Quantitative research of epileptogenicity biomarkers and early prognosis after stereoscopic electroencephalography guided radiofrequency thermocoagulation in drug-resistant epilepsy patients

Abstract

Stereotactic electroencephalography (SEEG) is an important invasive assessment method in epilepsy surgery. After electrode implantation, SEEG-guided radiofrequency thermocoagulation (RF-TC) is performed on the discharge initiation and rapid propagation areas by monitoring intracranial electroencephalography. High-frequency oscillations (HFOs) and spikes are quantifiable epileptogenic biomarkers before and after RF-TC. This study aimed to quantitatively assess the changes in electrophysiological biomarkers – spikes and HFOs – before and after SEEG-guided RF-TC in drug-resistant epilepsy patients. We also sought to determine whether these changes, along with clinical characteristics, could serve as predictive factors for postoperative seizure outcomes. Three-minute segments of SEEG signals were analyzed in 44 patients before and after RF-TC. We used Anywave software to quantify the rate of spikes, rate of HFOs (80–512 Hz), rate of HFOs (80–250 Hz), and rate of HFOs (250–512 Hz). We analyzed the differences both at an individual level (paired t test and percentage) and at a group level (Fisher exact test). Logistic regression was used to analyze the possible influencing factors. After SEEG-guided RF-TC, 44 patients were included in the study; 25 patients showed clinical improvement, on the contrary 19 patients did not show clinical improvement. At an individual level of 44 patients, in the epileptic zone (EZ), 23 patients (52.3%) showed a significant intra-individual reduction of spikes. In the EZ, an intra-individual decrease in spikes was significantly more frequent in clinically improved patients than in not clinically improved patients (17 [68%] vs 6 [31.6%], P = .017). Duration of epileptic seizures (t = −2.052 P = .046 95% CI [−131.19–−1.10]), frequency of seizure (χ = 8.636 P = .012), performance of magnetic resonance imaging (MRI) (χ = 3.889 P = .049) and spike of EZ (χ = 5.740 P = .017) had statistically significant effects on prognosis. Both faster frequency of seizure (OR = 0.025, 95% CI [0.001–0.469], P = .014) and positive performance of MRI (OR = 29.29, 95% CI [1.656–518.065], P = .021) presented a significant effect on clinically improved patients. Only both faster frequency of seizure (area under the curve = 0.739, 95% CI [0.588–0.890, P = .007) and spike ruduced of EZ (area under the curve = 0.682, 95% CI [0.520–0.844], P = .040) was predictive of clinical improvement. There may be difference in spikes in the EZ between clinically improved patients and clinically non-improved patients. Duration of seizure, frequency of seizure, positive MRI, and decreased spike rate in EZ after RF-TC were significantly associated with clinical improvement of seizures. More frequency of seizure and decreased spikes rate in EZ after RF-TC are significant in predicting the improvement of epileptic seizures.

1. Introduction

Currently, there are over 60 million epilepsy patients worldwide, of whom one-third have drug-refractory epilepsy (DRE). Stereotactic electroencephalography (SEEG) is an important invasive assessment method in epilepsy surgery. After electrode implantation, radiofrequency thermocoagulation (RF-TC) therapy is performed on the discharge initiation and rapid propagation areas by monitoring intracranial electroencephalography.^[1–3]

Multiple studies have shown that SEEG-guided RF-TC has good clinical efficacy and safety, but the improvement rate of epilepsy symptoms in patients after SEEG-guided RF-TC reported by various research centers varies greatly.^[4–6] Possible reasons may be related to the evaluation of the epileptic area and the selection of targets. Therefore, there is an increasing amount of research on electrophysiological markers that induce epilepsy.

High-frequency oscillations (HFOs) and spikes in intracranial electrode electroencephalography have long been considered important electrophysiological features of epileptic zone (EZ) and irritative zone (IZ). Studies have shown that quantified changes in the rate of spikes and HFOs can be observed after RF-TC, and the reduction of these markers correlates with a favorable clinical outcome after RF-TC and with successful resective surgery.^[7] This may suggest that interictal biomarker modifications after RF-TC can be clinically used to predict the effectiveness of the thermocoagulation procedure and the outcome of resective surgery.

The purpose of this study was to quantitatively evaluate the changes in epileptogenicity biomarkers, specifically spikes and HFOs, in patients with DRE before and after SEEG-guided RF-TC. We aimed to analyze these changes in different brain regions, including the epileptogenic zone (EZ), IZ, and thermocoagulated zone (TZ), and further identify clinical and electrophysiological predictors associated with postoperative seizure improvement. At the same time, the related factors affecting the improvement of epilepsy were analyzed.

2. Materials and methods

2.1. Inclusion and exclusion criteria

This study was approved by the Ethics Committee of Tianjin Huanhu Hospital (Approval No. 2021-059) and was registered in the hospital research registry under protocol ID TJHH-EP-RFTC-2021. We extracted from the database of the Epileptology Unit of Huanhu Hospital of Tianjin all the patients in the period January 2018 to December 2023 according to the following inclusion criteria: having undergone SEEG-guided RF-TC for DRE patients and the availability of at least 24 hours of SEEG recording just before and after the end of an RF-TC session sampled at 1024 Hz or more; approval for the procedure by the Ethics Committee of Tianjin Huanhu Hospital, with informed consent obtained from patients and their families.

Demographic and clinical data collected for analysis included age, gender, duration of epilepsy, frequency of seizure, MRI performance, number of electrodes/side, hypothetical EZ/side, number of contacts, epilepsy etiology, before and after RF-TC, subsequent surgical intervention, and clinical outcome. Detailed information is presented in Table 1.

Table 1.

Population’s clinical features and characteristics regarding the RF-TC guided by SEEG procedures (n = 44).

Variable	n	Range	Median (%)
Age (years old)	44	4–50	20.6 ± 1.9
Sex
Male	28
Female	14
Duration of epileptic seizures (yr)	44	0.1–39	6.5 (0.8, 11.3)
Seizure frequency (D/W/M)	10/13/21
Etiology of epilepsy
MCD	4
Polymicrogyria	1
FCD	3
NDT	3
HS	12
CT	2
CVD	2
Unspecified histology	6
Normal MRI	9
Lateralization of epilepsy (Rs/Ls/Bi)	26/16/2
Lateralization of electrode implantation
Right-sided	10
Left-sided	6
Bilateral with right hemispheric predominance	16
Bilateral with left hemispheric predominance	12
SEEG monitoring duration (d)		3–15	7.5 (6.8, 11)
Number of electrode implants		4–12	7.7 ± 0.4
Follow-up time (mo)	42	9–51	32 (13.5, 43.3)
Number of electrode implants		5–12	7.7 ± 0.4
Total contacts	2562	85–163	116 ± 8
Contacts in EZ	330	8–24	15.0 ± 1.7
Contacts in the irritative zone	456	9–33	21 ± 2
Contacts thermocoagulated	330	9–24	15.1 ± 1.6
Follow-up time (mo)	42	3–51	32 (13.5, 43.3)
Clinically improved patients	25
Not clinically improved patients	19

Bi = bilateral, CVD = cerebrovascular disease, CT = cranial trauma, D = daily, EZ = epileptogenic zone, FCD = focal cortical dysplasia, HS = hippocampal sclerosis, Ls = left-sided, M = monthly, MCD = malformations of cortical development, MRI = magnetic resonance imaging, NDT = neurodevelopmental tumors, RF-TC = radiofrequency thermocoagulation, Rs = right-sided, SEEG = stereotactic electroencephalography, W = weekly.

2.2. Electrode implantation design and execution

The electrode implantation plan was based on the patient’s seizure symptoms, 24-hour video electroencephalogram (VEEG) data, neuroimaging findings (3-dimensional T1-weighted sequence, PET), and other noninvasive evaluation results, including magnetic resonance angiography and magnetic resonance venography. These data were used to hypothesize the location of the epileptic network and to define the intracranial electrode coverage area. The hypothesized preoperative epileptic foci identified for SEEG included the temporal lobe (20 cases), frontal lobe (12 cases), parietal lobe (1 case), multiple lobes (8 cases), and cingulate gyrus (3 cases). Preoperative imaging included 3D-enhanced thin-layer MRI and thin-layer CT scans (1-mm slice thickness and interval), with imaging data integrated into the neurosurgery stereotactic surgical assistance system work station (SINOVATION, Sinovation Medical Technology Co., Ltd, Beijing, China) for multimodal image fusion. A multi-channel electrode implantation plan was then designed. During the 5-year period, the department employed the Leksell stereotactic headframe (ELKTA Ltd., Stockholm, Sweden) and surgical robots (SINO ROBOTICS, Sinovation Medical Technology Co., Ltd, Beijing, China) for electrode implantation.

2.3. SEEG recording and RF-TC procedures

We use the Talairach stereotactic implantation method of multi-contact electrodes from SINO (Sinovation Medical Technology Co., Ltd, Beijing, China) in the brain. Standard electrodes have 8 or 16 contacts. Each contact is 2 mm in length and 0.8 mm in diameter. The intercontact space is 1.5 mm. The position of the electrode is determined based on the hypothesis of the EZ from the first stage evaluation. After surgery, computed tomography (CT) and/or MRI scans of head were performed to verify the presence of complications such as intracranial hemorrhage and the accuracy of electrode position. We used Huake Precision Software (SINOVATION) for CT/MRI image registration to check the anatomical position of electrode trajectory. SEEG recordings were obtained using the Nihon Kohden system (Nihon Kohden, Inc., Tokyo, Japan), capturing at least 3 habitual seizures from patients over a period of 1 to 3 weeks. The contacts used for RF-TC were selected for each patient according to the 4 criteria: contacts sampling structures that belong to the EZ, as defined by visual analysis complemented by quantitative SEEG signal analysis using the Epileptogenicity Index (EI); contacts sampling structures that belong to the early propagation zone; contacts located within or at the MRI-visible borders of the lesion suspected to be epileptogenic; and/or induction of habitual ictal clinical phenomena by electrical stimulation of those contacts. If the contact point of thermal coagulation is adjacent to the functional area or closely related to blood vessels, thermal coagulation will not be performed. The parameters of the RF-TC device (Model No. R2000B-M1, BNS, Beijing, China) were set as follows: 7.5 W power applied in 2 sessions of 30 seconds each, with a 30-second pause between the sessions.

2.4. Analysis of SEEG signal and RF-TC

SEEG signal was recorded on a Nihon Kohden system (Nihon Kohden) with a 1024 Hz sampling rate and a 16-bit resolution. The system included a hardware high-pass filter (cutoff 0.16 Hz) and a low-pass filter (cutoff 513 Hz). Brain regions exhibiting focal low-amplitude fast rhythms on low-frequency rhythms were identified as typical seizure onset sites. After capturing 3 or more habitual seizures, cortical electrical stimulation was conducted to identify functional areas. Stimulation parameters commonly used included low-frequency (0.9 Hz) and high-frequency (50 Hz) stimulation, with a pulse width of 200 to 300 μs, intensity of 1 to 12 mA, duration of 2 to 5 seconds per stimulation, and intervals of 10 to 20 seconds between stimulations. During electrical stimulation, if habitual seizures were elicited at lower current intensities (1–4 mA) and the corresponding key electrode contacts matched the seizure initiation area recorded by SEEG, these contacts were confirmed as the EZ and used as targets for RF-TC. Contacts selected for RF-TC were determined based on 4 criteria^[2,8]: sampling structures in the EZ identified through visual analysis and quantitative SEEG analysis (e.g., EI); sampling structures in the early propagation zone; contacts located at or near MRI-visible lesion borders suspected to be epileptogenic; and induction of habitual ictal clinical phenomena during stimulation of these contacts. Thermal coagulation was avoided if the target contact was near functional areas or closely associated with blood vessels. RF-TC was performed using the R2000B-M1 RF generator (BNS). The power was set at 7.5 W, applied in 2 cycles of 30 seconds each with a 30-second pause between applications. The estimated contact point temperature during coagulation reached approximately 78 to 82°C, as monitored by the internal calibration system of the RF generator. Each selected bipolar contact pair underwent 1 coagulation session unless interrupted for safety. Contacts adjacent to eloquent cortex or critical vasculature were excluded from coagulation. No significant discomfort was reported during the procedure, although some patients described hearing abnormal intracranial sounds. About 24 hours post-procedure, some patients experienced tolerable headaches, possibly linked to brain edema near the RF-TC-treated tissue. All patients were discharged without significant new neurological deficits. The timing of surgical resection following RF-TC depended on seizure frequency: patients with daily seizures were observed for 1 week, and those with weekly seizures were observed for 3 months. Surgical intervention was recommended if seizure frequency did not reduce by 50%.

2.5. Follow-up and evaluation methods

A monthly seizure frequency was recorded for each patient after thermocoagulation, and a reduction in seizure frequency was calculated relative to the baseline frequency prior to RF-TC. Patients who underwent resection surgery were followed up until the time of surgery, while those who did not were followed up until August 2024. The modified Engel classification system was used to evaluate treatment efficacy: Engel I represented seizure cessation or occasional non-disabling simple partial seizures; Engel II indicated a reduction in seizure frequency by more than 90%, with occasional complex partial seizures; Engel III corresponded to a 50 to 90% reduction in seizure frequency; and Engel IV represented a reduction in seizure frequency of <50%. Engel I and Engel II are considered clinically improved.

To avoid potential confounding factors and for comparison purposes, we decided to analyze the same time before the beginning of the RF-TC session and after the RF-TC had been performed. All patients were awake during these periods of time. At last, 3-minute-long SEEG segments were manually chosen before and after the RF-TC. Signal analysis was computed using Anywave software,^[9–12] via which SEEG traces were pre-processed with a notch filter at 50 Hz and a low-pass filter at 512 Hz, and a high-pass filter at 0.16 Hz. A specific bipolar montage was created for each patient, removing channels with too many artifacts via a visual evaluation of the traces. Spikes and HFOs (80–512 Hz) were automatically quantified in Anywave using Delphos (Detector of Electrophysiological Oscillations and Spikes).^[10–12] We calculated the rates (number of events/min) before and after RF-TC of the following markers: spikes, HFOs, then, 2 surgeons and 1 electrophysiologist jointly determined the contacts of interest. First, the IZ is defined as the channels presenting, all along SEEG recordings, interictal activity. Second, the EZ is defined by the channels involved at seizure onset by analyzing the quantitative SEEG signal using EI and visual analysis. Finally, the TZ is composed of the thermocoagulated channels.^[8]

2.6. Analysis of 3 regions about epileptogenicity biomarkers

SEEG signal analysis was computed using AnyWave software. A specific montage was created for each patient, removing the channels with too many artifacts (including, in some cases, thermocoagulated contacts). Moreover, on the remaining channels, a visual evaluation of the traces before the analysis was performed to identify and remove the contacts, sections of electrodes with artifacts. The approach applied in AnyWave using Delphos has been elaborated.

We calculated the rates (number of events/min) before and after RF-TC of the following markers: spikes, HFOs (80–512 Hz), HFOs (80–250 Hz), and HFOs (250–512 Hz). Rates were first calculated from all channels and secondarily divided into 3 regions of interest (EZ, IZ, and TZ) as identified by an epileptologist during SEEG evaluation. In our patients, these zones could be overlapped with each other.

2.7. Statistics

The rates of all electrodes in each region were calculated and statistically analyzed using SPSS version 25.0 (IBM Corporation, Armonk). Paired-sample t tests were used to compare the rate changes (spikes and HFOs) before and after RF-TC in normally distributed data, which was confirmed using the Shapiro–Wilk test for normality. For data that did not meet the assumption of normality, the Wilcoxon signed-rank test was applied as a non-parametric alternative. Categorical variables were compared using chi-square tests or Fisher exact test when expected cell counts were < 5. Logistic regression analysis was conducted to assess possible influencing factors on clinical outcome. Receiver operating characteristic curves and area under the curve (AUC) values were used to evaluate the predictive value of key variables.

3. Results

Forty-four patients who underwent SEEG-RF-TC were followed up in our center. The results are shown in Table 1. Among them, there were 31 males and 13 females. Age of onset of epilepsy 20.6 ± 1.9 years old, duration of epileptic seizures 6.5 (0.8–11.3) years; Seizure frequency: 10 cases occur daily, 13 cases occur weekly, 21 cases occur monthly; Side of electrode implantation: 10 cases on the right side, 6 cases on the left side, 16 cases on both sides (left hemisphere dominance), 12 cases on both sides (right hemisphere dominance); number of electrodes implants are 7.7 ± 0.4. There were no electrode breakage, infection, cerebrospinal fluid leakage, or 1 case of intracerebral hematoma with a volume of approximately 3 mL, which did not affect SEEG monitoring. The hematoma was absorbed upon discharge and did not cause any adverse effects on the patient. We capture habitual epileptic seizures ≥ 3 times for each patient. Four cases underwent surgical treatment within 1 week after RF-TC, 3 cases underwent surgical treatment within 3 months, 4 cases underwent surgical treatment within 6 months, and 5 cases underwent surgical treatment within 1 year. The remaining patients did not undergo surgical treatment before the follow-up time. Follow-up time: 32 (13.5, 43.3) months, including 25 cases of clinically improved patients and 19 cases of not clinically improved patients.

At an individual level of 44 patients, in the EZ, 23 patients (52.3%) showed a significant intraindividual reduction of spikes, 15 patients (34.1%) of HFOs (80–250 Hz), 14 patients (31.8%) of HFOs (250–512 Hz), and 17 patients (38.6%) of HFOs (80–512 Hz). In the IZ, 24 patients (54.5%) showed a significant intraindividual reduction of spikes, 16 patients (36.3%) of HFOs (80–250 Hz), 15 patients (34.1%) of HFOs (250–512 Hz), and 17 patients (38.6%) of HFOs (80–512 Hz). In the TZ, 20 patients (45.5%) showed a significant intraindividual reduction of spikes, 13 patients (29.5%) of HFOs (80–250 Hz), 12 patients (27.3%) of HFOs (250–512 Hz), and 15 patients (34.1%) of HFOs (80–512 Hz).

Results of statistical analysis with regard to the clinically improved patients after RF-TC are shown in Table 2. In the EZ, an intra individual decrease in spikes was significantly more frequent in clinically improved patients than in not clinical improved patients (17 [68%] vs 6 [31.6%], P = .017), whereas this statistics difference was not significant for HFOs (80–250 Hz) (9 [36%] vs 6 [31.6%], HFOs (250–512 Hz) (9 [36%] vs 5 [26.3%], HFOs (80–512 Hz) (11 [44%] vs 6 [31.6%]). Within the IZ and TZ, we found a greater proportion of clinically improved patients with an intraindividual reduction of the spike and HFOs, but this did not reach significance.

Table 2.

Statistical analysis according to the presence of Engel grading after RF-TC guided by SEEG for 44 patients who showed significant intraindividual reduction of the different markers in the 3 areas.

Analyzed zone	Spike reduction	HFOs (80–250 Hz) reduction	HFOs (250–512 Hz) reduction	HFOs (80–512 Hz) reduction
EZ
Clinically improved patients	17 (68%)	9 (36%)	9 (36%)	11 (44%)
Not clinically improved patients	6 (31.6%)	6 (31.6%)	5 (26.3%)	6 (31.6%)
Fisher exact test	P = .017*	P = .759	P = .495	P = .402
IZ
Clinically improved patients	15 (60%)	10 (40%)	9 (36%)	10 (40%)
Not clinically improved patients	9 (47.4%)	6 (31.6%)	6 (31.6%)	7 (36.8%)
Fisher exact test	P = .405	P = .565	P = .759	P = .831
TZ
Clinically improved patients	13 (52%)	8 (32%)	8 (32%)	9 (36%)
Not clinically improved patients	7 (36.8%)	5 (26.3%)	4 (21.1%)	6 (31.6%)
Fisher exact test	P = .317	P = .682	P = .419	P = .759

Values are given as n (%). Also reported are the P values resulting from Fisher exact tests performed between clinically improved patients and not clinically improved patients for each marker in each area.

HFOs = high-frequency oscillations, RF-TC = radiofrequency thermocoagulation, SEEG = stereotactic electroencephalography.

^*Statistically significant P values.

We then analyze the related factors affecting prognosis and multiple logistic regression, including age, gender, duration of epileptic seizures, frequency of seizure, the number of electrodes implants, side of electrodes implants, MRI performance, scope of EZ, the number of contacts (total, in EZ, in IZ, in TZ), spike of EZ. The results (Tables 3 and 4) showed that duration of epileptic seizures, frequency of seizure, performance of MRI, and spike of EZ had statistically significant effects on prognosis. Only Both faster frequency of seizure (OR = 0.025, 95% CI [0.001–0.469], P = .014) and positive performance of MRI (OR = 29.29, 95% CI [1.656–518.065], P = .021) presented a significant effect on clinically improved patients. Only Both faster frequency of seizure (AUC = 0.739, 95% CI [0.588–0.890, P = .007) and spike ruduced of EZ (AUC = 0.682, 95% CI [0.520–0.844], P = .040) was predictive of clinical improvement. Sensitivity and specificity of faster frequency of seizure are 0.760 and 0.684. Sensitivity and specificity of spike reduced of EZ is 0.680 and 0.684 (Table 5 and Fig. 1).

Table 3.

Statistical analysis of the related factors affecting prognosis.

Factors		Clinically improved patients	Not clinically improved patients	t & χ	P value	95% CI
Age		29.16 ± 11.07	27.95 ± 10.77	0.364	.718	−5.508–7.933
Gender	M	18 (72%)	12 (63.2%)	0.389	.533
	F	7 (28%)	7 (36.8%)
Duration of epileptic seizures (mo)		70.96 ± 79.90	137.10 ± 132.87	−2.052	.046*	−131.19–−1.10
Seizure frequency (D/W/M)	D	9 (36%)	2 (10.5%)	8.636	.012*
	W	10 (40%)	4 (21.1%)
	M	6 (24%)	13 (68.4%)
Number of electrode implants		7.44 ± 2.00	7.73 ± 2.54	−0.434	.666	−1.67–−1.08
MRI	Ab.	23 (92%)	12 (63.2%)	3.889	.049*
	Nor.	2 (8%)	7 (36.8%)
Involving brain regions of epilepsy foci	Lo.	18 (72%)	13 (68.4%)	0.066	.797
	Ex.	7 (28%)	6 (31.6%)
Side of electrode implants	Unil.	7 (28%)	9 (47.4%)	1.750	.186
	Bi.	18 (72%)	10 (52.6%)
Total contacts		87.12 ± 26.75	90.58 ± 32.98	−0.384	.703	−21.63–14.71
Contacts in EZ		20.48 ± 11.20	20.47 ± 7.61	0.002	.998	−6.03–6.04
Contacts in IZ		22.12 ± 9.93	22.79 ± 9.87	−0.222	.825	−6.75–5.42
Contacts in TZ		15.16 ± 7.18	12.89 ± 4.97	1.176	.246	−1.62–6.15
Spike reduction	Y	17 (68%)	6 (31.6%)	5.74	.017*
	N	8 (32%)	13 (68.4%)

Ab = abnormal, Bi = bilateral, D = daily, Ex = extended, EZ = epileptogenic zone, IZ = irritative zone, Lo = local, M = monthly, MRI = magnetic resonance imaging, N = not, Nor = normal, Rs = right-sided, TZ = thermocoagulated zone, Unil = unilateral, W = weekly, Y = yes.

^*Statistically significant P values.

Table 4.

Logistic regression analysis results.

Predictor	B	P value	OR	95% CI
Duration of epileptic seizures	−0.01	.068	0.99	0.98–1.001
Frequency of seizure	−3.669	.014*	0.025	0.001–0.469
MRI	3.377	.021*	29.29	1.656–518.065
Spikes reduction	−0.394	.654	0.674	0.121–3.771

CI = confidence interval, MRI = magnetic resonance imaging.

^*Statistically significant P values.

Table 5.

ROC curve calculation results.

Predictor	Sensitivity	Specificity	AUC	95% CI	P value
Duration of epileptic seizures	0.640	0.526	0.646	0.48–0.813	.1
Frequency of seizure	0.760	0.684	0.739	0.588–0.89	.007*
MRI	0.920	0.368	0.644	0.474–0.815	.105
Spikes reduction	0.680	0.684	0.682	0.52–0.844	.04*

^*Statistically significant P values.

AUC = area under the curve, CI = confidence interval, MRI = magnetic resonance imaging, ROC = receiver operating characteristic.

Figure 1.

ROC curve. ROC = receiver operating characteristic.

4. Discussion

The treatment of DRE is achieved by accurately locating the EZ through anatomical, electrical, clinical, and SEEG, and then performing RF-TC or surgical resection of the epileptic lesion to cure the epilepsy.

However, the therapeutic effects after this procedure are uncertain,^[13] and the predictive factors of success remain poorly known, especially the potential electrophysiological predictors. The reduction of HFOs and spikes after RF-TC was reported, which corresponded to good clinical outcomes after the surgery, but many other possible influencing factors were not studied. This study evaluated the effect of RF-TC on several epileptic interictal biomarkers measured from SEEG records in clinically improved patients; meanwhile, the related factors affecting the efficacy of SEEG-guided RF-TC were analyzed.

Our results showed that only in EZ alone, the rate of spike reduction in patients with clinical improvement was significantly higher than that in patients with no clinical improvement, and there was no statistically significant difference in HFO between the 2 results. In IZ and TZ, both spikes and HFO showed no significant difference between the improved and non-improved groups. This is basically the same as the results reported in individual cases. Julia Scholly^[14] reported a case of para-ventricular gray matter ectopic, after SEEG-guided RF-TC, the spikes and HFOs in gray matter ectopic, hippocampus, and temporal cortex all decreased. However, they only distinguished the spikes, did not make a detailed division of affected brain regions, and did not divide the frequency band of HFOs. It is worth mentioning that Julia Scholly performed SEEG electrode implantation again 6 months after the first SEEG-guided RF-TC because the patient’s seizures were not significantly controlled, and they found a sharp spike reduction (>80%) in parventricular gray matter heterotopic and temporal cortex, while no significant differences were observed in the unthermocoagulated posterior hippocampus. At present, our research results show that the reduction of spikes in the EZ after SEEG-guided RF-TC is significant, while the reduction of spikes and HFOs in the IZ, TZ, and HFOs in the EZ is not exact. Future studies need to focus on a larger sample size analysis of a single disease.

Among the relevant factors affecting clinical outcomes, we found that the duration of epilepsy was significantly different between the improved group and the non-improved group. Pierre Bourdillon reported that the duration of epilepsy was an important factor affecting the outcome of surgical resection of epilepsy, and they believed that DRE with a longer duration of epilepsy had a poor prognosis. In the study, 41 out of 162 patients showed complete relief of epilepsy symptoms 2 months after SEEG-guided RF-TC, with a duration of 20.8 ± 10.6 years. Our results showed that the duration of epilepsy in the improvement group was 70.96 ± 79.90 months, while the duration of epilepsy in the non-improvement group was 137.10 ± 132.87 months. It can be seen that the average duration of seizure in the improvement group was about 6 years. Therefore, this time period may be a good time for treating DRE.

The frequency of seizures was also significantly different between the improved group and the non-improved group. Patients who had weekly and daily seizures had better epilepsy improvement than those who had monthly seizures. Logistic regression analysis showed that the more frequent the attacks, the better the clinical improvement after SEEG-guided RF-TC. This may be related to the partial destruction of EZ and IZ after RF-TC, and the decrease in discharge frequency of EZ in the improved group, which is also indicated by the obvious decrease of EZ after RF-TC. Julia Scholly stated that the improvement time of seizures after RF-TC was 9 months and 11.5 months in patients with weekly and daily seizures, respectively, while the improvement time was 18.5 months in patients with monthly seizures (a total of 4 cases), which seems to be contrary to our study results and may also be related to the small sample size. Statistical analysis of related factors was not carried out in this paper, so the effect of frequency of seizure on prognosis can not be confirmed.

The performance of MRI was statistically different between the improved group and the non-improved group. We concluded that the improvement of epilepsy after RF-TC in patients with MRI positive was significantly better than that in patients with MRI negative. This may be due to the presence of visible EZ in MRI-positive patients coupled with the fact that assessment of SEEG in the second stage allows for more accurate implantation of electrodes, thus more accurate localization of EZ. However, it does not mean that MRI negative is not suitable for SEEG-guided RF-TC. In the 9 patients with negative MRI, 2 patients had clinical improvement in epilepsy. This is consistent with other research reported in the literature.^[15] Logistic regression analysis showed that the MRI was positive, and the clinical improvement after RF-TC was better. Thus, positive of MRI is a highly predictive indicator of clinical improvement in epilepsy after RF-TC treatment for DRE. It seems that the preoperative epilepsy MRI sequence is crucial, and it must reach a certain quality and level, and try to find positive MRI results to meet the needs of epilepsy surgeons. These sequences include: 3DT1W, T2 (axial and crown position), FLAIR (axial and crown position), SWI. It is necessary to make full use of the technical characteristics of MRI scanning to show the lesions better and earlier, which can better improve the prognosis of patients with DRE.

In this study, factor analysis was also carried out on whether the spikes rate decreased after RF-TC, and there was a significant statistical difference. Our intention is to judge the prognosis of the patient based on the objective quantitative analysis of the patient’s SEEG after RF-TC. The presence of continuous or subcontinuous interictal epileptiform discharges (SCIEDs) indicates increased local cortical excitability and produces spontaneous hypersynchronous discharge, which is often an important basis for locating the origin of epilepsy before and during surgery.^[16] Dimova P and Catenoix H reported that SCIEDs in SEEG before RF-TC was a positive sign of clinical improvement after RF-TC; however, there was no clear causal relationship between the reduction of spike rate and the presence of SCIEDs and clinical improvement. In another study,^[16] 31% of patients after RF-TC experienced >80% reduction in spikes rate; however, this study was not based on SEEG electrophysiological quantitative analysis and did not find a significant correlation between spikes reduction and the clinical effect of RF-TC. Dimova P reported a reduction in spikes rate in 25% of clinically improved patients compared to 33.3% of patients in the non-improved group, but this difference was not significant. In our study, we showed that in patients with clinical improvement after RF-TC, there was a significant reduction of spikes rate in EZ. There was a statistical difference between the 2 groups.^[15,17,18]

The frequency of seizures and the decrease in spikes rate were predictive factors of the clinical improvement of seizures. The sensitivity and specificity of the former were 0.760 and 0.684, respectively. The sensitivity and specificity of the latter were 0.680 and 0.684. As mentioned above, EZ and IZ were partially destroyed after RF-TC, the discharge frequency of epileptic foci decreased, and the spikes in the EZ decreased significantly after RF-TC, which significantly improved the seizure of patients.

There are some limitations in this study. First of all, SEEG-guided RF-TC has been gradually developed in larger neurosurgery centers in recent years, and our sample size is small because of the small number of cases. It is even more difficult to carry out rigorous and detailed research on single epilepsy diseases. Further research will be carried out after the sample size is increased. Secondly, we only analyzed the quantitative analysis of electrophysiological markers, and did not research the electrophysiological characteristics of the onset of seizures in this study. In subsequent studies, we will expand the influencing factors and more completely analyze the factors that affect the improvement of seizures.

5. Conclusion

Quantified changes in the rate of spikes can be observed after SEEG-guided RF-TC in EZ. There may be differences between clinically improved patients and clinically non-improved patients. Duration of seizure, frequency of seizure, positive MRI, and decreased spikes rate in EZ after RF-TC were significantly associated with clinical improvement of seizures. More frequency of seizures and decreased spikes rate in EZ after RF-TC are significant in predicting the improvement of epileptic seizures.

Further research is needed. More evidence and study are necessary to determine the exact place of SEEG-guided RF-TC in the surgical management and prognosis of DRE.

Author contributions

Conceptualization: Shaoya Yin, Jingtao Yan, Yuhao Wang, Le Wang, Weipeng Jin, Deqiu Cui, Shaoya Yin.

Data curation: Shaoya Yin, Jingtao Yan, Yuhao Wang, Le Wang, Weipeng Jin, Deqiu Cui, Shaoya Yin.

Formal analysis: Shaoya Yin, Jingtao Yan, Yuhao Wang, Le Wang, Weipeng Jin, Deqiu Cui, Shaoya Yin.

Investigation: Shaoya Yin, Jingtao Yan, Yuhao Wang, Le Wang, Weipeng Jin, Deqiu Cui, Shaoya Yin.

Methodology: Shaoya Yin, Jingtao Yan, Yuhao Wang, Le Wang, Weipeng Jin, Deqiu Cui, Shaoya Yin.

Supervision: Yuhao Wang, Le Wang, Weipeng Jin.

Validation: Yuhao Wang, Le Wang, Weipeng Jin.

Visualization: Shaoya Yin.

Writing – original draft: Shaoya Yin, Jingtao Yan.

Writing – review & editing: Shaoya Yin, Jingtao Yan.