Abstract
INTRODUCTION: Atrial fibrillation (AF) is associated with an increased risk of stroke. Despite evidence linking cancer and thrombosis, cancer is not part of the CHA(2)DS(2)VASc score. HYPOTHESIS: Cancer is an independent risk factor for thromboembolic stroke in patients with AF. METHOD: The SEER database was utilized to identify patients with lung, colon, breast, and prostate cancers with AF and no prior diagnosis of stroke and. compared to controls within the dataset. The primary endpoint was rates of stroke per 100 person-years. Cox regression modeling and a nested model comparing CHA(2)DS(2)VASc score (Model 1) with a complete model including cancer diagnosis (Model 2) were performed. Models were compared using Akaike Information Criterion (AIC) and Net Reclassification Index (NRI). A propensity-matched cohort with equivalent CHA(2)DS(2)VASc scores determining stroke-free survival was also performed. RESULTS: A total of 101,185 patients were included in the analysis, with 48,242 in the Cancer and 52,943 in the Non-cancer Group. Stroke rate per 100 person-years was significantly higher in the Cancer Group. The CHA(2)DS(2)VASc model (Model 1) was compared against a model including cancer (Model 2) showing improved predictability as assessed by both NRI and AIC. Cox regression analysis calculated a hazard ratio of 1.085 for Cancer, which was comparable to age >75, female sex, and diabetes. Propensity matched Kaplan-Meier curve demonstrated a decreased probability of stroke-free survival in the Cancer Group. CONCLUSION: Cancers increase the risk of stroke in patients with AF. Consideration should be given to the addition of cancer to the clinical scoring system.