COVID-19-associated hyperviscosity: a link between inflammation and thrombophilia?

Reports of thrombotic complications in patients with COVID-19 are increasingly prominent, and these reports include patients receiving therapeutic anticoagulation.1, 2 At our institution, multiple occurrences of anticoagulation failure prompted us to search for alternative aetiologies contributing to refractory hypercoagulability. Here we describe COVID-19-associated hyperviscosity, a potentially severe consequence of infection with severe acute respiratory syndrome coronavirus 2, in 15 patients tested to date. This work was done ethically in accordance with institutional review board approval.

All patients were critically ill with COVID-19 pneumonia and admitted to the medical intensive care unit. 14 patients had acute respiratory distress syndrome requiring intubation, 14 patients were encephalopathic, 12 patients had shock requiring vasopressors, and 11 patients had renal failure requiring continuous renal replacement therapy (CRRT). All patients received anticoagulation according to an institutional protocol based on data suggesting increased venous thromboembolism rates when D-dimer concentrations exceed 3 μg/mL. Five patients with D-dimer concentrations of 3 μg/mL or higher and known (or highly suspected) thrombosis received therapeutic anticoagulation. Two of the patients received intravenous heparin, and three patients received a direct thrombin inhibitor (argatroban or bivalirudin) because of heparin resistance or concern for heparin-induced thrombocytopenia. Four patients with D-dimer concentrations below 3 μg/mL received low-dose thromboprophylaxis with low-molecular-weight heparin (LMWH) or subcutaneous heparin. Six patients with D-dimer concentrations of 3 μg/mL or more and without known thrombosis received intermediate dosing (ie, subtherapeutic) of LMWH or intravenous heparin.

The 15 patients had plasma viscosity exceeding 95% of normal, as determined by traditional capillary viscometry, ranging from 1·9–4·2 centipoise (cP; normal range 1·4–1·8). Notably, the four patients with plasma viscosity above 3·5 cP had a documented thrombotic complication: one patient had pulmonary embolism, one patient had limb ischaemia and suspected pulmonary embolism, and two patients had CRRT-related clotting. Plasma viscosity and Sequential Organ Failure Assessment scores, a measure of illness severity, were strongly correlated (Pearson's r=0·841, R 2=0·7072, p<0·001; appendix).

Hyperviscosity damages endothelium and is a known risk factor for thrombosis. It can result from increases in cellular components or plasma proteins, such as fibrinogen or immunoglobulin, as seen in Waldenström macroglobulinaemia. Consistent with reports of hyperfibrinogenaemia in patients with COVID-19, our patients had substantially increased fibrinogen concentrations (median 708 mg/dL, range 459–1188; normal reference range 200–393). Further study is needed to evaluate which plasma components, including acute phase proteins such as fibrinogen, contribute to COVID-19-associated hyperviscosity.

Our novel observation might provide an important link between inflammation and coagulopathy in critically ill patients with COVID-19. We are actively exploring any beneficial role of therapeutic plasma exchange, a highly effective treatment for symptomatic hyperviscosity in other conditions such as hypergammaglobulinaemia, in the clinical management of these patients. Finally, any causal relationship between hyperviscosity and thrombotic complications in COVID-19 warrants immediate investigation given the potential to impact clinical care.