Analysis of the diagnostic informativity of non-contrast computed tomography markers of intracerebral hemorrhage expansion in assessment of the individual risk of early neurological deterioration in patients with hemorrhagic hemispheric stroke

Abstract

The aim of the study was to evaluate the diagnostic informativity of non-contrast computed tomography (NCCT) markers of intracerebral hemorrhage (ICH) expansion in assessment of the individual risk of early neurological deterioration (END) in patients with hemorrhagic hemispheric stroke (HHS). Materials and methods. A prospective, cohort study was conducted involving 333 patients in the acute period of hypertensive spontaneous supratentorial ICH on the background of conservative therapy. The level of neurological deficit was assessed using the Full Outline of Unresponsiveness (FOUR) coma scale and the National Institute of Health Stroke Scale (NIHSS). The computed tomography of the brain was used to verify the ICH volume, the midline shift (MS), the secondary intraventricular hemorrhage volume (IVHV) and NCCT markers of intracerebral hemorrhage expansion. As a combined clinical endpoint, END was considered (decrease of the FOUR scale score ≥2 or/and increase of the NIHSS score ≥4 or/and lethal outcome within 48 hours of hospitalization). Results. Early neurological deterioration was registered in 112 patients. On the basis of a comparative analysis, it was established that the specific weight of END was significantly higher in subcohorts of patients with individual NCCT markers of intracerebral hemorrhage expansion, than it was in subcohorts of patients without corresponding NCCT signs (p ˂ 0.0001). It was established that the following NCCT markers of ICH expansion are the most sensitive predictors of END: hypodensity, swirl sign and irregular shape (sensitivity >90.0 %). The most specific NCCT markers were island sign, black hole sign, blend sign, satellite sign and heterogeneous density (specificity >87.0 %). In accordance with the multiple logistic regression analysis, hypodensity (OR (95 % CI) = 13.56 (4.54–40.49), p < 0.0001) and island sign (OR (95 % CI) = 5.94 (2.05–17.16), p = 0.0010) are independently associated with the risk of END. A highly sensitive multi-prediction logistic regression model was elaborated in order to predict END in patients with HHS which takes into account the most informative NCCT markers of ICH expansion (hypodensity, island sign) and quantitative neuroimaging indicators (MS, IVHV) (AUC ± SE (95 % CI) = 0.92 ± 0.02 (0.89–0.95), р ˂ 0.0001). Conclusions. Non-contrast computed tomography markers of ICH expansion are associated with increased risk of END in patients with HHS. Hypodensity and island sign are independent predictors of END. The integration of NCCT markers of ICH expansion with quantitative neuroimaging indicators (MS, IVHV) in the structure of the multi-prediction logistic regression model allows to assess the individual risk of END with an accuracy of >85.0 %.