A Multiple Logistic Regression Model as an Additional Mathematical Method for Predicting the Development of Ischemic Stroke in Patients with Atrial Fibrillation

Abstract

Prevention of thromboembolic complications in cases of atrial fibrillation (AF) and, above all, ischemic stroke (IS), represents a key problem of modern cardiology. The aim of the present study was to assess the feasibility of Multiple Logistic Regression Analysis in predicting the occurrence of IS in AF patients with the predictor genotypes of the FGB, GPIα, and GPIβα genes, in order to implement an approach to primary prevention and personalized treatment. Methods and Results: We examined 43 patients with atrial fibrillation and IS in their histories and 78 patients with AF without IS. A total of 188 persons without AF were included in the control group. The present study showed that the homozygote minor allele genotype (AA) of the FGB -455G/A SNP, the minor allele CT and TT genotypes of the GPIa 807C/T SNP, and the -5C/-5C and -5C/-5T genotypes of the GPIβα −5T/C polymorphism can be studied as genetic predictors of IS in AF patients. Logistic regression analysis was used to predict the development of IS in AF patients, depending on the presence of pathological genotypes of the FGB, GPIα, and GPIβα genes. The percentage of correct predictions for the absence of IS using this model was 99.5%. The development of IS was correctly predicted in 7.0% of cases. The overall percentage of correct predictions was 82.3%. Conclusion: The obtained logistic regression model is recommended as an additional method for assessing the risk of IS in young patients with isolated AF.

Keywords

• atrial fibrillation
• ischemic stroke
• multiple logistic regression analysis
• predictor genotypes