Прогнозирование течения мезиальной височной эпилепсии с использованием методов логистической регрессии и деревьев классификации

Abstract

Study Objective: To determine the independent role of the carriage of SNVs genes IL-1β, TNF, BDNF, NTRK-2, clinical, biochemical parameters in predicting the course of mesial temporal lobe epilepsy using statistical modeling methods. Study Design: Prospective study. Materials and Methods. 166 patients with mesial temporal lobe epilepsy were examined. The diagnosis of temporal lobe epilepsy was determined on the basis of clinical and anamnestic data, electroencephalography data, and neuroimaging data. Based on the molecular genetic analysis of the SNVs genes IL-1β, TNF, BDNF, NTRK-2, the study of the concentration of BDNF, NTRK-2, IL-1β, TNFα in blood plasma by enzyme immunoassay, neuroimaging data and clinical anamnestic data, a prognosis of the course was carried out mesial temporal lobe epilepsy using logistic regression analysis and classification trees. Study Results. The method of multiple logistic regression analysis revealed that the combined effect of hippocampal sclerosis and carriage of the A allele rs6265 of the BDNF gene leads to an uncompensated course of temporal lobe epilepsy. The accuracy was 80.3%, sensitivity — 66.7%, 95% CI: 46.0%; 83.5%, specificity — 91.4%, 95% CI: 76.9%; 98.2%. Using the CHAID algorithm, a classification tree was generated, allowing to assign patients with temporal lobe epilepsy to the group of compensated or uncompensated course of the disease. The accuracy of the model was 80.3%, sensitivity — 75.0%, 95% CI: 66.9%; 82.0%, specificity — 87.8%, 95% CI: 79.6%; 93.5%. Conclusion. Prognostically significant clinical characteristics of patients with VE, features of changes in the concentrations of circulating proinflammatory and neurotrophic factors, the carriage of single-nucleotide variants of the genes encoding them to predict the nature of the course of VE were identified which allows in practice to apply a personalized approach to the patient, especially in demand in the case of chronic disease. Keywords: mesial temporal lobe epilepsy, prognosis, logistic regression, classification trees.