Construction of a predictive model for assessing the risk of noncardioembolic ischemic stroke

Objective. To develop a predictive model to assess the risk of a first noncardioembolic ischemic stroke (IS). Subjects and methods. A case-control study enrolled 412 participants aged 45 to 80 years, including 206 patients who had experienced a first noncardioembolic IS and 206 healthy volunteers who had not and were matched for gender and age to those in the study group. In all the participants of the investigation, their blood was taken from the cubital vein in the morning on an empty stomach after 12-hour fasting, and was estimated using the concentrations of markers, such as glucose, insulin, total cholesterol, low- and high-density lipoprotein cholesterol, triglycerides, apolipoproteins B and A1, homocysteine, C-reactive protein, interleukins 1, 4, 6, 8, and 10, vascular endothelial growth factor A, tumor necrosis factor-а, adiponectin, uric acid, N-terminal propeptide of natriuretic hormone, creatinine, and cystatin C. All the participants underwent genotyping of 25 single nucleotide polymorphisms (SNPs): APOE (rs7412, rs429358, rs5174), APOA5 (rs34282181, rs619054), APOC4 (rs1132899), APON (rs4581), LPL (rs199675233), LPL (rs199675233), LP(a) (rs41267817), APOB (rs1042031, rs676210), APOD (rs7659), ANGPT4 (rs1044250), TNF-α (rs1800620), VEGFA (rs62401172), IL8 (rs1803205), IL6 (rs56383910), MTHFR (rs1801131, rs1801133), ADIPOQ-AS1 (rs17366743, rs185847354), ADIPOR2 (rs12342), GRM1 (rs1047005), GRM3 (rs2228595), and BDNF (rs6265). Analysis of allele recognition by a polymerase chain reaction assay using the ready-made TaqMan probes with Assey ID identification number (Applied Biosystems, USA). Results. The resulting model included the following independent variables: type 2 diabetes mellitus (DM2), adiponectin, ApoA1, IL6, ADIPOQ (rs17366743). The area under the curve (AUC) (95% confidence interval) was 0.947 (0.918; 0.976), the cut-off threshold was 0.565, while the sensitivity of the model was 87.1%, the specificity was 90.3%; the percentage of correct reclassification was 88.7%. Conclusion. The resulting predictive model includes clinical, biochemical, and molecular genetic parameters and is characterized by the high sensitivity, specificity, and accuracy of reclassification.

neurology, noncardioembolic ischemic stroke, predictive model