Surgery for Epilepsy Involving Rolandic and Peri-Rolandic Cortex

Purpose To evaluate the risk factors associated with motor decit following surgeries involving rolandic & peri-rolandic cortex and to introduce our surgical experiences dealing with lesions in this region. Methods We retrospectively reviewed patients who experienced drug-refractory epilepsies and received surgeries in our hospital. Medical records were carefully studied, and patients with lesions located in the rolandic & peri-rolandic cortex were screened. Those with detailed follow-up information were included. Lesion locations, resected regions, and invasive exploration techniques were studied to assess their relationship with the postoperative motor decit. Results A total of 41 patients with lesions located in the rolandic or peri-rolandic cortex were included in this study. Of all these patients, 23 (56.10%) patients suffered from a transient motor decit and 2 (4.88%) with a permanent disability after surgery. All eight patients with the anterior bank of precentral sulcus resected experienced motor decit, and six of them gradually recovered within half a year. Seven patients with the anterior half of precentral gyrus resected did not experience permanent disability. A total of 14 (34.15%) patients received invasive exploration, and one of them had a permanent disability. Conclusions The anterior bank of the central sulcus is indispensable for motor functions, and the destruction of this region would inevitably cause a motor decit. The upper part of the central sulcus could also be removed without signicant neurological impairment if there is an epileptogenic lesion.


Introduction
Epilepsy is one of the most common types of neurological disorder, and around 30% of them could not be controlled well despite many antileptic drugs (AEDs) applied [1,2]. For patients with drug-resistant epilepsies, surgery is the most effective method. For example, anterior temporal lobe resection is one of the most common epilepsy surgeries, with the seizure freedom rate of 50-70% for mesial temporal lobe epilepsies (MTLE) [3,4]. Other surgical techniques for epilepsy include cortical lesion resection, disconnection, and modulation surgeries [5,6].
The goal of epilepsy surgery is to eliminate seizures or limit their severity with or without medications. To achieve this goal, a complete resection or disconnection of the epileptogenic zone is needed [7][8][9]. For seizures arising from the non-eloquent cortex, to achieve a better surgical control rate, the potential epileptogenic cortex's maximum resection is encouraged [10][11][12][13][14]. However, for seizures arising from the rolandic and peri-rolandic cortex, we face the dilemma of maximum resection and maximum brain function protection [15,16].
Several studies concerning epilepsy surgery arise from rolandic and peri-rolandic cortex [17,18], and most of these studies focused on the pediatric population [19][20][21]. However, children tend to recover better than adults for surgeries involving the rolandic cortex because of brain plasticity [22]. So experience from children could not be applied to adult populations directly. Some other studies mainly focused on the prognosis of seizures. They rarely studied factors that will help preserve motor-related functions. One case series concerning precentral gyrus resection and neurological de cits [17]. However, the benign and malignant lesions are mixed within their case series. Since seizures due to benign and malignant reasons might have different surgical strategies, they should be considered separately. Therefore, we summarized our epilepsy surgical cases involving rolandic and peri-rolandic cortex with pathological proved benign lesions. Surgical considerations, together with factors that may in uence the protection of motor functions, were introduced.

Data recruitment and ethics
This study was approved by the institutional review board of Huashan Hospital. We retrospectively reviewed surgical cases with suspected epileptogenic zone located in the rolandic cortex and peri-rolandic cortex at Huashan Hospital from Dec. 2012 till Dec. 2019. To be included in this study, all patients need to be: 1) with suspected epileptogenic zone located in or immediately adjacent to precentral and/or postcentral gyrus; 2) all received resective surgery; 3) with detailed presurgical and postsurgical imaging information (MRI, CT or intraoperative images) to con rm seizure focus and the location of resected areas; 4) have detailed follow-up information. Patients with gliomas above grade II were excluded from this study due to surgical considerations.
Presurgical evaluation procedure A comprehensive presurgical evaluation was included in all patients, including 1) detailed seizure semiology analysis, 2) high-resolution MRI (including T1, T2, and Flair), 3) video-EEG analysis, 4) neuropsychological evaluation, and 5) FDG PET if possible. The operation strategies were made based on the conclusion reached in the preoperative multidisciplinary conference. Patients were referred to invasive evaluation if the available evidence is discordant.

Surgical strategies
All patients underwent epilepsy surgery by a single surgeon who specialized in functional neurosurgery. General anesthesia or awake anesthesia was applied. Direct brain stimulation (DCS) was performed for all awake surgical cases with an Ojemann stimulator. The stimulation parameters were as follows: biphasic square wave pulses (pulse width, 1ms) with the constant current were applied. Cortical mapping starts from 1mA with a step of 1 mA and up to a maximum of patients with lesions located in the motor-related region (primary motor & premotor).

Follow-up
Postoperative follow-up is one of the essential parts to assess the outcome of epilepsy surgeries. All patients were instructed to follow-up in the outpatient clinic at three months, six months, one year, and two years respectively, after the operation. Seizure outcome was assessed according to Engel's classi cation.
Postoperative MR images were usually acquired three months after surgery. AEDs were gradually tapped off if no seizure occurred for at least two years after surgery with a neurologist's instruction.

Image acquisition and processing procedures
High-resolution MRIs were acquired with a 3T MRI device. Freesurfer (https://surfer.nmr.mgh.harvard.edu/) was applied to create a surface model of the brain.
3D slicer was applied to show three-dimensional reconstructed models of the brain and related models. Resected areas were manually delineated with MRIcron (https://www.nitrc.org/projects/mricron) by two experienced neurosurgeons. The volume of interest (VOI) was then normalized to MNI space with SPM12 (http://www. l.ion.ucl.ac.uk/spm/). Each VOI was then binarized to create the resection mask and resection masks located within the right hemisphere were ipped to the left side. After that, all VOIs were added up to create the distribution map of all resected areas.

Demographics data
Forty-seven patients with drug-refractory epilepsies who received surgery in our center were included. However, 6 of them were excluded because detailed follow-up information or medical images were not available. Finally, a total of 41 cases were studied in our case series.
There are 28 (68.29%) male and 13 (31.71%) female within our case series, with a mean age of 23.88 ± 9.43 years. The mean duration of epilepsy was 12.96 ± 8.72 years. Of all these patients, 15 (36.59%) with lesions located in the left hemisphere, and 26 (63.41%) were located in the right hemisphere. Thirteen

Topography of resections
Of all 41 patients included, those resected cortexes were mainly located around the rolandic cortex or adjacent to the rolandic cortex, slightly extended to the posterior part of the frontal cortex and the anterior part of the parietal cortex ( Fig. 1).

Surgical interventions and motor functions
Of all these 41 patients, 25 (60.98%) suffered from a motor de cit after surgery, among which 23 with transient motor impairment and 2 with a permanent disability. As for those 23 patients with transit motor disability after surgery, 6 (26.09%) removed lesions located in the anterior half of the precentral gyrus, 3 (13.04%) with lesions resected in the central sulcus, 3 (13.04%) with lesions involved both precentral and postcentral gyrus, 4 (17.39%) with lesions in the parietal region, 4 (17.39%) with lesions in the premotor cortex and 3 (13.04%) with lesions in paracentral lobule or supplementary motor area (SMA). Transient motor de cit range from one week to 6 months ( Table 2). The precentral gyrus was destructed from the anterior bank to the posterior end around the hand knob region in both patients with a permanent motor de cit.
Awake craniotomy was performed in 12 of these patients. However, 10 of them suffered from transit motor impairment, none with a permanent disability with long-term follow-up. Fourteen patients were treated with invasive evaluation (subdural electrodes/SEEG) before resectioning the suspected epileptogenic zone, and the surgical decision was made according to intracranial EEG results. However, one of them (patient 24) suffered from monoplegia after surgery (Table 2).

Subjects with persistent disability
Two patients (patients 24 & 28) suffered from persistent motor disability in our case series. For patient 24 (Fig. 3), preoperative MRI indicated a lesion located in the left superior frontal gyrus and invaded the precentral gyrus's anterior bank. This patient received subdural electrodes exploration, and EEG results indicated the lesion to be the epilepsy responsible region. The surgical plan was made based on The posterior part of the superior frontal gyrus and the anterior part of the precentral gyrus (with the anterior bank of the central sulcus destructed in the hand knob region) were resected in order to prevent seizures. This patient suffered from monoplegia of the left arm immediately after surgery. Although he received active rehabilitation treatment after surgery, his left hand suffered from a permanent de cit (muscle strength grade II).
A total of 6 patients (patients 8, 13, 17, 21, 33, 36) with lesions mainly located within the central sulcus or paracentral lobule (Fig. 4). All patients received awake craniotomy. Navigation and intraoperative electrophysiological monitoring were applied during surgery. Transient monoplegia immediately after surgery was found in all of them, and ve patients gradually recovered within three months except one (patient 33) recovered in 6 months.

General Surgical Complications
One patient in our case series suffered from intracranial hemorrhage needs hematoma evacuation. Two patients suffered from intracranial infection and need antibiotics. No death occurred in this series.

Case illustration
This is a 47-year-old right-handed male patient (patient 33) who suffered from intractable seizures for four years. Most seizures consisted of left hand somatosensory symptoms, which he described as 'tightening' or 'shaking'. Seizure mainly involved the left hand and even the left limb. Sometimes may evolve into generalized tonic/clonic, especially during the night. Although multiple AEDs, such as valproate, carbamazepine, and clonazepam, were applied, the seizure is poorly controlled. He still experienced a daily seizure before admission.
A detailed preoperative evaluation was taken on admission. Structure MRI indicates lesion located within left central sulcus, slightly high signal in T2 Flair image ( Fig. 5 A, B & C). Surgery was performed under awake anesthesia. During the operation, direct stimulation of the brain was performed and electric stimulation of the lesion with muscle response of left hand. Detailed functional mapping was not available during the surgery because intraoperative seizures were induced by electric stimulation. The continuous motor evoked potential was applied to monitor motor functions. Surgical resection of the lesion was taken under great care. The patient experienced paralysis of the left upper limb after the operation. He could move his arm a little bit one week after surgery and gradually raise his left arm one month later. Three months after surgery, the muscle strength of his left hand reached grade V.

Discussion
Surgical removal of lesions located around the rolandic cortex remains a challenge for neurosurgeons because of the high risk of neurological de cit [18]. Techniques, such as awake anesthesia, could help preserve the brain's functions during surgery but not enough. In our case series, 41 patients received surgical removal of the epileptogenic cortex involving the rolandic and peri-rolandic cortex and 39 patients without a persistent motor de cit in long-term follow-up. This result is achieved based on a good understanding of the cortex's cytoarchitectonic basis and detailed peri-surgical evaluation techniques. We introduced our experiences dealing with cases around the rolandic cortex and hope our experience could help motor function protection in clinical practice.
The precentral gyrus as a whole is not all indispensable for motor function. Traditionally, the precentral gyrus was thought to be responsible for movement, and the postcentral gyrus was responsible for sensory information processing [23][24][25]. However, this is not always the case in the real setting. According to the cytoarchitectonic organization of the brain, Brodmann divided the brain into different sub-areas. Based on Brodmann's observation, BA4, characterized by giant pyramid neurons, is responsible for voluntary movement. BA6, situated anterior to the primary motor cortex, is responsible for planning complex movements [26]. Pen eld performed direct cortical stimulation to con rm functional representation regions with awake craniotomy and found motor response areas mainly situated around the central sulcus, but not purely in the precentral gyrus [23], which is similar to Brodmann's ndings. However, the boundary between BA4 and BA6 is not always consistent with anatomical landmarks. Ziles further studied the morphology and cytoarchitecture of the central sulcus with 32 human brains and found BA4 mainly located in the anterior bank of central sulcus [27]. That is, the anterior bank of the central sulcus might be the indispensable cortex for motor function. So, surgical removal of the precentral gyrus's anterior part might not lead to a persistent motor de cit.
The lower part of the precentral gyrus is also safe to remove, even located in the central sulcus's anterior bank. According to Pen eld's early observations, stimulation in the face representation area of the precentral gyrus could induce a bilateral facial motor response, but no ipsilateral motor response was observed with stimulation in the representation area of limb [23,28]. So, motor responses are not always controlled by the unilateral hemisphere. This could explain why surgical removal of the lower part of the precentral gyrus will not cause permanent disability. In addition to identifying the motor representation cortex with DCS, it is also crucial to bear in mind that we need to nd the boundary between the hand representation area and face representation area because the motor cortex below this boundary is relatively safe to remove.
An epileptogenic lesion which is located within the central sulcus is not always a contra indicator for surgery. The epileptic lesion might destroy the brain's function locally or at a considerable distance away [7]. This is most probably because of the interictal discharges and seizure spreading effects [29,30]. So the epileptogenic lesion, especially those benign lesions that occurred early during the lifespan, may not harbor any function, even if they are located in the center of the eloquent cortex. Much attention should be paid to the perioperative evaluation stage to delineate the boundary of the lesion and the eloquent cortex to achieve a better prognosis, which is the case with our case series.
Whether invasive exploration, such as subdural electrodes or SEEG, helps protect the brain's function for epilepsy surgeries involving the rolandic cortex is doubtful. In our early cases, subdural electrodes were applied to identify the potential epileptogenic zone and help us identify the eloquent cortex to avoid motor de cits. Surgical removal of the posterior part of the superior and middle frontal gyrus, together with the upper part of the precentral gyrus (the anterior bank of central sulcus was involved), was performed (supplemental gure ). Successful control of epilepsy was achieved after surgery. However, two patients (patient 6 & patient 24) experienced a persistent left hand movement de cit. This might be because the commercially available subdural strips/grid is not precise enough for functional mapping in the eloquent cortex. Currently, each contact's diameter is 2.5mm, and the distance between two adjacent subdural strips/grid contacts is 10mm, which is inexactitude for delineating the boundary of the eloquent cortex. We need a more precise method to delineate the epileptogenic zone boundary and eloquent zone, which is the core problem in epilepsy surgeries. High-density ECoG grid/strip might help this scenario, but lots more work needs to be done.
Intraoperative neuro-electrophysiological monitoring is crucial in identifying the boundaries of the eloquent cortex and preserving the intact functions of the brain [31]. From our perspective, SSEP and MEP monitoring is a must, while awake surgery with DCS is recommended but not essential for motor function protection. For seizure arising from the rolandic cortex, it is good to identify the motor representation cortex as described before. However, it is quite common to induce a seizure for DCS, especially in the motor cortex [32,33]. We need to be quite careful with DCS in the motor cortex. SSEP helps identify the central sulcus location, and continuous MEP monitoring is essential in helping protect motor functions. In our case series, all of them have SSEP and MEP monitoring. Twelve surgeries were performed under awake anesthesia, and none of them experienced long-term motor de cits.
Neuro-navigation is also recommended in epilepsy surgery involving rolandic cortex surgery but not enough. Brain shift due to loss of CSF is a big problem for neuro-navigations [34]. We can not rely on neuro-navigations to identify surgical boundaries if we could not correct brain shift related problems. 'Surface navigation' based on relative anatomical structures, such as vessels, bones, and gyrus, could supplement the current navigation system [35].
En bloc resection of the lesion is recommended. This is because the seizure is believed orientated from grey matter and corticectomy is enough for seizure control. Destructions of extra white matter in the rolandic region may lead to a persistent motor de cit. We need to nd the boundary between the grey matter and the white matter to remove the epileptic grey matter purely. Thus, we could start from the relatively safe areas where the interface between grey matter and white matter is straightforward. The resected grey matter thickness could serve as the reference for the remaining cortex to be removed. We could also follow the boundary from relative safe areas to highly eloquent areas.

Limitations
In this study, we only included epilepsy surgeries with benign lesions. Seizures due to malignant lesions, such as GBM, were not included. This is because, for benign lesions, we could only remove regions that are responsible for seizure generation. Thus, functional preservation is of vital importance with minimum resection. However, for malignant lesions, such as GBM, we need to balance between tumor recurrent and function preservation. Sometimes aggressive resection is needed, and we have to sacri ce some functions of the brain. So, our result might not be suitable for malignant tumor-induced epilepsies in the rolandic area. Secondly, a relatively short follow-up time is also a disadvantage of our study. We will still work on this topic and provide more cases with a longer follow-up time later.       Illustration of Brodmann's somatomotor and somatosensory representation areas. Brodmann areas were automatically parcellated with Freesurfer. Different areas were marked with different colors, with purple for BA1, pink for BA2, blue for BA3, green for BA4, and azure for BA6. Two-dimensional images (A, axial view. D, sagittal view, and E, coronal view) together with three-dimensional images (B, pial reconstruction. C, in ated image, and F, white matter surface) were

Abbreviations
shown. G, the enlarged section across the precentral gyrus with BA1 is shown in a 3D image.