key: cord-0889684-p1hctn1d
authors: Ketai, Loren H.
title: Deciphering the Risk of Thromboembolism in COVID-19
date: 2021-07-13
journal: Radiology
DOI: 10.1148/radiol.2021211563
sha: dee8b6567315efdaf4ca04a5e0fbdb8ba67d0ae7
doc_id: 889684
cord_uid: p1hctn1d

nan

Over the last year a half, the medical literature has demonstrated an increase in pulmonary embolic disease among patients with COVID-19. The results are convincing. Increases in incidence have been most marked among patients with COVID-19 admitted to the intensive care unit (ICU), where the frequency of pulmonary emboli exceeds that observed in other ICU patients or in patients hospitalized with other viral pneumonias (1, 2) . This is not surprising considering the thromboinflammation associated with COVID-19 infection. Thromboinflammation in COVID-19 manifests as elevated levels of procoagulants (such as von Willebrand factor) and endothelial dysfunction, which diminishes the protective antithrombotic activity of the endothelium.

In this issue of Radiology, the current study by Riyahi et al (3) reports the results of a detailed retrospective study of a very large cohort of patients hospitalized with COVID-19. The evaluation for pulmonary embolism (PE) was driven by conventional clinical assessment, with the vast majority of patients undergoing CT pulmonary angiography (CTPA). The overall incidence of PE among hospitalized patients, 25% (102 of 413) was commensurate with that seen in other metanalyses (1, 4) . The authors did not find evidence of a difference between the incidence of PE among ICU patients with COVID-19 (29%) versus non-ICU COVID-19 patients (24%) (p=.37). This differs from some meta-analyses that have observed a higher incidence of pulmonary embolism among those patients with COVID-19 admitted to the ICU compared to those imaged from the emergency department or general wards (4).

The authors also analyzed multiple variables potentially associated with pulmonary emboli among hospitalized patients with COVID-19 who underwent CTPA. In this subgroup of patients with COVID-19, analysis with a random forest model (a classification system constructed from multiple decision trees) identified d-dimer as the dominant predictor. Construction of a receiver operating characteristic curve and calculation of the Youden index identified a d-dimer>1600 ng/mL as the optimal cutoff for differentiating patients with PE from those without. In the derivation group, the sensitivity and specificity of this value were 82% and 68% respectively. When applied to a separate external validation group of patients with COVID-19 from another hospital, the sensitivity of that d-dimer threshold predicted PE with 100% and specificity 62%.

Echocardiography was performed in less than one-third of the patients whose CTPAs were positive for PE. The study demonstrated right ventricular (RV) strain in 26% of these patients and the presence of strain was associated with a higher semiquantitative (Qanadli) embolic burden. The authors did not report the presence of right ventricular strain (as defined by right to left ventricular diameter ratio) or other CT findings.

While this study confirms several important clinical observations regarding thromboembolism and COVID-19, key practical questions remain unanswered. Not the least important, is how to best identify patients with COVID-19 who will benefit from pulmonary CTPA. This is particularly difficult to determine given the large majority of patients in some series, 3718 in this paper, who undergo neither CTPA nor nuclear medicine perfusion scintigraphy scanning. A study reporting low rates of venous thromboembolism during the 90 days following hospitalization for COVID-19 suggests that the incidence of untreated PE among COVID-19 may be small (5) . Nevertheless, 90-day follow-up of cohorts such as the one reported in this paper would be important to confirm that the incidence of PE or deep vein thrombosis is low among patients not selected by clinical criteria to undergo imaging evaluation for pulmonary embolism.

Even when pulmonary emboli are detected by CTPA or perfusion scintigraphy, their clinical significance in the setting of COVID-19 may be ambiguous. Two-thirds or more of pulmonary emboli seen in patients with COVID-19 are segmental or subsegmental, as opposed to approximately half in other patients. This observation and the decreased frequency of concurrent peripheral deep venous thrombosis have led some authors to suggest that a subset of the observed pulmonary vascular clot represents inflammation-mediated in situ thrombosis rather than emboli (7) . Anticoagulation may be less effective in preventing clots caused by this thromboinflammation. Lower efficacy of anticoagulation to modulate this process may have contributed to the recent failure of more aggressive prophylactic anticoagulation to diminish mortality in patients with COVID-19(6).

Both in situ pulmonary thrombi and venous thromboembolism likely contribute to the widespread elevation of d-dimer seen in COVID-19. Thus, patients with COVID-19 have a higher threshold than patients without COVID-19 below which a d-dimer level helps "rule out" pulmonary emboli. For patients without COVID-19, an age-adjusted d-dimer level (500 ug/L for patients < 50 years old, age x 10 for patients >50 years old) can help exclude pulmonary embolism. The current study suggests that this threshold is approximately 3 times higher among patients with COVID-19. Unfortunately, the widespread elevation of d-dimer in COVID-19 also degrades d-dimer's specificity and therefore its utility as a screening laboratory study. The reported specificity of 68% in patients undergoing CTPA or perfusion scintigraphy in this series, would likely have been lower had the >1600 ng/mL threshold been applied to the entire 4131 inpatients testing positive for COVID-19. Prior studies suggest that thresholds above 3000 ng/mL would be needed to consider d-dimer a specific test for pulmonary emboli in COVID -19 if widely applied (4).

An underlying infection with COVID-19 may also confound the interpretation of signs of right ventricle (RV) strain seen on CT. RV dysfunction and pulmonary systolic hypertension are common in severely ill patients with COVID-19, including those without thromboembolic disease (8) . In that setting, RV dysfunction is a major predictor of mortality. Despite the association of RV strain seen on echocardiography with higher Qanadli scores observed in the current study, evidence of RV strain on echocardiography is unlikely to be driven solely by embolic load in patients with severe COVID-19. Also, the accuracy of CT findings of RV strain, as opposed to echocardiographic ones, is not yet well established in the setting of COVID-19 and may be complex. Even among patients with pulmonary emboli who do not have COVID-19, the association between a higher ratio of RV diameter to left ventricular (LV) diameter on CTPA and patient outcome is not binary. For instance, the risk of death or other adverse outcome is much greater among patients whose RV/LV ratio exceeds 1.2 compared to those whose RV/LV ratio exceeds 1. (9) . Among patients with pulmonary emboli and severe COVID-19 infection, RV size could be affected by one or both diseases, altering the correlation of specific RV/LV ratios with adverse patient outcomes. This paper and those that have preceded it demonstrate that thromboembolic disease in the setting of COVID-19 presents unique challenges to the medical community. CT evidence of pulmonary vascular thrombus, serum levels of d-dimer, and assessment of RV function can be interpreted somewhat differently in the setting of COVID 19 infection. Moreover, generalizable conclusions from the medical literature to date can be difficult given regional variations in patient demographics, reliance on retrospective studies with different diagnostic algorithms for imaging suspected emboli, and the use of varying anticoagulation regimens. Additional knowledge will accumulate, but for now, we will have to do what we can with what we've got where we are (10). 

Pulmonary Embolism at CT Pulmonary Angiography in Patients with COVID-19. Radiol Cardiothorac Imaging

Pulmonary Embolism and Deep Vein Thrombosis in COVID-19: A Systematic Review and Meta-Analysis

Pulmonary Embolism in Hospitalized Patients with COVID-19: A Multicenter Study

Pulmonary embolism in patients with COVID-19 and value of Ddimer assessment: a meta-analysis

Postdischarge venous thromboembolism following hospital admission with COVID-19

Effect of Intermediate-Dose vs Standard-Dose Prophylactic Anticoagulation on Thrombotic Events, Extracorporeal Membrane Oxygenation Treatment, or Mortality Among Patients With

Admitted to the Intensive Care Unit: The INSPIRATION Randomized Clinical Trial

Pulmonary Thromboembolism in COVID-19: Venous Thromboembolism or Arterial Thrombosis? Radiol Cardiothorac Imaging

Utility of Non-invasive Cardiac Imaging Assessment in Coronavirus Disease 2019. Front Cardiovasc Med

CT signs of right ventricular dysfunction: prognostic role in acute pulmonary embolism