1 Introduction

Drug-resistant epilepsy may be defined as the failure of adequate trials of two tolerated and appropriately chosen and used AED schedules (whether as monotherapies or in combination) to achieve sustained seizure freedom^[1]. In patients with drug-resistant epilepsy, the chance of achieving seizure freedom with subsequent antiepileptic drug treatments is modest^[1]. Compared with other palliative treatments^[2-4] such as vagus nerve stimulation (VNS) and deep brain stimulation (DBS), conventional open surgery is a better treatment option^{[5, 6]}.

For patients with drug-resistant partial epilepsy, the procedure is sometimes performed according to the findings of noninvasive examinations^[7], which consist of long-term scalp video-EEG monitoring (VEEG), neuropsychological testing, high-resolution magnetic resonance imaging (MRI), and positron emission tomography (PET). Some patients also need to undergo ictal single-photon emission computerized tomography (SPECT) and magnetic source imaging of interictal paroxysmal activity using magnetoencephalography (MEG). However, for patients with particularly complicated focal epilepsy, we frequently have to combine invasive preoperational EEG recordings with the above-mentioned noninvasive examinations, to delineate the optimal cortical resection^[8], as data obtained from noninvasive investigations do not suffice for the identification of the epileptogenic zone (EZ). Currently, invasive SEEG exploration is recommended for EZ investigations because it produces fewer complications and has a superior ability to explore deep regions in comparison to subdural grid electrodes^[9-12].

SEEG was first developed by Talairach and Bancaud in Paris in the late 1950s and its safety and efficacy have been proven over the last 60 years. Its implementation consists of stereotactic implantation of depth electrodes in the brain to identify the exact location of the EZ as well as the pathways of seizure propagation^[13]. Usually, the video-SEEG recording of spontaneous seizures leads to "a tailored resection" of the EZ. In 2004, Guénot et al.^[14] reported SEEG-guided radiofrequency thermocoagulation (RF-TC), which is gradually becoming a promising stereotactic procedure.

Stereotactic surgery was first introduced in the early 20th century^[15]. With the development and application of this technique in neurosurgery^[16], it was soon utilized to resect lesions by means of radiofrequency (RF). The first paper on stereotactic lesion on epilepsy was published in 1965^[17]. Since then, stereotactic lesion as a surgical treatment for focal epilepsy gained widespread use^[18], mainly as an alternative to conventional surgery in patients with mesial temporal lobe epilepsy (MTLE)^[19-21], especially with hippocampal sclerosis^[22-24] and malformation of cortical development (MCD)^[25-27]. The mesial temporal lobe structures (amygdala and hippocampus) have been targeted; however, due to the poor results in MTLE compared to conventional open surgery, the technique was almost entirely abandoned^[21].

With the recent development of modern imaging techniques and the increasing accuracy of stereotactic targeting, it has gradually obtained new popularity^[24] SEEG exploration is a powerful tool for defining the spatial and chronological evolution of an ictal discharge^[28]. Apart from the passive recording of brain electrical activity, it also enables accurate functional mapping of cortical and subcortical eloquent structures by using intracerebral electrical stimulations. Electrical stimulations may also elicit habitual ictal manifestations and provide a better definition of the organization of the EZ^[29]. Because of these advantages, its use has expanded very significantly over the last 5 years. Some patients, however, are still not eligible for surgery after SEEG because the EZ is multiple or located close to, or even inside, eloquent cortical areas.

Owing to the work of French researchers, nowadays, SEEG-guided RF-TC has become a widely used treatment in patients who are not eligible for resection. At present, the mechanism of the procedure is still unclear. RF-TC lesioning^[30] is based on the spreading of an RF current between the two poles of a dipole. This produces an oscillation in each point of the electric field (E-field) between these two poles, which induces the nearby charged ions in the electrolyte to move back and forth in space, at the same high frequency, thereby creating an ionic oscillation current named J. RF-TC is the result of the frictional heating within the tissue resulting from the J-field. The power deposition and the elevation of temperature are directly linked to the J-field. High temperature can degenerate proteins, and, consequently, ablate the lesion or blocked electrical propagation. In addition, coagulation might interfere with complex epileptogenic networks at crucial nodes and, thus, block initiation of the ictal discharge and/or its spread through the symptomatogenic area^[28].

This technique is characterized by the following advantages^[17]: (1) the RF-TC can be targeted on the seizure-onset zone delineated by SEEG recordings; (2) multiple lesions can be performed between adjacent contacts of the SEEG electrodes; (3) the electrodes used to perform the RF-TC are those implanted for the SEEG so that carrying out thermo-SEEG does not increase surgical risk, such as bleeding; (4) RF-TC can be preceded by a functional mapping through cortical stimulation, thereby preventing neurologic deficits; (5) these lesions are well tolerated by the patients and the procedure does not require anesthesia; (6) the clinical and electrophysiological condition of the patient can be monitored in real time before, during, and after the lesion is resected. Such monitoring allows for the interruption of the procedure as soon as a sensation of heat is reported^[31]; (7) RF-TC does not preclude subsequent conventional surgery in case of failure; (8) there are no additional costs for either consumables or prolonged hospitalization.

It has been recently reported that SEEG-guided RF-TC was a safe and efficient technique in many cases gathered from a large population over a 10-year period^[17]. In contrast to other palliative^{[2, 3]} treatments such as vagus nerve stimulation (30%-63% reduction of >50% of seizures)^[4-6], multiple subpial transection^[32], callosotomy^[33], or deep brain stimulation of the anterior nucleus of the thalamus (54% reduction of >50% of seizures)^[6], SEEG-guided RF-TC is a focal treatment targeted selectively at the epileptogenic zone (as is radiosurgery)^{[34, 35]}. Furthermore, SEEG-guided RF-TC efficacy is a predictor of outcome in patients who undergo conventional open surgery^[17].

2 Indications

All noninvasive presurgical investigations are not sufficiently congruent for localizing the epileptogenic zone; hence, SEEG is performed before surgery^[31]. For each patient, a customized arrangement of intracerebral electrodes is planned with the aim of verifying one or more working hypotheses based on the results of previous investigations. In the end of intracranial recording, it appears that patients are not eligible for surgery because of a larger size or a multiple epileptogenic focal zone. Some patients refuse to undergo resection because of greater risks. Furthermore, surgery is entailing a high risk because the epileptogenic zone is located inside eloquent cortical areas (primary language, motor, or visual zones) or is inaccessible to surgery. Patients may benefit from SEEG-guided RF-TC, and obtain a substantial improvement in seizure frequency. Of course, all patients should be fully informed, and they should give their informed consent.

In conclusion, the favorable risk-benefit ratio of SEEG-guided RF-TC in comparison to conventional surgery and other palliative treatments suggests expanding the indications of SEEG-guided RF-TC, either to patients for whom surgery is risky or even not viable, or as a predictive test before resective surgery^[17].

3 Equipment and parameters of RF-TC

In Italy, Cossu et al.^[28] adopted the following parameters: current power progressively increased from 1.5 W to 8.32 W within 60 seconds and current intensity (usually approximately 25 mA) variable according to impedance^{[7, 14, 28]}. They used a radiofrequency lesion generator equipment (NeuroN50 or NeuroN100 Stryker Leibinger, Freiburg, Germany with Dixi Medical and Alcis electrodes, respectively).

In France, Bourdillon et al.^[17] adopted the following parameters: 50 V/120 mA (6 W) for 10-20 s, while the lesion generator system was from a model RFG-3, manufactured by Radionics (Radionics Medical Products, Burlington, MA, U.S.A.), and electrodes were manufactured by Dixi (Dixi Medical, Besancon, France).

4 Selecting the RF-TC sites

Each RF-TC lesion lies between a pair of adjacent electrode contacts. The contacts selected for RF-TC are according to one or more of the following standards^[28]: (1) their involvement in the onset of the ictal discharge, including either spike wave discharges or low amplitude fast activity (LAFV); however, interictal paroxysmal activities are not considered for planning RF-TC sites^[7]; (2) an intralesional location; (3) induction of habitual ictal clinical manifestations by their electrical stimulations. Each selected site is first tested by low-and high-frequency bipolar electric stimulations between two adjacent contacts with increasing current up to a maximum of 3 mA.

If the selected contacts are positioned inside the functional area, as documented by SEEG functional mapping (such as movement, speech, and vision), they should be excluded from treatment because the resection of functionally critical areas will produce unacceptable neurological deficits. Moreover, coagulation should be performed far from vascular structures (>2 mm from selected contacts).

Catenoix et al.^[36] identified two factors, focal lowvoltage, high-frequency activity at seizure onset and lowered epileptogenic threshold in the coagulated area, that could be predictive of a favorable seizures outcome after RF-TC.

The anatomic localization and extent of the RF-TC are assessed in every patient by a brain MRI scan performed three months after the RF-TC procedure^[7]. Based on the MRI findings, the diameter of the lesions ranges from 5 to 7 mm and the size^[36] is about 6 mm×8 mm around the electrode contact. The volume of the lesion around the target is about 100 mm^3[28].

5 Efficiency

Cossu et al.^[28] reported that permanent seizure freedom occurred after RF-TC in 16 patients (18.0%). A permanent worthwhile improvement (decrease in seizure frequency was≥50%) was reported by 9 additional patients (10.1%). A total of 25 patients (28.1%) presented with persistently significant improvement in their seizures frequency or severity. All patients were followed-up over 12 months. Excellent results from studies in Italy have been reported for patients with nodular heterotopy and MTLE, particularly in patients with hippocampal sclerosis^[22-24]. Conversely, RF-TC results in patients eligible for resection proved inferior to those of surgery. These results showed that this procedure was not an alternative to resection surgery and urged the researchers to stop recommending RF-TC to patients eligible for conventional surgery, and especially to patients eligible for mesial temporal resection, which has a better surgical prognosis^{[7, 37]}.

A French article^[17] reported that 25% of patients were seizure-free at two months and 7% at one year. 67% of patients were classified as responders (decrease in seizure frequency was≥50%) at two months and 48% at one year; 58% of responders maintained their status during the long-term follow-up. The seizure outcome was significantly better when the SEEGguided RF-TC involved the occipital region. When the surgery followed an SEEG-guided RF-TC, the positive predictive value of being a responder two months after an SEEG-guided RF-TC, and to be Engel's class Ⅰ or Ⅱ after surgery was 93%. Therefore, the researchers routinely considered seizure control at two months after SEEG-guided RF-TC as a therapeutic attempt prior to any surgical decision. The only significant predictive factor to affect efficiency was the occipital location of the target, which predicted the probability of the patient to be responder at 12 months.

Catenoix et al.^[7] thought that due to the limited spatial sampling of SEEG recording sites, a complete destruction of the epileptogenic zone could not be achieved by RF-TC, so that RF-TC efficiency could not be linked with the completeness of RF-TC. However, increasing the RF-TC sites' density in an epileptogenic zone is likely to increase the efficacy of the procedure. Dysplasia might then have the better chances of ameliorating epilepsy by this way.

The following factors^[28] affecting efficiency should be included: sex, age at seizure onset, age at RF-TC, duration of epilepsy, seizure frequency, epilepsy related antecedents, preoperative MRI results (lesional or nonlesional), extension of lesion in magnetic resonance (MR) images, etiology of epilepsy (assessed by MRI or, for those who underwent surgery, histology), induction of seizures by low-or high-frequency intracerebral electrical stimulations, contribution of SEEG to EZ localization (localizing or nonlocalizing), site of the EZ (temporal or extratemporal), site and volume of RF-TC, total number of RF-TC sites, and number of RF-TC sites within the MRI-detected abnormality, time of RF delivery, speed of power increase, etc.

MRI data and epileptic duration, known to be important factors affecting the results of epileptic procedure, as well as the number of coagulations during SEEG-guided RF-TC, directly related to the lesion volume, have been considered as potential factors affecting effectiveness^[17]. There was no statistically significant difference in efficiency among these three parameters at 12 months, considering the last observation carried forward analysis. Its raw data suggests a higher efficiency at 12 months for dysembryoplastic neuroepithelial tumors (DNET)/ganglioglioma (n=6, 83% seizure-free and 100% responders at 12 months)^[17]. Other factors^[28] significantly correlated to seizure freedom included the patient's age (P=0.02885) and number of RF-TC sites (P=0.0271).

Another preliminary laboratory study found that slowing the increase of current power, usually resulted in larger coagulation volumes^[28].

6 Safety

Bourdillon et al.^[17] reported six (3.7%) of 162 patients experienced adverse events, which were divided according to the duration (transient or permanent) and expectancy (expected or not) of the symptoms. None of the patients showed any increase in seizure frequency. Due to the different locations of the effecting RF-TC, the six patients experienced different kinds of deficits, including hand palsy, ankle palsy, hemiparesis, gritty sensation in the mouth, brachiofacial palsy, and partial aphasia. The hand and ankle palsy were permanent, while the others were transient. Some deficits were expected, while others were unexpected. They thought that edema of a volume larger than that of the target had led to these expected deficits.

Cossu et al.^[28] reported that two (2.2%) of 89 patients experienced side effects. One patient developed an expected permanent motor deficit, the other had unexpected severe permanent neurological morbidity.

Concerning these etiologies of these unexpected post-RF-TC neurological deficits, different explanations could be offered^[17]: (1) Edema of a volume larger than that of the target may explain an unexpected transient deficit; (2) stimulation regions where the induced clinical impairment is difficult to test, has a poor predictive value regarding neurological deficits. There has been no report of vascular damage resulting from RF-TC thus far. Edema may result from vascular damage.

In general, the RF-TC procedures are well tolerated, and painful sensation is reported occasionally when the coagulation is performed closely to the dura. Furthermore, habitual seizures during RF-TC are not a major concern^[28]. Heat sensation can be due to the proximity of the lesion site to the pericerebral cistern and can be a warning sign against unintentional injury to the optic tract or brainstem^[31].

7 Resective surgery and medication

In case the seizure recurs, repeated SEEG-guided RF-TC might be a viable option^[17]. If patients do not benefit from RF-TC (no benefit at all or transient benefit), they can undergo surgery after a variable period of follow-up^[28]. Resection is performed with the aim of removing the EZ as defined by SEEG evaluation.

Could medical treatment be tapered and stopped in seizure free patients according to the same standard adopted for those who undergo microsurgical resection? Although the antiepileptic drug regimen was modified after the first 6 months of follow-up in some patients, these changes did not alter the seizure outcome^[7]. Modification of antiepileptic drug treatment in responders was usually not done in the year following SEEG-guided RF-TC^[17]. Thus, the patients with Engel's class Ⅰ at one year could be considered as patients to whom the epileptic network has been sufficiently damaged to become drug sensitive. Late recurrence of seizures occurred in two of the six seizure free patients; hence, caution in the tapering of antiepileptic drugs in seizure-free patients recommended after surgery^{[36, 38, 39]}, also applies for RF-TC.

8 Disadvantages and limitations

Naturally, this technique has its own disadvantages and limitations, too. Compared to other coagulation techniques, such as MRI-guided laser thermal ablation (LTA), SEEG-guided RF-TC lacks the real-time control of the progression of the lesion. Indeed, some experiments^[28] in vivo and in vitro have demonstrated that the present parameters are appropriate to RF-TC.

The main limitation^[31] of this procedure is the absence of real time monitoring of the local temperature. This is the reason that great attention must be paid to the moment when an abrupt decrease in the intensity of the current occurs (usually from 110-120 to 80-100 mA), indicating that the cerebral tissue surrounding the contacts is coagulated. All thermolesions could be obtained within 50 s.

In addition, another disadvantage is that the coagulated volume is not large enough to treat large-sized epileptogenic lesions.

9 Prospects

In the future, it may be possible to develop instruments for use with RF-TC compatible with thermal MRI sequences used for real-time control of coagulation, similarly to LTA. Furthermore, future implementation of the technique might be achieved by evaluating how different procedural parameters (including time of RF delivery, speed of power increase, number of coagulation sites, and differences in coagulated substrates) influence the anatomical results of RF-TC, as revealed by either MRI or histological examination of surgical specimens. This process enables accurate planning of the thermal lesion, which can be tailored to specific geometrical requirements with substantial safe margins^[28].

A prospective study could attempt to determine whether improved RF-TC efficacy could be achieved by increasing the number of intralesional-implanted electrodes, and, thus, the volume of RF-TC, in patients with non-operable malformations of cortical development^[7].

10 Conclusions

This technique is a relatively safe treatment for patients with complex drug-resistant focal epilepsy that requires invasive EEG evaluation. Compared to conventional surgery and other palliative treatments, the encouraging risk-benefit ratio of SEEG guided RF-TC suggests it has a larger indication. It can be used as a first line of treatment, even if a conventional surgery is required in the process. However, a randomized controlled trial is still needed to determine which patients are likely to benefit from SEEG-guided RF-TC^[31].

Conflict of interests

All contributing authors have no conflict of interests.