Risk factors and protective factors for cognitive outcomes after cerebral stroke: the results of statistical modeling using clinical data and neuroimaging

Abstract

Objective. To predict the dynamics of cognitive impairment (CI) in patients with ischemic stroke based on clinical and neuroimaging data using digital morphometry of “strategic zones” of the brain and a comprehensive neuropsychological study. Design and methods. Sixty patients in the early recovery period of ischemic stroke were examined including the following methods: morphometry (in mm) of hippocampus in the mediobasal parts of temporal lobes on the coronal section and the thalamus, an interview with a clinical psychologist, the MMSE mental status assessment scale, tests for assessing the frontal dysfunction FAB and MoC A. To consider information from the psychologist’s conclusion, text mining methods were used, the TF-IDF measure was calculated, which makes it possible to identify the main topic of messages and carry out their clustering (Ward’s method identified 3 clusters). For the analysis of CI in patients, logistic regression was used, where binarized values of the MMSE and MоCA scales were considered as target variables. Results. Based on the results of modeling with target variables, respectively, where the test results on the MMSE and MoCA scales are more or less than 24 points, we found that the results of the MoCA scale or the MMSE scale assessed in the first 6 months after stroke did not predict the risk of CI after stroke. The gender did not play any role for CI development after stroke in our study. Age < 65 years decreased the possibility of CI development after stroke by an average of 0,6–1,4 % (HR = 1,006 — for MoCA and HR = 1,014 — for MMSE assessment). The results of hippocampal morphometry according to neuroimaging data showed that the height of the left hippocampus greater than 6,8 mm increases the likelihood of the absence of CI after stroke by 1,11–1,24 times (HR = 1,11 (MoCA) and HR = 1,24 (MMSE)). Being assigned to the first or second clusters by a psychologist based on neuropsychological testing reduced the risk of developing CI by 2,62–6,19 times (HR = 6,19 (MoCA) and HR = 2,62 (MMSE)) and 3,36–9,02 times (HR = 9,02 (MoCA) and HR = 3,36 (MSSE)), respectively. Conclusions. Some indicators of brain morphometry seem to be informative and helpful regarding the diagnosis and further management of patients with post-stroke CI in the early recovery period of ischemic stroke.