Routinely available inflammation biomarkers to predict stroke and mortality in atrial fibrillation

Abstract

Background: This study aimed to assess the predictive value of 7 routinely available inflammation biomarkers for stroke and all-cause mortality in 229 non-valvular AF patients. Methods and results: C-reactive protein, Albumin (ALB), d-dimer, fibrinogen, the number of platelets, lymphocytes, monocyte and neutrophils were measured. The Multivariable Cox proportional hazard model was used to assess the predictive value of the inflammation biomarkers for stroke and all-cause mortality, the c-statistic, Net Reclassification Improvement (NRI), and Integrated Discrimination Improvement (IDI) were calculated. Lymphocyte Monocyte Ratio (LMR) was the most informative biomarker for predicting stroke, and adding LMR to the CHA2DS2-VASc score did not improve its predictive ability for stroke, but it did improve its reclassification ability. Similar results were observed when comparing LMR+ CHA2DS2-VASc score with the ABC stroke score. For all-cause mortality, a Systemic Inflammation Score (SIS) score was calculated using ALB, fibrinogen and LMR, and adding SIS to the CHA2DS2-VASc score showed a significant improvement in its predictive ability and reclassification ability. There were no significant differences in predictive and reclassification ability for all-cause mortality between SIS+CHA2DS2-VASc score and ABC death score. Conclusions: Adding routinely available inflammatory biomarkers to the CHA2DS2-VASc score increased its ability to predict all-cause mortality, suggesting this cost-effective biomarker strategy may help to improve decision support in AF.