Importance: Although the YEARS algorithm is a safe and efficient way to rule out acute pulmonary embolism (PE), robust evidence on its accuracy in patients with cancer is lacking, and current guidelines suggest proceeding directly to computed tomographic pulmonary angiography (CTPA).
Objective: To compare the safety and efficiency of the YEARS algorithm with CTPA only to rule out acute PE in patients with active cancer.
Design, setting, and participants: The Hydra study was an open-label, randomized, investigator-initiated, noninferiority trial with blinded central outcome adjudication conducted from August 22, 2019, to August 21, 2025, the date of final follow-up. Patients with active cancer and suspected acute PE were recruited from emergency departments or medical units in 21 hospitals in the Netherlands, Italy, Switzerland, Belgium, France, and Spain.
Interventions: Patients were randomly assigned in a 1:1 ratio to receive diagnostic management by the YEARS algorithm (n = 352)-consisting of assessing YEARS items, D-dimer levels, and performing risk-dependent CTPA-or by CTPA only (n = 346).
Main outcomes and measures: The primary outcome was centrally adjudicated symptomatic venous thromboembolism or (possible) PE-related death within 90 days after ruling out PE at baseline, assessed in a per-protocol noninferiority analysis with a 2.6% margin for the upper bound of a 1-sided 99.9% CI. The key secondary outcome was the proportion of negative CTPA results at baseline, assessed in a superiority analysis.
Results: A total of 698 patients were randomized (median age, 65 years [IQR, 56-72 years], 422 female [60%]), and 104 patients (15%) had PE diagnosed at baseline. One patient was lost to follow-up. Of those in whom PE was considered excluded, 5 of 282 patients (1.8%) in the per-protocol YEARS group vs 15 of 273 patients (5.5%) in the per-protocol CTPA-only group had a primary outcome (absolute risk difference, -3.7%; 99.9% CI, -8.8% to 1.4%; P = 3.4 × 10-5 for noninferiority). In the intention-to-diagnosis analysis, the absolute risk difference between the YEARS algorithm and CTPA only was -2.6% (99.9% CI, -7.5% to 2.4%; P = 5.9 × 10-4 for noninferiority). Diagnostic management of PE was carried out for 77 of 352 patients (22%) in the YEARS group without CTPA. No difference in the proportion of negative CTPA (P = .93) was observed between the groups.
Conclusions and relevance: In patients with cancer and suspected PE, a diagnostic strategy using the YEARS diagnostic algorithm was as safe as using CTPA only, thus, obviating the need to perform CTPA in 22% of patients.
Trial registration: ccmo.nl Identifier: NL-OMON52383.
