Electroencephalographic criteria of the functional outcome prognosis in the acute period of spontaneous supratentorial intracerebral hemorrhage

Матеріали та методи. Здійснили проспективне, когортне, порівняльне дослідження 114 пацієнтів (середній вік – 67,4 ± 0,9 року) в гострому періоді ССВМК на тлі консервативної терапії. Рівень неврологічного дефіциту оцінювали за National Institute of Health Stroke Scale. Електроенцефалографічне (ЕЕГ) дослідження виконали в перші 2 доби захворювання. Окремо для ураженої та інтактної гемісфер визначали рівні відносної спектральної потужності ритмів дельта(0,5–4,0 Гц), тета(4–8 Гц), альфа(8–13 Гц), бета (13–35 Гц) діапазонів, тета1(4–6 Гц), тета2(6–8 Гц), альфа1(8–10 Гц), альфа2(10–13 Гц), бета1(13–25 Гц) і бета2(25–35 Гц) піддіапазонів, а також лобово-потиличні градієнти, вираженість міжпівкульової асиметрії ритмів. Функціональний результат гострого періоду ССВМК оцінювали на 21 добу захворювання за модифікованою шкалою Ренкіна (мШР), несприятливий функціональний наслідок – у разі значення >3 балів за мШР.

One of the most promising ways of effective treatment for patients with SSICH is the implementation of personalized medicine concept, which provides a differentiated approach to choosing optimal patient management tactics based on individual prognosis [5,6], which justifies the vitality of studies aimed at the elaboration of criteria for the prognosis of SSICH outcome in the acute period [7]. Electroencephalography (EEG) is a very informative method to objectify a functional state of the cerebral structures [8]. Numerous studies have proved the informative value of the cerebral structures functional state assessment in acute brain lesions, mostly of ischemic origin [9][10][11]. Several studies have been focused only on the EEG pattern in patients with SSICH, mainly concerning aspects of the vital prognosis verification [12,13]. All of the aforementioned makes it important to develop the EEG criteria for the verification of the functional prognosis of the disease in the acute period.

Aim
The aim of this study was to define electroencephalographic predictors of unfavorable functional outcome of spontaneous supratentorial intracerebral hemorrhage in the acute period.

Materials and methods
In order to achieve this goal, a prospective, cohort, comparative clinical and paraclinical study was conducted among 114 patients (67 men and 47 women, mean age 67.4 ± 0.9 years). All the patients had a hypertensive first-ever SSICH confirmed by a clinical and neuroimaging examination. They were hospitalized to the Brain Circulation Disorders Department of the Municipal Institution "City Clinical Hospital No. 6" of the Zaporizhzhia City Council within the first 24 hours from the onset of the disease and underwent conservative therapy. A required informed consent for the study participation was signed by each patient.
Electroencephalographic examination was conducted within the first 48 hours from the onset of SSICH with the help of a 21-channel encephalograph "NeuroCom" by "ХАІ-Medica" (Ukraine). The electrodes were placed in accordance with the international system "10-20". Artifact-free 30-second epoch was selected in order to conduct a spectral analysis. The following features of the bioelectrical cerebral activity were determined: 1) relative spectral rhythm power (RSRP) of delta (0.5-4.0 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-35 Hz), theta1 (4-6 Hz), theta2 (6-8 Hz), alpha1 (8-10 Hz), alpha2 (10-13 Hz), beta1 (13-25 Hz), beta2 (25-35 Hz) bands (%) separately for an affected hemisphere (AH) and intact hemisphere (IH); 2) fronto-occipital rhythm gradients (FORG) of the aforementioned bands separately for the AH and IH in accordance with the following formula: FORG (fronto-occipital rhythm gradient) = (absolute spectral rhythm power of the selected band in the frontal area -absolute spectral rhythm power of the selected band in the occipital area)/ (absolute spectral rhythm power of the selected band in the frontal area + absolute spectral rhythm power of the selected band in the occipital area); 3) interhemispheric rhythm asymmetry (IHRA) of the aforementioned bands, including those taken separately for the frontal and occipital areas in accordance with the formula: interhemispheric rhythm asymmetry (IHRA) = (absolute spectral rhythm power of the selected band in the AH -absolute spectral rhythm power of the selected band in the IH)/(absolute spectral rhythm power of the selected band in the AH + absolute spectral rhythm power of the selected band in the IH).
All the patients received conservative therapy in accordance with the Unified Protocol of Medical Help for Patients with SSICH, approved by the order of the Ministry of Health of Ukraine No. 275 dated 17.04.2014.
Statistical analysis of the results was carried out using Statistica 13.0 software (StatSoft Inc., USA, series number JPZ804I382130ARCN10-J) and MedCalc (version 18.2.1). The normality of the studied parameters distribution was checked with the Shapiro-Wilk criterion. As most of indexes were different from normal, descriptive statistics were presented in the form of a median and an interquartile range. The Mann-Whitney test was used for the assessment of intergroup differences. In order to define independent predictors, factors with a significant prognostic value within the univariate logistic regression analysis and the significance level >0.05 based on the Hosmer-Lemeshow test were gradually included to the multivariate model. Threshold values of independent predictors with the optimal ratio of sensitivity and specificity were determined based on the ROC analysis. The results with a P value <0.05 were considered as statistically significant.

Results
Hypertension was the etiological factor of SSICH in all the examined patients. The values of clinical and neuroimaging parameters in the studied cohort on the 1 st day of the disease were as follows: total NIHSS score -11 (8; 16), intracerebral hemorrhage volume -8.5 (3.4; 15.5) ml, the midline shift -1 (0; 3) mm.
UFO in the acute period was determined in 53 (46.5 %) cases. It was found that patients with UFO of SSICH in the acute period had different levels of total NIHSS score on the 1 st day of the disease: 16 (14; 18) versus 8 (6; 10) Based on the structural analysis of the EEG pattern in comparison with SSICH acute period outcome, it was determined that patients with UFO in the first 2 days from the onset of the disease had following parameters: RSRP of delta band (45.6 (33.8; 56.9) % vs. 13.8 (9.7; 23.2) % in the AH and 42.9 (28.9; 51.3) % vs. 14.   (Fig. 2).
Based on the analysis of the interhemispheric organization of EEG pattern characteristics, it was determined that patients with UFO in the acute period of SSICH diring the first 2 days of the disease onset had greater interhemispheric asymmetry in the frontal areas by means of delta band rhythms The most significant indexes of RSRP, FORG and IHRA of EEG pattern were determined based on the univariate logistic regression analysis. All of these were associated with the risk of UFO in the acute period of SSICH ( Table 1).
The use of multivariate logistic regression analysis allowed to determine that the following indexes were associated with SSICH outcome in the acute period amongst the aforementioned spectral analysis indexes of the brain bioelectrical activity: RSRP of delta band in the AH (OR (95 % CI) = 1.18 (1.11-1.25), P ˂ 0.0001) and FORG of beta band in the AH (OR (95 % CІ) = 0.0007 (0.0001-0.0293), P = 0.0001). These indexes maintained a statistically significant relation with the functional outcome in the acute period of SSICH in case of the following clinical and neuroimaging criteria inclusion to the list of potential predictors: total NIHSS score, intracerebral hemorrhage volume, severity of midline shift.
A mathematical model for the determination of the individual risk of UFO in the acute period of SSICH was developed on the basis of the brain bioelectrical activity spectral analysis results in the first 2 days from the disease onset. It is represented by the following binary logistic regression equation: β = -4.67 + 0.16 × Pred1 -7.24 × Pred2, where -6.55 -β 0 (intercept); Pred1 -RSRP of delta band in the AH; Pred2 -FORG of beta band in the AH. On the basis of the ROC-analysis, it was determined that the value β>-0.2 was the integral electroencephalographic predictor of UFO in the acute period of SSICH (sensitivity = 83.0 %; specificity = 93.4 %; AUC (95 % CI) = 0.94 (0.88-0.98), P ˂ 0.0001) (Fig. 3).
The proportion of correctly classified observations using this model constituted 87.7 % (the level of statistical significance according to the Hosmer-Lemeshow test P = 0.8377).

Discussion
The comparative analysis of the brain bioelectrical activity spectral parameters showed a bilateral predomination of a slow-wave activity (mostly delta) in the EEG pattern in patients with UFO of SSICH in the acute period in the first 2 days upon the onset of the disease. The RSRP values of delta band in this group of patients were 3.0-3.3 times higher. Lower RSRP levels of alpha band (2.8-3.1 times) and beta band (2.3-3.2 times) were reciprocal in both hemispheres. The obtained data are consistent with the results of studies on the prognostic value of spectral analysis of brain bioelectrical activity in patients with acute brain lesions of ischemic origin, which determined the relation between the spectral power shift towards slow-wave activity and the worse outcome of the disease [10,11]. The specific features of the interhemispheric organization of the EEG pattern in comparison with the functional outcome of SSICH in the acute period were also determined. Bilateral reduction in FORG of alpha band combined with the inversion of FORG of beta band in the AH and beta2 band in the IG was typical for the subcohort of patients with unfavorable functional prognosis of SSICH in the acute period in the first 2 days upon the onset of the disease.
Thus, the reduction in FORG of alpha band showed the smoothing of differences between the frontal and occipital areas due to the levels of absolute spectral power of these rhythms. The obtained data are consistent with the results of a study conducted by M. Purandare et al. (2018) [13], which demonstrated the informative value of the reduction in the fronto-occipital gradient of rhythms as an electroencephalographic predictor of intracerebral hemorrhage unfavorable outcomes. In our opinion, the detected changes in FORG of alpha band in the AH and IH have different origins. Thus, the reduction in FORG of alpha band in the AH reflects ipsilateral depression of the absolute spectral rhythm power in the occipital areas due to focal lesions of specific brain systems; whereas the reduction in FORG of alpha band in the IH is due to the elevation of the absolute spectral rhythm power in the frontal area. The heterogeneous genesis of changes in FORG of alpha band in the AH and IH was also confirmed by the fact that the reduction in FORG of alpha1 band was detected in the AH, whereas the IH showed the reduction in FORG of alpha2 band. Statistically significant intergroup differences were also identified in RSRP of alpha2 band in the frontal area. The FORG inversion of beta band in the AH and FORG of beta2 band in the AH showed a more pronounced decrease in the absolute spectral power of the abovementioned rhythms in the frontal areas of both hemispheres in comparison with the occipital areas.
The detailed analysis of intergroup differences between IHA indexes in the frontal and occipital areas of the brain allowed to specify the changes in the interhemispheric EEG pattern organization and to find the following peculiarities of IHA, associated with the high risk of UFO in the acute period of SSICH: 1) greater IHRA of delta band in the frontal areas due to the aforementioned activity prevalence in the AH in comparison with the IH as for the absolute spectral power level; 2) smooth IHRA of alpha2 band in the frontal areas due to the aforementioned activity prevalence in the AH in comparison with the IH as for the absolute spectral power level; 3) negative IHRA of theta2 band in the occipital areas due to the aforementioned activity prevalence in the AH in comparison with the IH as for the absolute spectral power level; 4) negative IHRA of alpha1 band in the occipital areas due to the aforementioned activity prevalence in the AH in comparison with the IH as for the absolute spectral power level; 5) IHRA of beta band by means of beta2 band in the occipital areas due to the aforementioned activity prevalence in the AH in comparison with the IH as for the absolute spectral power level.
The detected significant differences between the studied groups as for clinical and neuroimaging parameters allow us to consider the abovementioned changes in the EEG pattern to be neurophysiological markers of the severity in focal lesions of the cerebral structures. The study provided the diagnostic value of FORG and IHRA of different bands (in the frontal and occipital areas) in order to determine the functional prognosis of SSICH in the acute period. Namely, unfavorable prognosis was illustrated by the bilateral reduction in zonal differences in alpha rhythm range (due to alpha1 band in the AH, alpha2 band in the IH), the FORG inversion of beta range in the AH, and the reduction in zonal rhythm differences in beta2 band in the IH with the formation of multidirectional interhemispheric asymmetry found in the frontal and occipital areas of the brain.
Based on the univariate logistic regression analysis, the most informative indicators of RSRP, FORG and IHRA were determined to predict the functional outcome of the acute period in SSICH. The multivariate logistic regression analysis allowed to verify independent electroencephalographic predictors (RSRP of delta band, FORG of beta band) and to integrate them into the mathematical model enabling clinicians to identify patients with increased risk of UFO in the acute period of SSICH with an accuracy of 87.7 %. The analysis of the electroencephalographic predictor coefficients, included in the mathematical model, led to the identification of a link between these indicators and the risk of lethal outcome. Thus, the coefficient of RSRP of delta band in the AH had a positive value, while the coefficient and FORG of beta band in the AH had a negative value. The abovementioned confirms that UFO of SSICH in the acute period was associated with high RSRP indexes of delta band in the AH and low FORG indexes (including negative) of beta band in the AH.
The value of AUC of the developed mathematical model constituted 0.94 (P ˂ 0.0001) indicating a high forecast accuracy (based on the assessment with the help of the scale of M. H. Zweig, G. Campbell (1993) [14]). It also has proved the viability if its use in patients with SSICH within the complex of diagnostic procedures for the functional prognosis verification of the acute period outcome in SSICH and as the basis for making decisions for further medical treatment issues.

Conclusions
1. Bilateral shift of EEG spectral power pattern of rhythms towards the slow-wave activity with the formation of interhemispheric asymmetry of delta band rhythms in the frontal areas, bilateral reduction in zonal differences of alpha band rhythms (by means of alpha2 band, alpha1 band), the inversion of fronto-occipital rhythm gradient of beta band in the affected hemisphere and the reduction in zonal differences in beta band rhythms due to beta2 band in the intact hemisphere with the formation of multidirectional interhemispheric asymmetry in the frontal and occipital parts of the brain in the first 2 days upon the onset of SSICH are considered as criteria for the unfavorable functional outcome in the acute period of the disease.
2. Independent association with the risk of unfavorable functional outcome of SSICH in the acute period was proved for the following indexes of the spectral analysis of EEG pattern in the first 2 days from the disease onset: the relative spectral rhythm power of delta band in the affected hemisphere (OR (95 % CI) = 1.18 (1.11-1.25), P ˂ 0.0001) and fronto-occipital rhythm gradient of beta band in the affected hemisphere (OR (95 % CI) = 0.0007 (0.0001-0.0293), P = 0.0001).
3. The mathematical model for the determination of an individual risk of unfavorable functional outcome of SSICH in the acute period was developed. It takes into consideration the most informative indexes of a spectral analysis of the brain bioelectrical activity in the first 2 days upon the onset of the disease (AUC (95 % CІ) = 0.94 (0.88-0.98), P ˂ 0.0001; forecast accuracy = 87.7 %).
The perspective for the further studies is to increase the forecast accuracy for the short-term functional prognosis in patients with SSICH based on the assessment of EEG pattern reorganization in the acute period of the disease.

Funding
The study is funded as a part of scientific research work at Zaporizhzhia State Medical University "Improving diagnosis and treatment of cerebral circulation disorders at different stages of the disease", number of state registration 0118U007145 (2018-2023).