Researchers from Massachusetts General Hospital in Boston have identified several markers at presentation that predict coagulation-associated complications in patients with COVID-19: elevated D-dimer, C-reactive protein (CRP), platelet count, and erythrocyte sedimentation rate. The findings were published in Blood by lead author Hanny Al Samkari, MD, and his colleagues. Incorporating these markers, which are easy to obtain in most centers, into predictive modeling at presentation may be crucial to reducing the substantial morbidity associated with bleeding events in patients hospitalized with a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, Dr. Al-Samkari noted.
"The treating clinician can integrate this information into decision-making regarding anticoagulation intensity, thrombosis monitoring, and other aspects of care," he added.
The study relied on patient data from the large multi-institutional Research Patient Data Registry at Partners Healthcare (now known as Mass General Brigham). Dr. Al-Samkari and colleagues retrospectively identified 400 adults in the registry who had a confirmed COVID-19 diagnosis from March 1 through April 5, 2020, and a D-dimer test performed at hospital presentation.
Data on patient demographics, comorbidities, bleeding events, types of anticoagulation administered, arterial and venous thrombotic events, and coagulation and inflammatory parameters were collected. In addition, the researchers gathered data on clinically significant non-vessel thrombotic complications that were possibly reflective of a systemic hypercoagulable state.
Coagulation-based laboratory parameters, laboratory inflammation measures, and high-sensitivity cardiac troponin levels were compared between patients with bleeding or thrombotic complications (i.e., composite of venous thromboembolism [VTE], arterial thromboembolism, and clinically significant non-vessel thrombotic complications) and those without bleeding or thrombotic complications.
The study included a total of 256 non-critically ill patients (mean age = 60 years) and 144 critically ill patients (mean age = 65 years), who had a total of 1,608 patient-days and 1,618 patient-days available for analysis, respectively. Most of the non-critically ill patients (80.1%) had been discharged by the end of the study period, whereas only 12.5% of the critically ill patients had been discharged by this time. A higher proportion of patients who were critically ill had died during the study period compared with non-critically ill patients (18.8% vs. 0.8%, respectively).
In addition, more patients who were not critically ill were admitted with cardiovascular disease (32.0% vs. 29.9%), chronic lung disease (20.7% vs. 17.4%), and chronic liver disease (5.1% vs. 4.2%), while more non-critically ill patients had resided in a nursing home or long-term care facility (8.2% vs. 4.9%). Critically ill patients presented with more diabetes (40.3% vs. 25.4%), immune compromise or suppression (13.9% vs. 11.3%), and chronic kidney disease on dialysis (4.2% vs. 1.6%).
Among the entire study population, the radiographically confirmed rate of VTE was 4.8% (or 4.13 events per 100 patient-weeks), while the overall rate of VTE, including suspected VTE, was 6.0% (or 5.22 per 100 patient-weeks). All patients except for 1 were treated with standard prophylactic doses of unfractionated heparin or low-molecular-weight heparin at the time of their VTE event.
The overall rate of arterial thrombosis was 2.8% (or 2.39 per 100 patient-weeks), and the researchers reported an overall thrombotic complication rate of 9.5% (or 9.78 per 100 patient-weeks). Approximately 10% of patients (n=41) shifted from prophylactic doses of anticoagulation to therapeutic doses during admission to manage thrombotic complications or new-onset atrial fibrillation.
The researchers also noted that 4.8% of patients experienced a bleeding event: 3.1% of non-critically ill patients and 7.6% of critically ill patients. Major bleeding occurred in 2.3% of patients.
Thrombocytopenia was relatively uncommon, the researchers added. Approximately 10.3% of patients had a platelet count <100Ã—109/L, and 2.5% had a platelet count <50Ã—109/L during their hospital course. Four patients with a platelet count <50Ã—109/L experienced grade â‰¥2 bleeding events.
Next, the investigators conducted a multivariable analysis to determine which factors at presentation were associated with thrombotic complications and observed that 3 factors were predictive:
- D-dimer >2,500 ng/mL (adjusted odds ratio [aOR] = 5.28; p=0.003)
- platelet count >450Ã—109/L (aOR = 3.29; p=0.045)
- CRP elevation >100 mg/dL (aOR = 3.04; p=0.015)
Dr. Al-Samkari noted that it is difficult to determine how D-dimer levels are being used in clinical practice, and the way each center approaches this is different and rapidly evolving. "Centers across the world are incorporating D-dimer levels into all manner of institutional pathways for the management of COVID-19, from determination of prophylactic anticoagulation dose to level-of-care triage," he said. "Some centers are not using D-dimer to make clinical decisions regarding these patients."
Limitations of this study include its retrospective and observational design, the inclusion of "suspected" VTE events that were not radiologically confirmed, and the lack of uniformly available data for every evaluated laboratory measure.
"It is important to recognize that many critically ill patients with COVID-19 cannot undergo radiologic imaging to confirm VTE due to clinical instability and logistical reasons, so some VTE events may have been missed," added Dr. Al-Samkari. Also, the study may have missed some pulmonary hemorrhages, which could have been misinterpreted as worsening pneumonia. "Routine diagnostic bronchoscopies to diagnose hemorrhage were likely not being performed with typical frequency in these patients due to the concerns for aerosolizing virus particles during the procedure," he said.
The authors report no relevant conflicts of interest.
Al-Samkari H, Karp Leaf RS, Dzik WH, et al. COVID and coagulation: Bleeding and thrombotic manifestations of SARS-CoV2 infection. Blood. 2020 June 3. [Epub ahead of print]
According to the report, elevations of D-dimer, CRP, and sedimentation rate were predictive of thrombosis, critical illness, and death. Interestingly, a D-dimer level greater than 5 times the upper limit of normal also was associated with a risk of bleeding, which may indicate that these patients were treated with a different intensity of anticoagulation.
Bleeding outcomes can be difficult to identify without manual chart review and the initial reports of patients with COVID-19 focused on thrombosis, with wide variation in incidence reported. This study adds to the literature about the bleeding complications in this setting, with Dr. Al-Samkari and co-authors finding that 3.5% of patients admitted to the hospital ward had a confirmed or suspected VTE and 3.1% had bleeding events. Only one major bleed occurred.
The results from the study confirm that critically ill patients have higher risk for both bleeding and clotting complications. Bleeding events occurred in patients not treated with anticoagulation and in patients treated with therapeutic-intensity anticoagulation. Other treatment decisions and patient characteristics also may influence the risk of bleeding and clotting, which cannot be accounted for in observational studies.
Evidence from randomized trials is needed to understand an anticoagulation intensity that balances the risks of bleeding and clotting in patients with COVID-19, and if anticoagulation can prevent progression of organ failure. Other strategies, including antiplatelet medications and immune modulation, are also being studied. A mechanistic understanding of how viral infection, inflammation, and thrombosis occurs may help us determine why some people can have asymptomatic SARS CoV-2 infection and others become critically ill.
Lisa Baumann Kreuziger, MD, MS
Versiti Blood Research Institute
Medical College of Wisconsin