Cancer incidence during follow-up in patients with new-onset atrial fibrillation treated with DOACs and its impact on bleeding risk

Abstract

Abstract Background Cancer is increasingly recognized as strictly related to atrial fibrillation (AF). In patients with AF, the relationship between cancer and cardioembolic or bleeding risk during oral anticoagulant therapy is unknown. Purpose To assess the bleeding and ischaemic burden of a baseline or newly diagnosed cancer in patients treated with direct oral anticoagulants (DOACs) for non-valvular atrial fibrillation (NVAF). Methods All consecutive patients treated with DOACs were enrolled among those with new-onset atrial fibrillation and indication for oral anticoagulant between January 2017 and March 2019. During follow-up, bleeding events, newly diagnosed primitive or metastatic malignancy and major cardiovascular events (MACE) were evaluated. At baseline, CHA2DS2-VASc, HAS-BLED, ATRIA, and ORBIT scores were used to assess the hemorrhagic and ischaemic risk. Major bleedings (MB) were defined according to the ISTH definition. Anemia was defined as haemoglobin levels below 11 g/dL in women and 12 mg/dL in men. Results 1258 patients constituted the study population and followed for a mean time of 21.6±9.5 months. Overall, 66 patients (5.2%) were affected by malignant neoplasia at baseline, whereas 59 (4.7%) were diagnosed with a malignancy during follow-up. Among baseline characteristics, anemia was associated with cancer at enrolment (43.9% vs 22.5%, p<0.001) but not at follow up (29.3% vs 23.4%, p=0.341). MACEs were not associated with cancer at baseline (5.3% vs 5.2%, p=1.0) and at follow up (5% vs 4.9%, p=1.0). No association was observed between major ischaemic events and cancer at enrolment or follow up (5.3% vs 4.4%, p=0.83 and 4.4% vs 5%, p=0.82). Despite no statistically significant differences in haemorrhagic risk at baseline, the overall bleeding events and MB were associated with newly diagnosed cancer (9.2% vs 3.9%, p=0.001 and 13.8% vs 4.5%, p=0.001, respectively) but not at baseline (5.2% vs 5.5%, p=0.82 and 9.2% vs 5.2%, p=0.162). At multivariate analysis adjusted for age, hypertension and renal function, anemia and a newly diagnosed cancer during follow up remained independent predictors of MB (respectively, HR 1.27, 95% CI 1.52–1.06, p=0.009 and HR 3.53, 95% CI 7.71–1.62, p=0.001). Conclusion Bleeding risk assessment is an ongoing challenge in patients with NVAF on DOACs. During follow-up, newly diagnosed primitive or metastatic cancer is a strong predictor of bleeding regardless of baseline haemorrhagic risk assessment. In contrast, such association is not observed with malignancy at baseline. A proper diagnosis and treatment could therefore decrease cancer-related bleeding risk. On the contrary, our study shows that cancer is not an ischaemic risk modifier, either diagnosed at baseline or follow-up. Funding Acknowledgement Type of funding sources: None.