Pulmonary thrombosis associated with COVID-19 pneumonia: Beyond classical pulmonary thromboembolism

Abstract

Background: Classical pulmonary thromboembolism (TE) and local pulmonary thrombosis (PT) have been suggested as mechanisms of thrombosis in COVID-19. However, robust evidence is still lacking because this was mainly based on retrospective studies, in which patients were included when TE was suspected. Methods: All patients with COVID-19 pneumonia underwent computed tomography and pulmonary angiography in a prospective study. The main objective was to determine the number and percentage of thrombi surrounded by lung opacification (TSO) in each patient, as well as their relationship with percentage of lung involvement (TLI), to distinguish classical TE (with a random location of thrombi that should correspond to a percentage of TSO equivalent to the TLI) from PT. We determined TLI by artificial intelligence. Analyses at patient level (TLI and percentage of TSO) and at thrombi level (TLI and TSO) were performed. Results: We diagnosed TE in 70 out of 184 patients. Three (2-8) thrombi/patient were detected. The percentage of TSO was 100% (75-100) per patient, and TLI was 19.9% (4.6-35.2). Sixty-five patients (92.9%) were above the random scenario with higher percentage of TSO than TLI. Most thrombi were TSO (n = 299, 75.1%). When evaluating by TLI (<10%, 10%-20%, 20%-30% and >30%), percentage of TSO was higher in most groups. Thrombi were mainly in subsegmental/segmental arteries, and percentage of TSO was higher in all locations. Conclusions: Thrombi in COVID-19 were found within lung opacities in a higher percentage than lung involvement, regardless of TLI and clot location, supporting the hypothesis of local PT rather than "classic TE".