Background: EPIPHANY is the first algorithm to predict serious complications in both suspected and unsuspected cancer-associated pulmonary embolism (PE), overcoming limitations of previous models. Methods: PERSEO is a prospective multicenter study. We recruited cancer patients with both incidental and symptomatic PE treated between Oct. 2017 and Dec. 2019. The primary aim was to determine the percentage of serious complications in patients at low predicted risk, with at least 3% accuracy. We also compared the predictive parameters of EPIPHANY with other available scores for prediction 15-day serious complications and 30-day mortality. Results: Cohort includes 831 patients (men, 58.6%; median age, 66 years). Most frequent tumors were lung (27.1%), colorectal (19%) and breast (7.8%). 78.6% had stage IV disease, and 77.6% were receiving antineoplastic treatment. EPIPHANY classified 27%, 24% and 49% of patients as low, medium and high risk, respectively. The rate of 15-day serious complications increased significantly across these prognostic categories: 2.67 (95% CI 0.6 - 4.8), 8.9% (95% CI 0.5 - 12.8), and 25.9% (95% CI 21.7 - 30.2), for low, intermediate, and high risk patients, respectively (p<0.001, linear-by-linear test). In comparison with other validated scores, EPIPHANY has a higher negative predictive value, lower negative likelihood-ratio, and comparable sensitivity (Table). Conclusions: The EPIPHANY index is able to identify a subgroup of patients with cancer-associated pulmonary embolism at very low risk of serious complications or short-term mortality, with potential implications for decision making.

*For the calculation of the SCORE in these models, incidental EPs have not been taken into account since they are scales developed for symptomatic EP.