Bleeding and New Malignancy Diagnoses After Anticoagulation for Atrial Fibrillation: A Population-Based Cohort Study

Abstract

BACKGROUND: Bleeding after starting anticoagulation for atrial fibrillation (AF) may be the first sign of malignancy, especially in elderly individuals. There are no recommendations to guide investigations for malignancy after new-onset bleeding after anticoagulation for AF. Our objective was to determine the association of bleeding after starting oral anticoagulation for AF with new diagnoses of malignancy in a population-wide sample. METHODS: We conducted a population-based cohort study using linked administrative data sets of people ≥66 years of age who newly initiated warfarin or direct oral anticoagulants after diagnosis with AF between 2008 and 2022. Follow-up was 2 years after starting anticoagulation. We excluded patients with valvular disease, chronic dialysis, venous thromboembolism, previous cancer, or previously documented bleeding. Bleeding was identified from hospital/emergency department discharge records and physician billings, then handled as a time-varying covariate in cause-specific regression models while adjusting for baseline characteristics. The primary outcome was incident malignancy. We also determined the site of origin of the malignancy and the stage at diagnosis if indicated in the Ontario Cancer Registry. Analyses were repeated while limiting the exposure to specific bleeding sites. RESULTS: Among 119 480 people (mean age, 77.4 years; 52% men) who started anticoagulants, 26 037 (21.8%) had documented bleeding, and 5800 (4.9%) were diagnosed with malignancy within the next 2 years. Bleeding was associated with a higher hazard of cancer diagnosis with a hazard ratio (HR) of 4.0 (95% CI, 3.8–4.3). The HRs for any malignancy were 5.0 (95% CI, 4.6–5.5) for gastrointestinal, 5.0 (95% CI, 4.4–5.7) for genitourinary, 4.0 (95% CI, 3.5–4.6) for respiratory, 1.8 (95% CI, 1.4–2.2) for intracranial, and 1.5 (95% CI, 1.2–2.0) for nasopharyngeal bleeds. The HRs were substantially higher for cancers concordant with the bleeding site (gastrointestinal, 15.4; genitourinary, 11.8; respiratory, 10.1). Cancers were diagnosed at an earlier stage after bleeding (27.6% stage 4 after bleeding versus 31.3% without bleeding; P =0.029). CONCLUSIONS: In anticoagulated patients with AF, bleeding was strongly associated with new cancer diagnoses. Antecedent bleeding was associated with cancer diagnosis at an earlier stage. This highlights the importance of timely investigations in patients with bleeding after anticoagulation for AF, rather than attributing bleeding as an expected adverse effect.