Microthrombi as a Major Cause of Cardiac Injury in COVID-19

Myocardial injury is a common phenomenon in hospitalized patients with coronavirus disease 2019 (COVID-19) and is associated with worse outcomes. In 1 study in which troponin I was measured within 24 hours of admission to assess myocardial damage, 36% of patients had elevated troponin I concentrations.¹ After adjusting for disease severity and relevant risk factor differences, even small amounts of myocardial injury were associated with increased risk of death.

The cause of myocardial injury in patients with COVID-19 has not been previously elucidated in a systematic manner. Various pathophysiological mechanisms have been hypothesized, including direct viral invasion of the heart or immune-mediated cardiac injury causing myocarditis, stress-cardiomyopathy, myocardial supply-demand mismatch leading to type II myocardial infarction, cytokine storm and hypercoagulability resulting in either increased risk of epicardial coronary thrombosis, pulmonary emboli, and microvascular thrombi.² We recently reported a case of a young woman dying of ST-segment–elevation myocardial infarction (STEMI) and cardiogenic shock who was found to have multiple myocardial microthrombi at autopsy, but the prevalence of such findings among hospitalized patients with COVID-19 has not been reported.³ Understanding the exact nature of cardiac injury in patients with COVID-19 may affect public health strategies, diagnostic testing, and new therapeutic trials for those infected with COVID-19.

Here, we report the analysis of 40 cardiac autopsies conducted on hospitalized patients dying of COVID-19 during the height of the pandemic in February 2020, in Bergamo, Italy. Our goals were to systematically determine (1) the frequency of cardiac injury in hospitalized patients as assessed by autopsy findings of myocyte necrosis, (2) the major causes and risk factors for myocardial injury, and (3) the pathophysiology of cardiac necrosis in subjects infected with COVID-19.

Methods

The current analysis was performed in the frame of a systematic pathology study to assess cardiac injury by COVID-19 infection. The data that support the findings of this study are available from the corresponding author on reasonable request.

The hearts of 40 unselected patients dying from COVID-19 at Ospedale Papa Giovanni XXIII in Bergamo, Italy, and undergoing autopsy were collected and sent to CVPath Institute (Gaithersburg, MD) for detailed pathological analysis. The study protocol was reviewed and approved by the ethical committee at Ospedale Papa Giovanni XXIII, Bergamo, Italy (2020-0056), and by the CVPath Institute Institutional Review Board (RP0112), Gaithersburg, MD, and registered at https://clinicaltrials.gov; Unique identifier: NCT04367792.

All patients had a laboratory-confirmed diagnosis of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection either by nasal swab test on admission (39/40 patients) or by postmortem reverse transcription polymerase chain reaction. Demographic data, medical history, clinical presentation, treatment, and in-hospital course were derived from medical records. All patients were managed according to local clinical practice.

The specimens were anonymized before shipping, and the examining pathologist was blinded to the clinical details. All tissue specimens were fixed in 10% buffered formalin for at least 72 to 96 hours before shipment. Whole hearts and lungs (either paraffin blocks or tissues) were sent to the CVPath Institute for pathological examination.

Pathological Analysis

Hearts were radiographed, and coronary arteries with significant calcification were removed from the heart, radiographed, decalcified as necessary before dehydration, and sectioned at 3- to 4-mm intervals at the time of embedding in paraffin to determine the extent of luminal narrowing. Coronary arteries without any significant calcification were cut on the heart at 3- to 4-mm intervals and sections submitted with the maximal narrowing from the 4 major coronary arteries (left main, left anterior descending, left circumflex, and right coronary arteries). If no significant narrowing was observed, sections from the proximal 4 major coronary arteries were submitted. A total of 14 transmural sections of myocardium (anterior, inferior, and lateral left ventricle [LV], ventricular septum, anterior, and inferior wall of the right ventricle [RV], at mid and apex, and 1 each from the left and right atrium) were dehydrated, embedded in paraffin, sectioned at 4- to 6-μm intervals, and stained with hematoxylin and eosin and Masson’s trichrome stain for histological evaluation. Acute myocardial infarction was defined as areas of myocyte necrosis ≥1 cm², whereas focal myocyte necrosis was defined as necrosis of >20 myocytes with an area of necrosis ≥0.05 mm² but <1 cm² (Figure 1).

• Download figure
• Download PowerPoint

Additional sections from the slides with areas of interest were cut and stained by immunoperoxidase stains with anti-CD61 (prediluted, 760–4247, Ventana Medical Systems, Inc, Oro Valley, AZ), anti–von Willebrand Factor (dilution 1:4000, A0082, Agilent Technologies, Santa Clara, CA), anti–fibrin II (dilution 1:100, NYBT2G1, Accurate Chemicals, Westbury, NY), anti–d-dimer (dilution 1:100, NBP1-05045, Novus Biologicals), and anticomplement C5b-9 (dilution 1:800, ab66768, abcam, Cambridge, MA). Area of percent positive staining was quantified by ZEN (Zeiss) or HALO software (Indica Labs) as previously described.⁴

Total RNA was extracted from myocardial tissue and coronary artery samples. Concentration and quality of RNA samples were measured. Quantitative reverse transcription polymerase chain reaction was performed using specifically designed primers for SARS-CoV-2 (N1 and N2 from CDC EUA assay, Integrated DNA Technologies, Coralville, IA). Ribonuclease P was used as a control, and all RNA samples from the lung and heart had a cycle threshold ≤40. The detection of SARS-CoV-2 RNA was determined by the amplification at cycle threshold ≤40.⁵

To detect SARS-CoV-2 within formalin-fixed paraffin-embedded tissue sections, RNA scope in situ hybridization (Advanced Cell Diagnostics, Hayward, CA) was visualized by confocal microscopy (LSM800, Zeiss, Oberkochen, Germany). SARS-CoV-2 viral RNA–positive cells were visualized by the colocalization of SARS-CoV-2 positive-sense (genomic; violet) and negative-sense (replicative intermediate; turquoise) probes. Immunofluorescence staining for CD61/CD42b and vascular endothelial cadherin was performed using the same or adjacent sections as in situ hybridization.

For transmission electronic microscopy (TEM), heart samples were fixed with 2.5% glutaraldehyde in 0.1 M sodium phosphate buffer at 4°, and then 1% osmium tetroxide was used for postfixation. Tissues were dehydrated by graded alcohol and embedded in EPON Resin. Ultrathin sections were cut at 80 nm with a microtome. Specimens were assessed with the TEM (Hitachi H-7650, Hitachi Science Systems Ltd, Tokyo, Japan).

Analysis of Coronary Artery Aspirates

A comparative analysis of the constituents of thrombus was performed between microthrombi derived from autopsy cases from subjects with COVID-19 (n=5), intramyocardial thromboemboli from COVID-19–negative subjects (n=5) obtained at random from the CVPath Sudden Cardiac Death Registry, and thrombus aspirates retrieved from the culprit lesion of coronary arteries in COVID-19–positive and COVID-19–negative STEMI cases (5 of each) presenting with STEMI, and undergoing primary percutaneous coronary intervention in the same time span of the pathology data collection. Manual thrombus aspiration was performed according to clinical practice (indication and procedure). Patients with an occluded culprit coronary artery TIMI (Thrombolysis in Myocardial Infarction) flow rate 0 to 1 at baseline coronary angiography, or patients with an open artery (TIMI 2–3) but with large filling defect or haziness in a segment reachable by a thrombectomy catheter, were selected for the study. Aspirates were fixed in 10% formalin and shipped to CVPath for analysis. This protocol was reviewed and approved by the ethical committee at Ospedale Papa Giovanni XXIII, Bergamo, Italy (2020-082), and by the CVPath Institute Institutional Review Board, Gaithersburg, MD (RP0112). Aspirates were embedded in paraffin, cut, and stained with the indicated antibodies as described earlier.

Statistical Analysis

Normality of data was checked using the Shapiro-Wilk test. For normally distributed data, ANOVA or Student t test was used to compare data. For multiple comparisons testing, Tukey-Kramer post hoc testing was used. Categorical values were analyzed by the χ² or Fisher exact test as appropriate. For nonnormally distributed data, the Wilcoxon rank-sum test and post hoc Steel-Dwass tests were applied. A P value <0.05 was considered significant. JMP (version 14, SAS, Cary, NC) or GraphPad Prism (GraphPad Software version 8.2.1, La Jolla, CA) software was used for statistical analysis as appropriate.

Results

Clinical characteristics of the 40 autopsy cases examined are shown in Table 1. The average age was 74 years, with the majority of patients (29/40, 72.5%) being male. Most (90%) were admitted for severe respiratory failure (median Pao₂/fraction of inspired oxygen ratio, 123), whereas 4 patients (10%) presented with a cardiovascular emergency (3 STEMI, 1 stroke). Except for the patients admitted with STEMI, no other cases of clinical myocardial infarction were recorded during hospitalization. However, 20% of electrocardiograms in the overall study group showed abnormalities suggestive of ischemia (ie, ST-segment–elevation/depression >0.1 mV, new left bundle-branch block, or inverted T wave), although not associated with chest pain.

Hearts were examined blindly at CVPath Institute without knowledge of the clinical finding except that all subjects had died of COVID-19 (as confirmed by laboratory testing during hospitalization). An analysis revealed the presence of myocyte necrosis in 14 out of the 40 (35%) cases. Of these 14 cases, 3 (21.4%) had evidence of acute myocardial infarction defined as an area of necrosis ≥1 cm², whereas 11 (78.6%) had focal myocyte necrosis (ie, ≥0.05 mm² but <1 cm² contiguous area). There was no difference in the presence of severe coronary artery disease (defined as epicardial coronary stenosis >75% cross-sectional area narrowing) between cases with and without myocardial necrosis. Details of the pathological findings of the heart and lungs are shown in Table 2. Overall, heart weight, LV cavity size, and LV and RV wall thickness were not significantly different between those with and without myocardial necrosis (Figure I in the Data Supplement). Similarly, lung pathology such as diffuse alveolar damage, pulmonary artery thrombi, or microthrombi in alveolar septa were not different between cases with and without necrosis (Table 2). None of the hearts examined had evidence of myocarditis as defined by the European Society of Cardiology.⁶ Other significant cardiovascular findings such as the presence of amyloid, hypertensive heart disease (ie, hypertrophy), and valvular disease are listed in Table 2.

Once the pathological findings were characterized, the clinical and laboratory findings were unblinded and divided according to the presence and absence of myocardial necrosis as shown in Tables 1 and 3. Overall, subjects with necrosis were more frequently female compared with the patients without necrosis (50% versus 84.6%, P=0.03), had a greater prevalence of chronic kidney disease (35.7% versus 7.7%, P=0.04), had a higher rate of STEMI at presentation (21.4% versus 0%, P=0.04), and had a shorter interval between symptom onset and hospital admission (4.0 versus 6.5 days, P=0.02). Subjects with myocardial necrosis had lower values of hemoglobin and C-reactive protein, greater high-sensitivity troponin I values (23 386 versus 226, P=0.03), and a higher rate of ischemic EKG changes (42.9% versus 8.3%, P=0.03; Table 3). No difference in the degree of respiratory impairment was detected, as confirmed by the similar values of Pao₂/Fio₂ ratio and similar degrees of ventilatory support. In-hospital therapy and rate of adverse events were also similar between the subjects with versus without myocardial necrosis (Table I in the Data Supplement). Molecular analysis of RNA extracted from the lungs and various areas of the heart revealed the presence of virus in the lungs as detected by polymerase chain reaction in 34 of the 40 cases (85%), but it was detectable in the heart in only 8 cases (20%) (Table II in the Data Supplement). There was no difference in virus presence in cases of necrosis versus no necrosis.

Table 4 lists the pathological findings in the 14 subjects with myocyte necrosis. Two cases had evidence of epicardial coronary thrombosis in the setting of severe coronary atherosclerosis and underwent percutaneous coronary intervention. Four of 14 (28.6%) cases had evidence of RV strain as indicated by RV necrosis (Table 4). Of the 11 cases with focal myocyte necrosis, 8 (72.7%) had microthrombi, whereas 3 (27.3%) showed no microthrombi. Microthrombi were observed in 9 of 14 (64.3%) cases with myocyte necrosis, 8 with focal necrosis, and 1 with acute myocardial infarction. We measured the distribution and extent of focal myocyte necrosis and microvessel thrombosis in subjects with COVID-19. Overall, focal myocardial necrosis was more common in the LV inferior and lateral wall and ventricular septum as follows: inferior 55%, lateral 36%, septum 36%, inferior RV 36%, anterior LV 18%, atria 9%, and anterior RV 0% (Figure II in the Data Supplement). The distribution pattern of microthrombi were similar to focal myocyte necrosis (inferior LV 67%, lateral LV 56%, ventricular septum 44%, inferior RV 44%, anterior LV 22%, anterior RV 11%, and atria 11%) (Figure II in the Data Supplement). Focal necrosis was found in 13% of histology sections (22/154 sections from 11 patients). Similarly, microthrombi were observed in 27.7% of histology sections (31/126 sections from 9 patients). The presence of microthrombi was significantly associated with the presence of focal necrosis in a section-based analysis (P<0.01, Fisher exact test, Figure III in the Data Supplement). A cumulative distribution curve showing the percentage area of microthrombi per 1 mm² myocardial tissue is shown in Figure IV in the Data Supplement. Only 1 case of microthrombi was associated with acute myocardial infarction. In this case, the subject presented with STEMI involving the inferior and lateral regions of the LV and posterior wall of the RV accompanied by profound cardiogenic shock causing global myocyte necrosis as described previously.³ Coronary artery disease was generally mild to moderate in almost all cases of microthrombi. None of the cases with myocyte necrosis had an unstable coronary plaque phenotype.

Because recent reports have suggested that SARS-CoV-2 may be found in endothelial cells in the hearts of subjects dying of COVID-19, we conducted in situ hybridization and indirect immunofluorescence as well as TEM to look for virus presence in COVID-19 hearts with evidence of microthrombi. In the 3 hearts examined by in situ hybridization, rare virus presence could be found in cardiac myocytes (without associated inflammation) but not in microvascular endothelial cells with evidence of microthrombi (Figure 2). Similarly, by TEM, no evidence of virus particles could be found within endothelial cells in vessels with and without microthrombi in the 5 hearts examined (Figure 2). These data suggest direct endothelial infection is likely not a major mechanism of cardiac microthrombi formation in subjects with COVID-19 infection.

• Download figure
• Download PowerPoint

We compared the constituents of thrombi between coronary thrombus aspirates retrieved from the culprit lesions of a separate group of COVID-19–positive and COVID-19–negative STEMI cases (5 of each) treated during the height of the pandemic to 5 of the autopsy cases with myocardial microthrombi already described and to 5 additional cases from COVID-19–negative patients with evidence of intramyocardial thromboemboli (associated with epicardial coronary thrombosis in all cases) from the CVPath Sudden Coronary Death Registry. Clinical characteristics of the 4 groups are shown in Table III in the Data Supplement. Given the previous connection between COVID-19 and hypercoagulability, we examined factors known to be involved with clotting using antibodies against platelets (CD61), fibrin II, von Willebrand Factor, D-dimer, and terminal complement complex C5b-9, also known as the membrane attack complex. Microthrombi from subjects with COVID-19 were richer in fibrin II compared with both myocardial thromboemboli from COVID-19–negative autopsies and aspirates from patients with STEMI (both non–COVID-19 and COVID-19) (fibrin II % area: microthrombi from COVID-19–positive subjects, 71.8 [interquartile range, 57.4–85.1]; non–COVID-19 thromboemboli, 35.6 [interquartile range, 10.1–58.7]; non–COVID-19 STEMI thrombus aspirates, 21.3 [interquartile range, 12.3–24.0]; COVID-19 STEMI thrombus aspirates, 36.7 [interquartile range, 12.4–52.1]; overall P<0.0001). Microthrombi from COVID-19–positive subjects were also richer in C5b-9 compared with non–COVID-19 myocardial thromboemboli and with COVID-19–negative and COVID-19–positive STEMI thrombus aspirates (C5b-9 % area: microthrombi from COVID-19–positive subjects, 19.3 [interquartile range, 6.7–39.2]; non–COVID-19 thromboemboli, 0.001 [interquartile range, 0–0.047]; non–COVID-19 STEMI thrombus aspirates, 0.22 [interquartile range, 0.036–0.28]; and COVID-19 STEMI thrombus aspirates, 0.027 [interquartile range, 0.014–8.5]; overall P<0.0001) (Figure 3). However, thromboemboli associated with epicardial coronary thrombosis from COVID-19–negative cases had significantly greater percent area of CD61 (platelet) staining (overall P=0.0002) compared with the 3 other groups of COVID-19–positive and –negative thrombus aspirated from patients with STEMI and COVID-19–positive microthrombi. Constituents of thrombi aspirated from COVID-19 and non–COVID-19 patients with STEMI were numerically similar.

• Download figure
• Download PowerPoint

Discussion

COVID-19 infection continues to be a major cause of mortality throughout the world. Evidence of cardiac injury as indicated by elevated levels of cardiac troponin is not uncommon in hospitalized patients and increases risk of death. In 1 case series from New York City involving 18 patients with confirmed diagnosis of COVID-19 and ST-segment–elevation, 44% received a diagnosis of acute coronary thrombosis causing myocardial infarction, whereas 56% had evidence of noncoronary myocardial injury (defined as nonobstructive disease on coronary angiography).⁷ Of the patients with noncoronary myocardial injury, 90% died, suggesting a desperate need for better understanding its pathogenesis and how best to treat them. Here, we report the first systematic analysis of the causes of cardiac injury in subjects dying of COVID-19 infection. In our series, 35% of subjects had evidence of cardiac injury as indicated by myocardial necrosis at autopsy. The most common cause of necrosis was microthrombi, found in 64% of cases with myocyte necrosis. Microthrombi were distinctly different in composition compared with epicardial coronary thrombus aspirates from STEMI cases, consisting of higher levels of fibrin and terminal complement.

Although many studies have focused on pulmonary findings of COVID-19, few pathology studies have been conducted specifically examining the effects of COVID-19 on the heart, and most of these did not describe specific findings related to myocardial injury. In the largest series of autopsies conducted in a New York hospital, hearts from 25 cases were evaluated. Most hearts showed evidence of preexisting atherosclerotic or hypertensive heart disease, with 60% of cases showing nonspecific patchy mild interstitial chronic inflammation within the myocardium without associated myonecrosis.⁸ More recently, Basso et al reported the findings from 21 autopsy cases collected from 4 hospitals around the world and reported a 14.2% (3/21 cases) incidence of myocarditis.⁹ Other pathological series have documented rare findings of lymphocytic myocarditis without clear clinical sequelae, whereas clinical case reports have described myocardial injury presumed to be consistent with myocarditis but without actual tissue diagnosis.^10,11 Indeed, after quantification of viral load in 39 consecutive autopsy cases from Germany, SARS-CoV-2 could be documented in 24 of 39 (61.5%), with 26 of 39 (41%) having copy numbers >1000 copies per microgram RNA.¹² There was no difference in inflammatory infiltrate or leukocyte numbers between individuals with and without cardiac infection. We were able to recover viral RNA in only 20% of hearts studied despite it being detectable in lungs in the vast majority of cases (85%). In our series, there were no differences in the percentage of cases with viral RNA detected in the heart with and without myocardial necrosis (14.3% versus 23.1%, P=0.51), suggesting direct viral invasion of the heart does not play a major role in the development of necrosis.

Although direct infection of the lungs with resulting multifocal pneumonia is thought to be the major cause of death in patients with COVID-19, inflammatory cytokine syndrome may also be an important cause of morbidity and mortality. Li et al reported significantly increased levels of interleukin (IL)–8, IL-6, tumor necrosis factor–α, macrophage chemotactic protein-1, and regulated upon activation, normal T cell expressed and presumably secreted in severe COVID-19 cases, and IL-6 and IL-8 were associated with disease progression.¹³ It is thought that severely ill patients with COVID-19 are at an increased risk for thromboembolic events, including pulmonary microthrombi as well as venous thrombosis, perhaps resulting from cytokine storm. We show the presence of microthrombi in the hearts of COVID-19–related deaths as the leading cause of cardiac injury. An autopsy case series of 32 cases from New York City reported 77% had elevated troponin I. Pathologically, 1 case showed epicardial coronary artery thrombi, and 19% had intramyocardial small vessel thrombi.¹⁴ Furthermore, Bois et al found nonocclusive microthrombi in the small intramyocardial vasculature in 80% (12/15) of COVID-19 autopsy cases.¹⁵ Of note, they reported that only 2 out of 15 patients (13.3%) had acute ischemic injury. A multicenter autopsy study by Basso et al reported acute myocyte injury in the RV and thrombi in small vessels of myocardium in 4 out of 21 cases (19% in both cases).⁹ However, their distribution and association with necrosis were not described. We had previously reported 1 case of microthrombi in a COVID-19–infected young female patient who died of acute myocardial infarction and cardiogenic shock, but now we have expanded our findings in a larger series of cases (n=40) and found that microthrombi are the leading cause of cardiac injury (defined as myocyte necrosis). Rapkiewicz et al also reported autopsy findings of 7 cases with COVID-19, with 5 of the 7 being hospitalized at the time of death and 2 having sudden cardiac death at home.¹⁶ In all 7 cases, fibrin microthrombi were identified in the heart; however, the extent of microthrombi and their location were not specifically identified, nor was the extent of myocardial necrosis. Although staining for complement (C4d) was performed to rule out complement-mediated myocyte damage, and was negative in all cases tested, complement in microthrombi was not specifically examined.

Cardiac microthrombi would not be detectable clinically because no laboratory test can specifically detect microthrombi, but future studies should be directed toward developing methods and laboratory testing to diagnose this type of injury. Overall, there was a significant difference in ischemic EKG changes in those with versus without myocyte necrosis, although sensitivity of such findings was poor, because only 6 of the 14 (42.9%) cases showed myocardial injury by EKG. Clinical presentation may also be misleading, because chest pain may be absent or significantly underreported, especially in patients with from severe respiratory impairment.

We found that microthrombi were distinct in composition compared with epicardial thrombus aspirates from patients with STEMI with and without COVID-19 infection with higher levels of fibrin and terminal complement. A recent study of SARS-CoV, which is closely related to SARS-CoV-2, found disease exacerbation was related to the activation of complement C3.¹⁷ Others have reported that SARS-CoV-2 autoactivates MASP-2 (mannan-binding lectin-associated serine protease 2), the primary enzymatic initiator of the lectin pathway.¹⁸ MASP-2 activation leads to generation of C3 convertase and activation of the membrane attack complex (C5b-9).¹⁹ Moreover, alteration of the MASP-2–binding motif, either by Masp2 deletion or blocking the MASP-2–N protein interaction, attenuated lung injury.¹⁸ Immunohistochemistry analysis of pulmonary autopsy samples revealed MASP-2 and C5b-9 deposition localized in interalveolar septa. These data, along with human proteomic studies, suggest that coronavirus infections are associated with the activation of multiple complement pathways.^20–22

Moreover, inhibiting complement pathway seems to have a therapeutic effect, at least in experimental models.^17,21 C3-deficient mice infected with SARS-CoV exhibited less respiratory problems despite similar viral loads in the lung, with lower levels of cytokines found in the lung and serum.¹⁷ The reduction in lung neutrophils reduced intrapulmonary and plasma IL-6 levels. At present, C3 blockade with agents such as AMY-101 are undergoing clinical trials in patients infected with COVID-19 (NCT04395456).

We should also address some of the limitations of this study. Autopsy material from subjects dying of COVID-19 infection may have its own biases and may not be reflective of cardiac findings in those who survive COVID-19 infection. Thus, the true incidence of cardiac injury and microthrombi may be different in subjects who survive COVID-19 infection. Moreover, the lack of troponin levels in all subjects also may limit the clinical translation of myocardial necrosis detected at autopsy. Although we surveyed all regions of the heart at multiple levels, small areas of necrosis and microthrombi may have been missed because of sampling. Nonetheless, we believe this study reveals important and novel insights about the nature of cardiac injury in subjects dying of COVID-19 infection.

In conclusion, our study is the first to examine systematically the causes of cardiac injury in patients dying of COVID-19 infection. Here, we report that 35% of subjects dying with COVID-19 had evidence of cardiac injury as identified by the presence of myocyte necrosis, with the majority (78.6%) having focal myocyte necrosis. The major cause of myocyte necrosis was cardiac microthrombi, occurring in 64.3% of those with myocyte necrosis. Microthrombi were different in thrombus constituents with richer fibrin and complement compared with intramyocardial thrombi from COVID-19–negative subjects and from thrombi aspirated from coronary arteries of both COVID-19–positive and –negative STEMI cases. Our data suggest that microvascular thrombosis should be entertained as a likely cause of cardiac injury in hospitalized patients with COVID-19 and that further investigation of antiplatelet, anticoagulant, and anticomplement therapies that specifically target microthrombi should be examined in clinical trials.

Supplemental Materials

Data Supplement Figures I–IV

Data Supplement Tables I–III

Footnotes