Risk Stratification of Nonvalvular Atrial Fibrillation

Abstract

We read with interest the article by Keceoglu et al evaluating the possible association of elevated eosinophil count with thrombosis in 89 consecutive patients with nonvalvular atrial fibrillation (NVAF). The 40 patients in whom a thrombus in the left atrium or left atrial appendage was found through transesophageal echocardiography presented higher eosinophil count levels (a 2-fold increase: 233.0 + 30.7 10/mL vs 118.9 + 11.8 10/mL; P < .001). Furthermore, eosinophil count remained an independent predictor of the presence of thrombus on binary logistic regression (odds ratio 1⁄4 1.008; confidence interval95% 1.00-1.01; P < .001). As the authors say in the article, ‘‘the traditional thromboembolic risk classification methods concentrate on the hemodynamics of the left atrium but do not focus sufficiently on the components forming the thrombus.’’ It has been known for long that dense spontaneous echocardiographic contrast, a precursor and strong predictor of the formation of thrombi in left atrium or left atrial appendage, seems to be composed of erythrocyte, platelet, and leukocyte aggregates. Therefore, there is a strong rationale for assessing the relationship of blood cells with the prothrombotic status in NVAF. The use of biomarkers in the risk stratification of NVAF is currently a hot topic, with the recent demonstration of troponin I and NTproBNP as independent predictors of thromboembolic events and as capable of improving the discriminative capability of current risk classification schemes. Our group has recently found an association between increased mean platelet volume or mean corpuscular volume and markers of left atrial stasis in NVAF: left atrial appendage thrombi (as also described by the article authors, concerning mean platelet volume), dense spontaneous echocardiographic contrast, and reduced flow velocities in the left atrial appendage. However, we found no association between eosinophil count and markers of left atrial thrombi in any of the investigations, most probably due to a much lower prevalence of this echocardiographic finding in our samples (8.5% to 12.2% vs 44.9%) which may have led to lack of statistical power to assess differences concerning eosinophil count as a predictor. One point that may merit clarification is the role of other white blood cells (eg, basophils, neutrophils, monocytes, lymphocytes, etc) in thrombogenesis, since the white blood cell count is significantly increased in patients with thrombi (9389 + 574 vs 8108 + 320; P < .001), eosinophils account only for a very small percentage of that count, and no information is provided concerning the other white blood cell types. Another aspect that could justify evaluation is the levels of C-reactive protein, since they were recently shown to be associated with transesophageal echocardiogram changes in this subset of patients. It would be interesting to clarify whether the association of eosinophils with thrombi is independent of this marker. These findings confirm a possible role of blood cells in the thrombogenic process of NVAF. At first, the most immediate clinical application may be the prediction of left atrial thrombi, in order to select patients with a higher probability of thrombi for transesophageal echocardiogram before cardioversion or percutaneous ablation. Second, these biomarkers may have a role in decisions concerning anticoagulation. However, despite knowing that the presence of markers of left atrial stasis is a predictor of thromboembolism and adverse prognosis in the NVAF setting, the association of these markers (both cell counts and morphologic aspects) with thromboembolic events still remains to be categorically proven. Therefore, studies with clinical end points to elucidate the role of blood cells for the refinement of the existing thromboembolic risk classifications (CHADS2 and CHA2DS2-VASc), which have, at best, a low to moderate discriminative capability, are an important topic of research to pursuit in the next few years.