Major bleeding in patients with acute pulmonary embolism: results from the COPE study

Background In patients with acute pulmonary embolism (PE), major bleeding (MB) is the counterbalance for anticoagulant and thrombolytic treatment. We aimed to assess independent predictors of MB, evaluate the performance of available bleeding risk scores, and identify bleeding-phenotypes in patients with acute PE from a prospective, multicentre study. Methods The primary outcome was MB defined according to ISTH criteria occurring during 30-day follow-up. MBs were classified as early (<7 days) and late (≥7 days) based on days from PE diagnosis. Results Among 5213 patients with acute PE, 159 had MB (3.1%), 78 as early, and 81 as late events. MB occurred in 12.9% and 2.5% of patients treated versus not treated with thrombolysis. Active cancer, previous bleeding, dementia, anemia, syncope, and thrombolysis were independent predictors of early MB and of MB at 30 days; active cancer, previous bleeding and anemia were independent predictors of late MB. Four bleeding-phenotypes were identified; phenotype#1=previous VTE/estrogen users, #2=unprovoked PE, #3=cancer patients, #4=fragile patients with comorbidities. The risk of MB was lower in the previous VTE (#1) bleeding-phenotype in comparison to the others, with a trend toward increased risk from phenotype #1 to #4. Performance of phenotypes to predict MB at 30-days was similar to that of PE-SARD, BACS and VTE-Bleed. Conclusions In patients with acute PE, risk factors for early and late MB within 30 days differ. Compared to stratification strategies that solely based on bleeding predictors, phenotyping appears to offer similar accuracy in identifying patients at risk for MB.