane angiotensin-converting 67 enzyme 2 (ACE2) and is then cleaved by the serine protease TMPRSS2 to coordinate entry into the host 68 cell 4,5 . Therefore, co-expression of ACE2 and TMPRSS2 may be important for host cell entry and 69 infectivity of SARS-CoV-2. Importantly, human tissue distribution of ACE2 and TMPRSS2 mirrors organ 70 system involvement in COVID-19 and includes the lungs 6-11 , vascular endothelium 9-12 , heart 11,13,14 , 71 kidneys 8,10,13 , liver 8,10 , digestive tract 8,10,11,15 , nasal epithelium 7,10,11 and central nervous system 10,14 .  A recent report demonstrated that COVID-19 patients have a divergent platelet transcriptome from 82 healthy individuals, and aspirin suppresses COVID-19 platelet activation in vitro 19 . The platelet surface 83 receptor for SARS-CoV-2 was not clarified in this study, while a similar investigation by another group 84 identified mRNA for SARS-CoV-2 in human platelets 20 . Thus, our goal was to determine if platelets 85 express known SARS-CoV-2 receptor proteins and, as with influenza previously, contribute to thrombotic 86 events in patients. In the absence of clinical trial data, we sought to evaluate the potential benefit in 87 mitigating thrombotic responses in vivo with use of aspirin or other NSAID antiplatelet therapies by 88 propensity matching patients using real-world data.

100
Immunoblotting 101 Washed platelets from healthy subjects or patients with coronary artery disease (CAD) enrolled at the 102 Cleveland Clinic main campus in Ohio were isolated and proteins separated by SDS-PAGE as we have 103 previously documented 21,22 and in accordance with IRB protocols (#19-1451 for patients and #20-413 104 5 for healthy volunteers). We utilized human brain lysate, human placenta, and engineered human heart 105 tissue as positive controls for TMPRSS2 and ACE2. Human brain lysate is commercially available (Novus 106 #NB820-59177). Human placenta lysate was prepared as follows: placental villous tissue was collected 107 immediately upon uncomplicated, full-term (37-42 weeks' gestation), elective C-section deliveries at 108 MetroHealth Hospital in Cleveland, Ohio and approved by the Cleveland Clinic and MetroHealth IRB 109 (#16-1311 and #16-00335, respectively). This tissue was normally discarded placentas with intact fetal 110 membranes, and following inclusion in the study no protected health information, identifiers, or clinical 111 data were collected. A waiver of consent was approved by the Cleveland Clinic Foundation IRB as 112 the placentas were collected anonymously. Engineered human heart tissue was obtained as follows:

Statistical Analysis 148
Categorical factors are summarized using frequencies and percentages, while continuous factors are 149 described using median and ranges. Initial descriptive analyses were performed. Comparisons were 150 made between those with known death status and those with missing death information to identify if 151 any differences exist in these cohorts. Then among those with known death status, differences in COVID 7 positive and COVID negative patients were assessed. Finally, after stratifying by COVID status, 153 comparisons of those with and without aspirin use were performed. For all tables, continuous measures 154 were compared using nonparametric Wilcoxon rank sum tests, while categorical factors were compared 155 using Pearson chi-square tests or Fisher exact tests, for rare events.

157
Given the differences across many covariates, propensity score matching was performed to account for 158 differences between those with and without aspirin use. This approach used two steps. First, multiple 159 imputation was performed on all demographic and covariate measures within COVID status stratified 160 datasets, using fully conditional specification methods. Ten imputed datasets were created. Then

Results 177
Expression of ACE2 (n=6) and TMPRSS2 (n=3) on the platelet surface was observed by confocal 178 microscopy ( Figure 1). Expression of TMPRSS2 in healthy subjects (mean age 40.1 ± 2.8 years, n=20) was 179 also confirmed by immunoblotting at the expected molecular weight of ~50 KDa. Utilizing human brain as a positive control, TMPRSS2 expression was standardized to a loading control 184 with no correlation between age and platelet TMPRSS2 expression (Figure 2A; r 2 =0.058, p=0.30). Since 185 ACE2 exists as multiple glycosylated proteins of variable molecular weight 26-28 , human brain 29 , human 186 placenta 30 , and engineered heart tissue 31 were utilized as positive controls to confirm predominant 187 migration at ~100 kDa as expected. Given that patients with confirmed CAD receive antiplatelet 188 medications according to established guidelines, TMPRSS2 expression for healthy controls (n=20) was 189 compared to patients with coronary artery disease (CAD, n=10) and, while numerically greater in CAD,

195
Platelet ACE2 in healthy subjects (n=20) was compared to patients with CAD (n=10) and, again, while 196 numerically higher in CAD, was without a statistical difference ( Figure 2B, p=0.11). Further, we did not 204 were not exposed and 285 patients were exposed to aspirin. In an attempt to differentiate an anti-205 platelet drug effect with aspirin from a more general NSAID class effect, we propensity-matched 206 patients 1,445 patients not exposed and 465 patients exposed to NSAID therapy (Figure 3).   The 248 propensity-matched patients either treated with aspirin or not demonstrated no significant 217 group differences in demographics or clinical covariates. Aspirin therapy did not alter mortality (13.3% 218 vs 15.3%, p=0.53). The 444 propensity-matched patients either exposed or not to NSAIDs demonstrated 219 no significant group differences in demographics or clinical covariates. NSAID therapy did not alter 220 mortality (7.0% vs 7.2%, p=0.90). In propensity-matched patients treated with aspirin, the incidence of 221 MI (2.0% vs 0.81%, p=0.27) and VTE (4.0% vs 1.6%, p=0.12) were not significantly different, but aspirin 222 therapy was associated with an increased risk of thrombotic stroke (3.6% vs 0.40%, p=0.036). In    showing aspirin has no mortality benefit in patients with COVID-19, and, in fact, displays a slightly 247 increased signal for harm driven mostly by thrombotic stroke. Platelet reactivity data in vitro is often 248 extrapolated to suggest a risk for harm, but it is important to acknowledge that the behavior of anti-249 platelet medications in vivo can be markedly different from in vitro studies. Our goal was to clarify this 250 concern by using real-life data with both mortality and thrombotic end points.

252
The failure to show a protective effect of the antiplatelet medication aspirin in patients with COVID-19 253 may be related to the dose administered, an insensitivity to aspirin's mechanism of platelet inhibition in

Study Limitations 305
The observational nature of this study from just two hospitals has intrinsic limitations, and the small 306 patient sample to allow for propensity matching limits generalizability of our findings. A few patients 307 testing positive for SARS-CoV-2 were ambulatory and we relied on physician prescriptions making it 308 impossible to confirm compliance to aspirin therapy.

507
The scale bar is noted. 508 509