Biomarkers of Drug Resistance in Temporal Lobe Epilepsy in Adults

Temporal lobe epilepsy (TLE) is the most common type of focal epilepsy in adults. Experimental and clinical data indicate that neuroinflammation and neurodegeneration accompanying epileptogenesis make a significant contribution to the chronicity of epilepsy and the development of drug resistance in TLE cases. Changes in plasma and serum concentrations of proteins associated with neuroinflammation and neurodegeneration can be predictive biomarkers of the course of the disease. This study used an enzyme-linked immunosorbent assay of the following plasma proteins: brain-derived neurotrophic factor (BDNF), tumor necrosis factor alpha (TNFa), and high-mobility group protein B1 (HMGB1) in patients with mesial TLE to search for biomarkers of the disease. The objective of the study was to examine biomarkers of the neuroinflammation and neurodegeneration of plasma: BDNF, TNFa, and HMGB1. The aim of the study was to identify changes in the concentration of circulating pro-inflammatory and neurotrophic factors that are prognostically significant for the development of drug resistance and the course of TLE. A decrease in the concentration of BDNF, TNFa, and HMGB1 was registered in the group of patients with TLE compared with the control group. A significant decrease in the concentration of HMGB1 in patients with drug-resistant TLE was observed. Aberrations in plasma concentrations of BDNF, TNFa, and HMGB1 in patients with TLE compared with the controls have been confirmed by earlier studies. A decrease in the expression of the three biomarkers may be the result of neurodegenerative processes caused by the long course of the disease. The results of the study may indicate the acceptability of using HMGB1 and TNFa as prognostic biological markers to indicate the severity of the disease course and the risk of developing drug resistance.


Introduction
Temporal lobe epilepsy (TLE) is characterized by recurrent seizures with an onset involving the amygdalohippocampal complex and parahippocampal place area (PHA) [1]; it is the most common type of focal epilepsy in adults [2].
Approximately 30-50% of patients with TLE develop drug resistance [3]. According to the International League Against Epilepsy (ILAE) [4], drug-resistant epilepsy is defined "as failure of adequate trials of two tolerated and appropriately chosen and used antiepileptic drugs (AED) schedules (whether as monotherapies or in combination) to achieve sustained seizure freedom". One of the major causes of drug resistance in temporal lobe epilepsy is the presence of hippocampal sclerosis (HS) [5]. Hippocampal sclerosis is caused by: a decrease in the volume of the hippocampus; dystrophic changes in neurons and a decrease in their number (mainly in the CA1 layer), relative to an increase in the number of glial cells; and demyelination of fibers [6]. Hippocampal sclerosis is characterized by the presence of the following signs according to MRI data: an increased signal from the structure in T2 and T2 FLAIR modes; and a reduction in the size of the structure in the T1 and T1 "inversion-recovery" modes by more than 30% [7].
Factors that often lead to the development of TLE (trauma, tumor, infection, etc.) are accompanied by massive neuroinflammation and neuronal loss [8,9]. Inflammation, as already known, significantly contributes to the pathogenesis of epilepsy [10]. Post-seizure inflammation can be a physiological response to stress and brain tissue damage, helping in its reparation and in balancing the enhanced metabolic demand during increased synchronic neuronal activity. Nonetheless, if the neuroinflammatory response is protracted or too intense, it can become maladaptive, leading to cellular dysfunction that is seen in epilepsy and other neurological diseases [11]. The expression of inflammatory mediators, which are induced by seizure activity in astrocytes and microglia, can cause a cascade of subsequent inflammatory events [12][13][14]. This leads to a change in the excitability of neuronal networks, which increases the likelihood of seizures [11] and further neurodegenerative processes [15].
Experimental and clinical data demonstrate that neuroinflammation and neurodegeneration involved in epileptogenesis make a significant contribution to the development of drug resistance and the chronicity of TLE [16,17]. The mechanism of development of neuroinflammation and neurodegeneration in TLE has been studied in experimental models as well as in clinical samples of biological fluids and tissues of patients [17][18][19]. In TLE cases with hippocampal sclerosis (HS), there is a progressive loss of neurons and gliosis in the hippocampus and amygdala, which may be caused by uncontrolled processes of neuroinflammation and neurodegeneration accompanied by disruptions in the blood-brain barrier and damage to brain cells [20].
Neuroinflammation can break the permeability of the blood-brain barrier, so the peripheral inflammatory status should reflect some characteristics of the central inflammatory status. Peripheral biomarkers of neuroinflammation have been proposed, [21] although a clear comparison between central and systemic inflammation is still lacking.
A problem in modern neurology is the search for a significant association between the biomarkers of neuroinflammation and the neurodegeneration that occurs during the course of the pathological process at different stages of TLE, which would permit the prediction of drug resistance and selection of further treatment tactics [22,23].
Protein biomarkers for various neurological diseases [24,25] have been previously discussed, as have other molecular biomarkers for temporal lobe epilepsy [26] in earlier studies.
The objective of the study was to identify prognostic biomarkers for the course of TLE and the risk of developing drug resistance.

Materials and Methods
The study was carried out as part of a comprehensive study on the topic "Management of orphan diseases" under registration number AAAA-A19-119031990004-3 at the Department of Medical Genetics and Clinical Neurophysiology of the Institute of Postgraduate Education of Prof. V.F. Voino-Yasenetsky Krasnoyarsk State Medical University, Russia.
The study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the Local Ethics Committee of Prof. V.F. Voino-Yasenetsky Krasnoyarsk State Medical University (extract from protocol No. 85/2018, dated 27 September 2018).

Patients
We consecutively enrolled TLE patients with hippocampal sclerosis (HS) based on the following criteria: (1) diagnosis of epilepsy, (2) history of focal seizures consistent with TLE semiology, and (3) typical brain MRI features of HS, including increased signal intensity on fluid-attenuated inversion recovery (FLAIR) imaging with atrophy in the hippocampus.
TLE and drug-resistant TLE were diagnosed by a professional specializing in the neurology of epileptic disorders and epilepsy neuroimaging.
Patients were included in the study upon signing an informed consent form.  The main characteristics of the patients are compiled in Table 1.

Methods
Clinical interrogation involved the study of the neurological status, assessment of the frequency and severity of epileptic seizures (including using the NHS-3), analysis of ongoing antiepileptic therapy, and response to AEDs. The study of neurological status was conducted according to the NHS-3, EEG video monitoring, and MRI imaging of the brain.
The determination of the concentration of BDNF and TNF-α in blood plasma was carried out using ElisaKit kits: SEA011Hu, SEA133Hu, SEC183Hu, SEA563Hu (Cloud-Clone Corp., Katy, TX, USA) and a model 3300 StatFax biochemical analyzer.
The immunobiochemical study of HMGB1 proteins in blood plasma was carried out using an enzyme-linked immunosorbent assay using the Luminex fluorescent technique with multiplex magnetic beads (Milliplex MAP Human Cytokine/Chemokine Magnetic Bead Panel IV).

Data Analysis
The median and 25-75% (Me [LQ; UQ]) were used to describe the amount of data with abnormal accumulation. The study used percentages and 95% confidence intervals (95% CI) to describe the statistics. A non-parametric analysis of variance (the Kruskal-Wallis test) followed by a posteriori pairwise comparison of the groups with each other were carried out to compare several groups on a quantitative basis. The Mann-Whitney test was carried out to compare the two groups.
The odds ratio (OR, 95% CI) or hazard ratio (HR, 95% CI) were used to assess the risk factors associated with the development of TLE. Spearman's correlation coefficient (r) was used to assess the relationship between quantitative traits with non-normal distribution. Intergroup differences were recognized as statistically significant at p < 0.05.
An ROC analysis was used with the determination of the area under the curve (AUC) and results being statistically significant at p < 0.05 to assess the quality of the classification. A logistic regression was used as a forecast to assess the influence of a combination of factors.

Results
In the group of patients with TLE, a lower concentration of BDNF, TNFa, and HMGB1 was detected compared with the control group ( (Tables 2 and 3).  Figure 1).  There was no relationship between the HMGB1 level and the clinical characteristics of epilepsy (the course of TLE, type of structural disorders, nature of EEG changes, type of AED therapy, and duration of TLE) (p > 0.05).
A statistically significant decrease in the concentration of TNFa was registered in patients with a TLE duration of more than 10 years (10.58 [8.70; 16.30] pg/mL) and a disease duration of less than 10 years (14.30 [11.22; 26.85] pg/mL) compared with the control group (p < 0.001).
In patients with a TLE duration of less than 5 years, the BDNF concentration was higher compared with patients with a duration of epileptic seizures in the range of 10-20 years (30.4 [20.8; 30.4] ng/mL vs. 23.8 [19.8; 27.3] ng/mL), p = 0.025 by median test. However, according to the Kruskal-Wallis test followed by a pairwise comparison using the Mann-Whitney test, there were no statistically significant differences, p = 0.068 ( Figure  2B). There was no relationship between the HMGB1 level and the clinical characteristics of epilepsy (the course of TLE, type of structural disorders, nature of EEG changes, type of AED therapy, and duration of TLE) (p > 0.05).
A statistically significant decrease in the concentration of TNFa was registered in patients with a TLE duration of more than 10 years (10.58 [8.70; 16.30] pg/mL) and a disease duration of less than 10 years (14.30 [11.22; 26.85] pg/mL) compared with the control group (p < 0.001).
In patients with a TLE duration of less than 5 years, the BDNF concentration was higher compared with patients with a duration of epileptic seizures in the range of 10-20 years (30.4 [20.8; 30.4] ng/mL vs. 23.8 [19.8; 27.3] ng/mL), p = 0.025 by median test. However, according to the Kruskal-Wallis test followed by a pairwise comparison using the Mann-Whitney test, there were no statistically significant differences, p = 0.068 ( Figure 2B).
To assess the contribution of changes in the concentration of BDNF and TNFa to the pathological process, an ROC analysis was performed.
The area under the ROC curve for BDNF and TNFa was 0.975, CI 95% [81.12; 97.82%] and 0.895, CI 95% [63.56%; 98.54%], respectively (p < 0.001). The threshold value for BDNF at the cut-off point was 37.5 ng/mL, and the value for TNFa was 33.5 pg/mL; values equal to or less than this indicator were recorded in patients with a long experience of TLE.
The sensitivity for BDNF concentration was 92.16%, and the specificity was 88.24%. The sensitivity for TNFa concentration was 93.94%, and the specificity was 88.89% ( Figure 3A). To assess the contribution of changes in the concentration of BDNF and TNFa to the pathological process, an ROC analysis was performed.
The area under the ROC curve for BDNF and TNFa was 0.975, CI 95% [81.12; 97.82%] and 0.895, CI 95% [63.56%; 98.54%], respectively (p < 0.001). The threshold value for BDNF at the cut-off point was 37.5 ng/mL, and the value for TNFa was 33.5 pg/mL; values equal to or less than this indicator were recorded in patients with a long experience of TLE. The sensitivity for BDNF concentration was 92.16%, and the specificity was 88.24%. The sensitivity for TNFa concentration was 93.94%, and the specificity was 88.89% ( Figure 3A).
The area under the ROC curve, which corresponds to the relationship between the development of drug-resistant TLE and the concentration of HMGB1, was 0.664, CI 95% [45.12%; 79.60%]; the threshold value at the cut-off point was 146.5 pg/mL. Values equal to or below this indicator predict a high risk of developing a drug-resistant course of epilepsy. The resulting models were statistically significant (p < 0.001). The sensitivity for HMGB1 concentration was 63.64%, and the specificity was 60.00% ( Figure 3B). The results of the ROC analysis are compiled in Table 4.   There is no correlation between the severity of epileptic seizures on the NHS-3 scale and the levels of BDNF, TNFa, and HMGB1 (Spearman's correlation coefficient r = −0.14, −0.14, and 0.15, respectively, p > 0.05). The area under the ROC curve, which corresponds to the relationship between the development of drug-resistant TLE and the concentration of HMGB1, was 0.664, CI 95% [45.12%; 79.60%]; the threshold value at the cut-off point was 146.5 pg/mL. Values equal to or below this indicator predict a high risk of developing a drug-resistant course of epilepsy. The resulting models were statistically significant (p < 0.001). The sensitivity for HMGB1 concentration was 63.64%, and the specificity was 60.00% ( Figure 3B). The results of the ROC analysis are compiled in Table 4. There is no correlation between the severity of epileptic seizures on the NHS-3 scale and the levels of BDNF, TNFa, and HMGB1 (Spearman's correlation coefficient r = −0.14, −0.14, and 0.15, respectively, p > 0.05).

Discussion
Aberrations in the plasma markers of neuroinflammation and neurodegeneration observed in our study have also been found in a number of previous studies, which are described in the literature. Our research revealed a lower concentration of BDNF, TNFa, and HMGB1 proteins in the group of patients with TLE compared with the control group. There were no statistically significant differences in other markers associated with epileptogenesis in the group of patients with TLE compared with the control group.
A number of studies have shown a correlation between the severity of epileptic seizures and increased levels of cytokines in the blood plasma (HMGB-1, TLR4, IL-1β, IL-1R1, and TNF-α) of patients with drug-resistant epilepsy [17,27]; increased levels of cytokines have also been observed in cerebrospinal fluid (IL-1β) [16,28]. The impairment of the profile of pro-inflammatory cytokines in plasma (IL-1β, IL-6, IL-8, TNF-α) also correlates with the frequency of seizures [29]. However, the level of cytokines has not been studied early when monitoring the effectiveness of treatment in TLE cases [29,30].
It has also been reported that pro-inflammatory cytokines are upregulated in patients with TLE compared with patients with extratemporal epilepsy [31][32][33].
We found a decrease in the level of HMGB1 concentration in patients with drugresistant TLE compared with the control group (p < 0.001) and patients without drug resistance (p = 0.049). According to Yang W. et al. [29], HMGB1 is involved in the pathogenesis of TLE. The level of its expression in the hippocampal tissues is high. The studies of Walker L. E. et al. [34] and Kamaşak et al. [35] demonstrated that patients with refractory epilepsy with structural changes had a higher total level of HMGB1 than the control group and the group with treatable epilepsy. The study by Kamaşak et al. [35] found a correlation between the severity of epileptic seizures and cytokines plasma level, including HMGB1, in children with drug-resistant epilepsy. These results are in contrast to our data, which is probably due to the heterogeneous sample of patients with different forms and durations of epilepsy.
A decrease in the level of pro-inflammatory cytokines, as observed in our study, was also recorded in cohort patients with TLE who had a longer course of the disease [36]. This can be explained by the fact that the long duration of the disease can induce an inadequate systemic anti-inflammatory immune response. Furthermore, a decrease in the level of pro-inflammatory cytokines is characteristic of patients taking sodium valproate and leviteracetam [31].
A decrease in the level of the pro-inflammatory cytokines TNFa and HMGB1 may be associated with long-term chronic neuroinflammation in TLE, as well as with other pathophysiological mechanisms with a predominant involvement of T-cells; downregulation was observed in the peripheral blood of patients with mesial TLE [32] with a depletion of the immune defense mechanisms of the phagocytic link or endothelial factors [33,37].
TNFa is a pro-inflammatory cytokine [30] synthesized by neurons and glial or endothelial cells of the blood-brain barrier [36,38]; they possess neuromodulatory properties, thereby contributing to rapid changes in neuronal excitability [39]. TNFa affects seizure susceptibility in animal models [39,40]. Limbic seizures also cause an increase in TNFa expression in brain microglia and endothelial cells [41].
BDNF is a modulator of excitatory and inhibitory synaptic transmission [32]. A number of studies have demonstrated a dramatic increase in BDNF mRNA and protein expression in both animals and humans with epilepsy [37,42,43]. BDNF enhances the efficiency of excitatory synapses that connect the main neurons [43]. BDNF-mediated TrkB activation can also disrupt GABA-mediated inhibition [44]. Intrahippocampal infusion of BDNF or transgenic overexpression of BDNF or TrkB in transgenic mice resulted in an increased susceptibility to seizures or their severity [44]. Immunohistochemical evidence of increased TrkB activation, manifested as an increased pTrk immunoreactivity in the hippocampal mossy fiber pathway, has been demonstrated after seizure induction in rats and mice with various electrical stimulation models and various chemoconvulsants [45].
Decreased BDNF expression has also been observed in plasma in adult TLE patients compared with controls [42]. Presumably, the decrease in plasma BDNF levels in adult patients with TLE may be mediated by the use of AEDs. One study showed that phenobarbital, valproate, and phenytoin reduced the BDNF mRNA in the cingulate cortex, hippocampus, and thalamus [45]. A decrease in the level of BDNF may indicate the suppression of neuroplasticity and neurogenesis in TLE patients compared with healthy people.
Upregulation in cytokine expression is usually observed in the acute phase of epileptogenesis [46][47][48]. Despite the fact that in cases of chronic epilepsy, there is also an increase in the expression of cytokines [49,50], the duration is nevertheless not comparable in longitude with the disease duration in our study.
Neuronal death can directly lead to a decrease in cytokine synthesis and, consequently, the hypoexpression found in patients with TLE compared with controls. According to one of the hypotheses for the development of drug resistance in epilepsy [51], massive degeneration and remodeling of the neural network caused by seizures suppresses the endogenous anticonvulsant system and prevents the AED access to neuronal targets. Therefore, a decrease in the level of cytokines, caused by massive neuronal death and remodeling, can be a marker for the development of drug resistance.
A limitation of our study is that we were conditioned by heterogeneous plasma sample of patients with different forms and durations of epilepsy. Furthermore, the BBB disruption has not been calculated.

Conclusions
The concentration of the three markers of neuroinflammation in the blood plasma of patients with TLE was reduced compared with the control group.
Thus, HMGB1 can be used as a biomarker of drug resistance in patients with TLE. In our study, a decrease in the concentration of BDNF and TNFa reflected the duration of TLE; particularly, the concentration of TNFa decreases with an increase in an onset duration of more than 10 years. The results of the study may indicate the acceptability of the use of HMGB1 and TNFa as alternative markers of TLE, and the use of these markers may indicate the risk of developing drug-resistant TLE.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to internal regulations.