No Evidence for Classic Thrombotic Microangiopathy in COVID-19

Background: Coronavirus disease-2019 (COVID-19) triggers systemic infection with involvement of the respiratory tract. There are some patients developing haemostatic abnormalities during their infection with a considerably increased risk of death. Materials and Methods: Patients (n = 85) with SARS-CoV-2 infection attending the University Medical Center, Mainz, from 3 March to 15 May 2020 were retrospectively included in this study. Data regarding demography, clinical features, treatment and laboratory parameters were analyzed. Twenty patients were excluded for assessment of disseminated intravascular coagulation (DIC) and thrombotic microangiopathy (TMA) due to lack of laboratory data. Results: COVID-19 patients (n = 65) were investigated, 19 with uncomplicated, 29 with complicated, and 17 with critical course; nine (13.8%) died. Seven patients showed overt DIC according to the ISTH criteria. The fibrinogen levels dropped significantly in these patients, although not below 100 mg/dl. Hallmarks of TMA, such as thrombocytopenia and microangiopathic haemolytic anaemia, were not detected in any of our COVID-19 patients. ADAMTS13 activity was mildly to moderately reduced in 4/22 patients, all having strongly elevated procalcitonin levels. Conclusion: DIC occurred in 7/65 COVID-19 patients but fibrinogen and platelet consumption were compensated in almost all. ADAMTS13 assays excluded TTP and hallmarks of classic TMA were absent in all investigated patients. We hypothesize that the lacking erythrocyte fragmentation and only mild platelet consumption in severe COVID-19 are due to a microangiopathy predominantly localized to the alveolar microcirculation with a low blood pressure gradient.

A massive release of VWF from the endothelial cells, as it occurs in severe inflammatory states and systemic infections, can lead to a mild decrease of VWF-cleaving protease, a disintegrin and metalloprotease with thrombospondin type 1 motif 13 (ADAMTS13) [23]. Whether the resulting VWF/ADAMTS13 dysbalance contributes to the pathophysiology of certain TMAs in a similar way as in classic TTP characterized by a very severe deficiency of ADAMTS13 activity (<5-10% of normal) remains unclear.
We investigated clinical and laboratory patterns in this observational study to understand more about the underlying coagulopathy during COVID-19. In particular, we focused on clinical and laboratory features of DIC and classic TMAs.

Materials and Methods
Patients (n = 85) with confirmed SARS-CoV-2 infection who were seen at the University Medical Center, Mainz, Germany, between 3 March and 15 May 2020 were assessed in this observational study. Patient characteristics and laboratory findings were reviewed retrospectively through the electronic hospital information systems (i.s.h.med ® , SAP, Weinheim Germany, Nexus Swisslab, Berlin, Germany). The retrospective study was approved by German law [Landeskrankenhausgesetz §36 and §37] in accordance with the Declaration of Helsinki and by the local Ethics Committee of "Landesärztekammer Rheinland-Pfalz" (reference numbers: 2020-14988_2).
Severity of COVID-19 was classified by respiratory function into an uncomplicated, complicated, and critical clinical course. Patients with uncomplicated disease required neither monitoring nor oxygen supplementation, whereas patients affected by complicated COVID-19 were in need for oxygen supplementation and critically ill COVID-19 patients needed invasive ventilation. The categorization into the individual COVID-19 severity stages was done retrospectively based on the clinical course during hospitalization.
SARS-CoV-2 infection was confirmed by polymerase chain reaction (PCR) from respiratory samples, employing a PCR kit specific for SARS CoV-2 (Altona Diagnostics GmbH, Hamburg, Germany). All other laboratory assays were performed in the accredited (DIN-ISO 15.189) Institute of Clinical Chemistry and Laboratory Medicine of the University Medical Center, Mainz. Renal injury was assessed based on Acute Kidney Injury Network (AKIN) criteria [24,25]. Presence of DIC was determined according to ISTH guidelines by the DIC score, incorporating platelet count, D-dimer, INR and fibrinogen level [26]. D-Dimer, derived fibrinogen and prothrombin time (PT/INR) were performed on ACL TOP 750 instruments (Instrumentation Laboratory Company, IL, Bedford, MA, USA) using IL reagents (HemosIL D-Dimer HS 500 and HemosIL RecombiPlasTin 2G) and following the manufacturer's instructions. ADAMTS13 activity was examined by the fluorescence resonance energy transfer system (FRETS-VWF73) method [27] modified according to Kremer-Hovinga et al. [28].
Statistical analyses employed SPSS version 22.0 (IBM GmbH, Ehningen, Germany). Descriptive statistics included frequency, mean, standard deviation, median, interquartile range (IQR), minimum and maximum. Explorative group comparisons were performed by t-test or Mann-Whitney-U-test for continuous variables and by Chi-squared test or Fisher's exact test for categorical variables, accordingly. p-values are two tailed, and p values < 0.05 were considered statistically significant.

Patient Characteristics
A total of 85 patients with proven SARS-CoV-2 infection were seen between 3 March and 15 May 2020. For the assessment of haemostatic alterations, 20 patients were excluded due to insufficient laboratory data ( Table 1 and Table S1.1).
As shown in Table 1, 57/65 patients (88%) were treated with prophylactic or therapeutic doses of LMW-heparin from the time of hospitalization. In particular, all patients who developed thromboembolic events during COVID-19 received therapeutic doses of LMW-heparin. One patient had been under rivaroxaban because of deep vein thrombosis and lung embolism that had occurred prior to COVID-19. No major bleeding event was observed during hospitalization and heparin-induced thrombocytopenia did not occur.

Patients with DIC
Overt disseminated intravascular coagulation (DIC), as defined by a DIC Score ≥5 points (26), was present in 7/65 (11%) patients (Table 3 and Table S3, Figure 2).  The dynamics of laboratory values included into the DIC score are shown for the seven patients who developed DIC over the course of hospitalisation (Figure 2). The DIC score became positive after a median of 21 (range 6.5-40) days post hospital admission. DIC diagnosis was mainly based on moderately (0.5-2.0 mg/L) or extremely (>2.0 mg/L) elevated D-dimers and an INR above 1.7. Patients with positive DIC score included only four patients (Pat.ID 24/ 50/ 74/ 79) with thrombocytopenia (platelets < 100/nL) ( Figure 2). Fibrinogen levels were elevated in all seven patients at hospitalisation, followed by a decrease in fibrinogen levels over time ( Figure 2). However, fibrinogen remained ≥100 mg/dl in all seven patients meeting the criteria for DIC (Table 3 and Figure 2). DIC was only diagnosed in male patients with complicated or critical COVID-19, three patients died (Table 1 and Figure 2). DIC patients had more pronounced LDH, total bilirubin and creatinine elevations as compared to the non-DIC group ( Table 2 and Table S2.2). Two patients received low molecular weight heparin in prophylactic doses, three patients in therapeutic doses and one patient continued rivaroxaban introduced prior to COVID-19 infection (Table 1).
It is remarkable that in the four patients with mild to moderately reduced ADAMTS13 activity (18-48%) by far the highest procalcitonin (PCT) values were found. Bacterial coinfections were confirmed in two cases by Escherichia coli and Staph epidermidis isolates from the lower respiratory system and blood stream, respectively.
In sum, there were no diagnostic clues for classic TMA (lacking schistocytes, normal or elevated haptoglobin levels, no severe thrombocytopenia) in any of our patients whereas the consistently high VWF levels were compatible with an endothelial activation and/or damage.

Discussion
We present here our single-centre cohort of all consecutive, retrospectively included patients diagnosed with COVID-19 at the University Medical Center Mainz during the first wave of the pandemic from 3 March to 15 May 2020. Our report focuses on the clinical and laboratory abnormalities related to thrombosis and haemostasis. Multiple scientific publications pointing to the high thrombotic risk in COVID-19 [9][10][11][12][13]32], trying to understand its pathophysiology [33][34][35], suggesting the prognostic value of haemostatic laboratory parameters [4,36,37] and proposing prophylactic and/or therapeutic measures to improve the outcome have been provided [11,38].
From 65 of the 85 registered COVID-19 patients sufficient laboratory data were available to delineate the frequency and extent of haemostatic abnormalities ( Table 1). Nine of the patients died, 3 of 29 with complicated and 6 of 17 with critical disease. No autopsies were performed. The incidence of clinically manifest thromboembolic events was rather low and included three venous TE, one mesenteric arterial infarction and three acute coronary syndromes needing percutaneous coronary intervention (Table 1), which is substantially lower than reported by several authors [12,13]. Whereas initially published cohorts from China had not received TE prophylaxis [9], other series of patients had a cumulative incidence of TE events up to about 15-60% (dependent on COVID-19 severity and length of hospitalisation) despite prophylactic or even higher-than-prophylactic doses of LMWH given [13]. Major bleeding events in patients with or without anticoagulation were generally rare [39].
Routine haemostatic laboratory parameters showed strongly elevated fibrinogen and D-dimer levels, the latter being exceedingly high in those with critical disease (Table 2, Figure 1). Prothrombin times expressed as INR were mildly elevated in critical patients and platelet counts were mostly normal and sometimes subnormal independent of COVID-19 severity. Calculating the DIC score [26] taking into account the highest D-dimer and INR and the lowest fibrinogen and platelet count values showed seven of 65 patients having a score ≥5 signalling overt DIC (Table 3 and Table S3). The course of laboratory parameters in these seven individual patients shows a notable difference of their "DIC pattern" as compared to "typical DIC" associated with bacterial sepsis, obstetric complications and other inflammatory conditions [40,41]. Six of our DIC patients showed a mild drop of fibrinogen (but none <100 mg/dL), platelet count fell below 50/nL in only two, and none of the seven showed any abnormal bleeding tendency. These data resemble observations by other groups [20,37] and new designations for the COVID-19 associated haemostatic disturbances have been proposed, i.e., "COVID-19 associated coagulopathy (CAC)" (18) or "Pulmonary intravascular coagulopathy (PIC)" [42] to stress the discrepancy to "classic DIC" [43]. Experts in this field have discussed the current evidence and suggested that CAC during severe COVD-19 should be considered as a prothrombotic phenotype of DIC [44].
Other investigators have labelled the coagulopathy in COVID-19 as thrombotic microangiopathy (TMA) [30,31,[45][46][47][48]. Nevertheless, the hallmarks of classic TMA, consumptive thrombocytopenia and microangiopathic haemolysis with erythrocyte fragmentation resulting in schistocytes in the peripheral blood smear [21], have neither been found in any of our 22 patients subjected to detailed investigation for TMA (Table 4) nor in most studies from other investigators [49,50]. Mildly or moderately decreased ADAMTS13 activity (18-48% of normal in 4 of the 22 patients, the other 18 displaying normal activity >50%) ( Table 4) as well as normal [50] or subnormal to normal levels [30,45,49] in several previously described patients clearly ruled out thrombotic thrombocytopenic purpura (TTP), which is characterized by <10% (and indeed often <1%) ADAMTS13 activity [51]. Most notable findings in the 22 patients were the highly elevated values of VWF activity and antigen and the increased ratio of VWF:antigen/ADAMTS13 activity (Table 4). Whether this imbalance between high VWF concentrations and low levels of its size-regulating protease in our patients, features that have also been described in patients with severe sepsis or septic shock [23,52], is pathophysiologically relevant in COVID-19 remains questionable, especially in the absence of consumptive thrombocytopenia. Henry et al. reported that a decreasing ratio of ADAMTS13:act/ VWF:Ag in 52 COVID-19 patients at presentation to the emergency room was predictive for the development of acute kidney injury and a severe form of COVID-19 [48]. However, the decreased ratio of ADAMTS13:act/VWF:Ag in 12 of 52 patients, which did not correlate with platelet count, was mainly due to high VWF:Ag levels. This may, in turn, primarily reflect an augmented VWF release caused by endothelial injury [48]. It is well possible that the highly elevated FVIII:C levels, paralleling the increased VWF values in COVID-19 [49,50] are equally or more relevant in mediating a prothrombotic effect. The explanation for the consistently and often massively elevated VWF values is likely explained by endothelial activation and damage [14,17,50], the latter caused by direct endothelial invasion by the SARS-CoV-2 [14,53,54]. The autopsy study by Ackermann and colleagues comparing the lungs of seven patients who died from COVID-19 with those of seven who died from acute respiratory distress syndrome caused by influenza A infection and 10 age-matched control lungs showed distinctive morphologic vascular features in COVID-19. These findings included severe endothelial injury, widespread alveolar capillary fibrinous microthrombi (9-times more prevalent than in influenza-infected lungs) and marked features of (mainly intussusceptive) angiogenesis with formation of new vessels growing into the lumen of existing vessels [14]. Beigee et al. described diffuse alveolar damage and thrombotic microangiopathies in lung biopsies of 31 patients who had died from COVID-19 [16]. Moreover, other autopsy studies confirmed extensive fibrinous microthrombi in the lungs [53,55] and also in some skin lesions [53], from patients succumbed to severe COVID-19. The latter study demonstrated, in addition, significant microvascular deposition of complement activation products, C5b-9, C4d and mannose binding lectin-associated serine protease 2, in colocalization with SARS-Cov-2, hinting at systemic activation of the alternative and lectin-mediated complement pathways [53]. Thus, these studies provide unequivocal proof of "thrombotic microangiopathy", mainly in the lungs. A study of 50 children with COVID-19 including 18 with a multisystem inflammatory syndrome in children (MIS-C) suggested the presence of TMA in 17 of 19 with complete laboratory evaluation. Many showed evidence of systemic complement activation, i.e., increased sC5b-9 levels [47]. Furthermore, first studies describe the successful use of complement inhibitors in COVID-19 patients [56].
Whether the (almost uniformly) lacking laboratory hallmarks of TMA in our COVID-19 patients and those described in other cohorts [49,50] is explained by the predominantly pulmonary microangiopathy in COVID-19 remains to be investigated. While both VWFmediated platelet adhesion and aggregation mainly in the microcirculation of brain, kidney and heart as seen in TTP [51] and fibrin microthrombi located predominantly in the kidney as seen in HUS [21,22], will result in erythrocyte fragmentation in the partially occluded microcirculation with the high blood pressure gradient in the arterial circulation, it may be hypothesized that the "COVID-19-associated thrombotic microangiopathy that is often restricted to the pulmonary microcirculation will not produce schistocytes owing to the much lower blood pressure gradient".

Limitations of the Study
Our study has several limitations. Due to the retrospective approach, relevant parameters were not monitored throughout the course of COVID-19. In particular, we did not assess markers of complement activation, which may play a pathophysiologic role in the prothrombotic alterations of COVID-19 [53] and inhibition of complement activation may be targeted therapeutically [56]. In addition, several other tests, such as viscoelastic methods to study clot formation and fibrinolysis were not performed in this COVID-19 cohort. Whether these latter methods could be useful to identify the thrombotic risk in individual patients and/or to guide antithrombotic treatment remains to be further investigated [57].
Another limitation is a bias regarding the severity of COVID-19. Mostly severely symptomatic patients presented at our University Medical Center and were admitted as inpatients and laboratory data on haemostatic alterations were not available from all patients. Therefore, our observational study is merely hypothesis-generating.

Conclusions
In sum, our observational data on the haemostaseologic abnormalities in real world adult COVID-19 patients admitted to a single academic centre add results suggesting that the haemostatic alterations in severe COVID-19 are not fully fitting into the established categories of DIC and classic TMA. Further pathophysiologic research is needed to provide clinically useful targets for intervention to improve patient outcome.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical and privacy restrictions.