Surgical approaches to refractory central lobule epilepsy: a systematic review on the role of resection, ablation, and stimulation in the contemporary era

OBJECTIVE Epilepsy originating from the central lobule (i


FIG. 1.
Anatomy and functional connectivity of the central lobule.The central lobule consists of the precentral gyrus (PreCG) and the postcentral gyrus (PostCG) on the lateral surface, and their continuation in the medial brain surface, the paracentral lobule.The precentral and postcentral gyri are continuous at the bottom of the central sulcus (CS) and occasionally at the superficial level.Motor representation can also be found in the postcentral gyrus.The primary motor cortex has been further subdivided into an anterior and posterior part, which serve distinct roles.The most important efferent pathway of the central nervous system in humans, the corticospinal tract (CST), has been shown to originate from the precentral gyrus, the postcentral gyrus, the premotor cortex, and the supplementary motor area (SMA).The functional connectivity of the precentral and postcentral gyri is also shown.Modified from an image found in Wikipedia (https://en.wikipedia.org/wiki/Cerebrum;CC BY-SA 4.0 [https://creativecommons.org/licenses/by-sa/4.0]).That image was modified itself from Gray, Henry.Anatomy of the Human Body.Philadelphia: Lea & Febiger, 1918; public domain.Figure is available in color online only.
been elicited during electrical stimulation of the postcentral gyrus in > 80% of children with focal epilepsy (up to 25% of hand stimulations in the landmark Penfield paper 1 ). 6In addition, contemporary diffusion tensor imaging experiments have confirmed the contribution of multiple cortical areas to the corticospinal tract formation (i.e., M1 [37%], S1 [32%], supplementary motor area [25%], and premotor cortex [7%]). 7hile the true incidence remains largely unknown, older community-based studies have estimated that central lobule epilepsy may account for nearly one-third of all focal epilepsies. 8A significant portion of these cases are refractory to antiseizure medications, thereby necessitating neurosurgical intervention.Among the first to describe central lobe epilepsy resections, Penfield and his team described motor cortex resections in pediatric patients with cerebral palsy and drug-resistant epilepsy, without postoperative worsening of their motor deficits. 9Later, Pilcher et al. presented their series of 41 patients who underwent extensive corticectomies for lesional and nonlesional central lobe epilepsy. 10After more than 5 years of follow-up (n = 36), 8 patients (22%) were free of seizures.Furthermore, 15 patients (41.7%) did not develop a postoperative neurological deficit, while 17 patients (47.2%) experienced mild weakness, and 4 patients (11.1%) had moderate to severe residual weakness. 10hile open resection is a well-established treatment for drug-resistant rolandic epilepsy, it can be associated with considerable postoperative morbidity.Occurrence of a deficit depends on surgical technique, intraoperative mapping, neural plasticity, and associated cortical reorganization.2][13] With the advent of less-invasive modalities, there is an emerging interest in deciphering the comparative effectiveness of available treatment options to optimally inform patients, providers, and researchers.Here, we present a systematic review of available evidence on the safety and efficacy of resective, ablative, and stimulation treatment approaches for refractory central lobule epilepsy and provide recommendations for practice and research.

Search Strategy
A librarian with a master's degree conducted an electronic literature search using Medline/PubMed, EMBASE, and Scopus for all years available until June 16, 2021.The electronic search strategy used was applied to the title, abstract, or keywords: ("sensorimotor cortex" OR "Rolandic" OR "Primary Motor" OR "Primary Sensory") AND ("Epilepsy" or "Seizures").The compiled reference lists were reviewed for potential relevance.The bibliographies of included studies were also searched for missed articles.Article selection was performed by two authors (P.K. and B.N.L.), and conflicts were resolved by consensus.This study complied with the guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).

Selection Criteria
Studies were eligible for this systematic review if they fulfilled the following eligibility criteria.1) Seizures originated from the primary sensorimotor cortex.2) The abstract was written in the English language.3) The full-text article was available.4) The surgical technique and seizure freedom outcome were reported.Case reports were considered eligible, whereas abstracts, commentaries, and editorials were excluded.

Outcomes of Interest
The primary outcome of interest was the seizure freedom rate at last follow-up.Seizure freedom was defined as Engel class I outcome: completely seizure free since surgery (IA) or auras only since surgery (IB) or some seizures after surgery, but seizure free for at least 2 years (IC) or atypical generalized convulsion with antiepileptic drug withdrawal only (ID).When available, we also evaluated class Engel II (almost seizure free), Engel class III (worthwhile improvement), and Engel class IV (no worthwhile improvement) outcomes, as well as reduction in seizure frequency per month (when seizure freedom was not achieved).The secondary outcome consisted of permanent or temporary postoperative neurological deficit (motor, sensory, or language).

Data Extraction and Processing
The following data were extracted: first author and year of study, country, number of patients, patient demographics (age [pediatric patients defined as age < 18 years old] and sex), duration of epilepsy before surgery, duration of postoperative follow-up, imaging modalities used for seizure workup (MRI, PET-FDG, SPECT, and magnetoencephalography), invasive extraoperative EEG recording (subdural strips/grids/depth electrodes or stereoelectroencephalography (SEEG) or both) to record the patient habitual seizures and mapping, intraoperative electrocorticography (ECoG), type of procedure, and pathology.Data were extracted directly from the text, tables, or figures.When information was not directly stated, variables were inferred from careful examination of the article.

Statistical Analysis
Continuous and categorical variables were pooled and are presented as means and frequencies with proportions, respectively.Given the limited amount of information and the lack of patient-level data in most studies, univariate analysis for predictors of seizure freedom was not feasible.Subgroup analysis on open resection for pediatric and nonlesional cases was performed based on a prior decision.We also examined the association of invasive intraoperative EEG recording and seizure freedom using random-effects meta-regression (only studies with at least 10 patients).Statistical analysis was performed using R Statistical Computing software version 3.1.2(https://www.R-project.org/) and OpenMeta [Analyst] open-source software. 15Level of statistical significance was established at 0.05.

Literature Search Results
Our search strategy identified a total of 218 articles.After removal of duplicates, a total of 130 papers remained.
The inclusion and exclusion criteria were then applied to titles and abstracts, yielding 50 articles.pen Resective Surgery A total of 400 patients from 24 studies were analyzed (Table 1). 16-19, 23-27, 29, 33, 35, 36, 41, 43, 44, 46, 48, 53-55, 57, 61,62The mean age at surgery was 17.3 years (n = 377 with available data), and the mean age at first seizure was 6.4 years (n = 293 with available data).A preexisting motor deficit was present in 31% (n = 273 with available data).A lesion on MRI was present in 80%, and invasive EEG recording to record seizures was carried out in 49%.Resection was performed in the precentral gyrus in 64% patients and included the postcentral gyrus in 62%.The most common pathology was focal cortical dysplasia (47%), followed by tumoral lesions (22%) and other pathologies (28%).At a mean follow-up of 48 months, seizure freedom was achieved in 62% of patients.A permanent postoperative neurological deficit was observed in 27% of patients (new or worsened motor deficit in 19% and sensory deficit in 10%).In addition, a temporary motor deficit was present in 26% of patients and temporary sensory deficits in 6%.Among those studies that provided details on the specific somatotopic area resected during surgery (13 studies, 154 patients), the most common type of deficit was hemiparesis (9%), followed by upper-extremity weakness (6.5%), facial weakness (5%), and lower-extremity weakness (3%).

Subgroup Analysis: Nonlesional Epilepsy
There were 5 studies (34 patients, 71% males) that analyzed nonlesional central lobe epilepsy. 18,23,24,33,63 Te mean age at surgery was 22.4 years, and the mean age at first seizure was 9 years.All patients underwent invasive EEG recording with subdural grids to record seizures, and an overlap between seizure focus and eloquent cortex was found in 68%.A portion of the precentral gyrus was resected in 27 patients (79.4%) and the postcentral gyrus in 24 patients (71%).At a mean follow-up of 25.9 months, seizure freedom was observed in 62%, with permanent or temporary postoperative neurological deficit rates in 29% and 44%, respectively.

Subgroup Analysis: Pediatric Cases
There were 8 studies with 185 patients (56% males) that exclusively analyzed pediatric cases of central lobule epilepsy. 16-19, 27, 36, 41,48A preexisting motor deficit was noted in 30%.MRI revealed a lesion in 90%.Intraoperative ECoG was performed in 30% and invasive EEG monitoring to record seizures in 42%.Focal cortical dysplasia was the most common pathology (59.4%).At last follow-up, seizure freedom was noted in 69%.A new or worsened permanent motor deficit was observed in 16% and a temporary motor deficit in 28%.

Multiple Subpial Transections
38,42,45,46,[48][49][50] The mean age was 20.3 years, and 48% were pediatric patients.A lesion on MRI was present in 31% (n = 32 with available data).MSTs were performed in the precentral gyrus in 86% and in the postcentral gyrus in 68% of patients (both gyri in 59%) (n = 44 with available data).At a mean follow-up of 26.3 months, seizure free-dom was observed in 16%.Engel class IV outcome was noted in 31%, while the remaining patients had Engel class II (27%) or III (27%) outcomes.A temporary neurological deficit developed in 17% and a permanent deficit in 12%.

Ablative Procedures
A summary of baseline characteristics and postoperative outcomes for ablative procedures is presented in Table 2.There were 6 patients (mean age 34.8 years, only 1 pediatric patient) from two studies who underwent LITT for central lobe epilepsy. 13,56All were lesional cases (3 cavernomas, 1 cortical dysplasia, and 2 glioses), and 50% had invasive EEG recording.Three (50%) and 4 (67%) patients had ablation of the precentral and postcentral gyrus, respectively (both gyri treated in 1 patient).At a mean follow-up of 19.5 months, seizure freedom was achieved in 4 patients (67%), while the other 2 patients had Engel class II and III outcomes.One patient had a permanent motor deficit, and 3 patients developed intracranial blood products (subarachnoid or subdural hemorrhage).
Within the RFA group, there were 5 patients (mean age 21.6 years, 2 pediatric cases) from three studies. 12,28,31Four cases (80%) were lesional and all had SEEG recording.Three patients (60%) had ablation of the precentral gyrus, and 2 patients (40%) had ablation of the postcentral gyrus.At a mean follow-up of 49.4 months, seizure freedom was achieved in 1 patient (20%), whereas Engel III and IV outcomes were observed in 2 patients each.

Electrical Stimulation Procedures
A summary of baseline characteristics and postoperative outcomes for electrical stimulation procedures is presented in Table 3.A total of 27 patients (mean age 23.6 years, 7 pediatric patients) from 6 studies (plus one institutional case, yielding a total of 28 patients) underwent RNS for central lobule epilepsy. 47,51,52,58,59 Lsional MRI was observed in 3 patients (of 8 patients with available data).The precentral and postcentral gyri were stimulated in 26 (92.8%) and 8 (29%) patients, correspondingly.At a mean follow-up of 17.7 months, seizure freedom was observed in 1 patient (3.6%).Among patients who were not seizure free, worthwhile improvement (i.e., Engel class III) was the most common outcome (70%) (n = 10 with available data on Engel outcome).Overall, the average seizure frequency reduction was noted to be 79%.In the largest series of RNS patients by Jobst et al. (n = 17), a median reduction in seizure frequency of 83% and responder rate of 65% were observed. 58Few studies provided information on neurological outcomes, and no deficits were noted among those.
Within the CSCS group, there were 15 patients (mean age 24.5 years, 4 pediatric patients) from 4 studies. 11,14,30,39 n patients (66.7%) had lesional epilepsy, and all underwent invasive EEG recording with trial stimulation prior to permanent implantation (subdural grids in 14, SEEG in 1).At a mean follow-up of 52 months, seizure freedom was observed in 2 patients (13%), 80% of patients had an Engel class II outcome, and 6.7% had an Engel class IV outcome.The mean seizure frequency reduction was 90%.There were no cases of temporary or permanent neurological deficit.
Finally, we identified 4 patients (mean age 16.3 years, 3 pediatric patients) from two studies who underwent DBS for intractable central lobe epilepsy. 22,40All cases had stimulation of the subthalamic nucleus (bilateral in 2).No patient achieved seizure freedom at a mean follow-up of 28.8 months.All patients had an Engel class III outcome, with a mean seizure frequency reduction of 73% (range 67%-75%).There were no cases with temporary or permanent neurological deficits.

Discussion
In this review, we analyzed 52 studies and 504 patients to summarize available literature on epilepsy surgery for medically refractory central lobule epilepsy (Table 4).As expected, the majority of evidence is based on open resection, yielding a total of 400 patients, of whom 62% achieved seizure freedom at a mean follow-up of 48 months.Temporary and permanent motor deficits developed in 26% and 19%, respectively.We identified 46 patients who underwent MSTs of the rolandic cortex, of whom 16% achieved seizure freedom and 30% developed a neurological deficit (permanent in 12%).Finally, the small series of patients in the ablation and stimulation groups had lower seizure freedom (Engel class III being the most common outcome) but also lower neurological deficit rates compared with the open surgery group.Based on our analysis, we devised an algorithm on how to surgically approach intractable epilepsy rising from the primary sensorimotor cortex (Fig. 3).

Resection, Ablation, or Stimulation?
The field of epilepsy surgery is one of the fastest growing subspecialties in neurosurgery.While open resection remains the procedure with the highest probability of seizure freedom for drug-resistant central lobule epilepsy, novel palliative approaches are being developed to reduce seizure frequency, protect eloquent cortex, minimize the occurrence of neurological deficit, and preserve neurocognitive function.Whether resection, ablation, or stimulation should be offered is a subject of ongoing investigation and largely depends on shared decision-making with patients and/or their families and the expertise of the care team.Within this context, Yan and Ibrahim recently proposed a framework with the following 4 main factors to be considered when discussing surgical approaches to eloquent cortex epilepsy: 64 1) goals of treatment (employment, independence, driving, etc.); 2) long-term outcomes (some procedures have higher seizure freedom rates than others); 3) reversibility/timing (e.g., explore resection after failed RNS); and 4) access to technology (e.g., RNS and LITT are only approved in the United States).
The results of our systematic review are consistent with this approach.It is critical that patients and their families understand the relative risks and likelihood of seizure outcomes.To choose the optimal surgical treatment for patients with intractable central lobule epilepsy, we identified the following 4 main questions to be answered.1) Is the case lesional or nonlesional?2) Does the epileptogenic zone overlap with eloquent cortex?3) Is palliation with significant seizure reduction acceptable, or is less than seizure freedom considered a failure?4) Is a neurological deficit acceptable by the patient and/or the family?
Depending on these questions, different approaches can be pursued.In general, if the goal is seizure freedom and a neurological deficit is acceptable, then open resection can be attempted with adjunct of intraoperative ECoG and mapping.If seizure frequency reduction (i.e., a palliative approach) is acceptable and a deficit is not acceptable, an electrical stimulation treatment can be offered.Ablative procedures have more limited evidence and can be associated with complications related to exposure (e.g., burr hole opening) and tissue destruction.The optimal modality of FDA-approved electrical brain stimulation, RNS versus DBS, for central lobule epilepsy is not known.Furthermore, consideration of off-label stimulation approaches such as CSCS can be considered, but the optimal approach remains to be determined in future studies.When a case is nonlesional, invasive EEG recording to capture the patient's habitual seizures is strongly recommended, with the majority of literature utilizing subdural grids and depth electrodes rather than SEEG.Each approach has its own advantages and disadvantages, as discussed in the next paragraph.

Which Invasive EEG Recording Modality to Choose?
Extraoperative invasive EEG recording is of particular importance in eloquent cortex epilepsy, as this method allows for more refined and precise delineation of the epileptogenic zone. 16Corroborating previous work, we did estimate a higher proportion of seizure-free patients among studies that utilized higher rates of invasive extraoperative EEG recording.While SEEG technology is safer and allows for sampling of cortical and subcortical structures connected to the rolandic cortex (e.g., premotor and supplementary motor areas, thalamus, and insula), subdural grids are superior in superficial neocortical gyri coverage and FIG. 3. Proposed algorithm for management of epilepsy originating in the primary sensorimotor (rolandic) cortex.EZ = epileptogenic zone.
higher spatial resolution of sensorimotor mapping, which can be particularly insightful in cases of origin of the corticospinal tracts from the postcentral gyrus (up to 7% of pediatric patients). 65If SEEG is preferred, the electrodes can also be used to perform focal thermocoagulation of the presumed epileptogenic zone and provide evidence for higher probability of seizure freedom if subsequent resection is attempted. 66Extraoperative mapping with SEEG has been shown to be comparable to that reported with subdural grids, but in most cases the spatial resolution is limited by SEEG sampling.

What Predicts Seizure Freedom?
The limited information provided by the included studies (only 4 studies had more than 10 patients per group) did not allow for univariate statistical analysis with regard to seizure freedom.Factors that were found in the original articles to be associated with increased probability of being free of seizures included the following: 1) younger age, specific histology (vs nonspecific gliosis), and completeness of resection; 29 2) absence of residual ECoG spikes, particularly close to the margins of the resection cavity; 46 and 3) absence of interictal nonrolandic spikes on preoperative EEG. 41Interestingly, a lesion on MRI was not found to have a significant impact in the aforementioned work. 41receding seminal work as well as our subgroup analysis (62% seizure freedom in the nonlesional group compared with 61% in the overall cohort) corroborates the excellent outcomes on MRI-negative epilepsy in carefully selected patients when invasive monitoring is performed and identifies a focal, resectable seizure onset zone. 67hen all etiologies are considered, focal lesions (e.g., cavernomas, low-grade tumors, and focal cortical dysplasia) tend to have better outcomes compared with more diffuse pathologies. 46Cortical dysplasia type I can be either occult or associated with multilobar involvement or concomitant pathology (e.g., hippocampal sclerosis) on MRI and worse seizure freedom rates. 68While most included studies did not separately report on various FCD types, future work should investigate the association between FCD subtype and selection of a resection compared with other treatments and postoperative outcome.
Whether intraoperative ECoG spikes need to be examined and associated areas should be resected is a controversial issue and a topic of ongoing investigation.In a recent review by Roessler et al., the benefit of postresection ECoG spikes on long-term seizure control showed equivocal results, despite being used to guide resection. 69Detection of pathological ripples (high-frequency oscillations) seems to be more clinically relevant and is actively being explored. 70 There a Role for Multiple Subpial Transections?
Originally described by Morrell et al. in 1989, MSTs were developed as a procedure to treat drug-resistant epilepsy originating from seizure focus in eloquent cortex by transecting horizontal fibers (at 5-mm intervals) while minimizing damage to ascending and descending fibers. 71n the authors' series, the procedure was found to be associated with a 55% seizure freedom rate and a nearly 0% permanent motor deficit rate.Despite the initial en-thusiasm, a later meta-analysis demonstrated worse outcomes. 72Pooling data from 34 studies and 212 patients, Rolston et al. estimated a seizure freedom rate of 24% at a mean follow-up of 33 months when MSTs were performed alone and 55% when performed with resection. 72Currently, many authors consider MST a palliative procedure or an adjunct to a limited resection, as illustrated in the present analysis.Patients should be informed of the potential risk of neurological decline (mostly temporary) and the high risk of seizure relapse in the long term (up to 20% risk of increased seizure frequency after 2 years). 73In this study, pooled analysis suggested a similar rate of permanent morbidity compared with resection but a significantly lower seizure freedom rate.

Is Extraoperative Mapping With Invasive Electrodes Always Necessary?
Not all patients are amenable to prolonged extraoperative monitoring and mapping.More recently, there have been published studies on the utility of transcranial magnetic stimulation for noninvasive mapping of motor and language functions in patients with epilepsy. 74Schramm et al. described the safety and effectiveness of transcranial magnetic stimulation in localizing the primary cortex in 16 pediatric epilepsy patients; no adverse events were noted and no epileptic seizures were provoked. 75The estimated precision for motor mapping for epilepsy has been found to be approximately 11 mm for hand muscles compared with invasive electrical cortical stimulation; relative to direct cortical stimulation, precision was found to be 6 mm for tumor-related mapping. 74

Limitations
Our study has limitations.First, like any systematic review, the validity of results is limited by the retrospective nature of included articles, limited sample size in most series, and the individual surgical technique.Given the considerable difference in the number of patients between the different procedure types, a significant portion of our discussion and inference was built on careful examination of the articles and the authors' thought processes during decision-making rather than direct statistical comparisons.Second, there was a notable lack of neurocognitive evaluation and outcomes.Third, we were not able to perform robust univariate or multivariable analysis due to limited information provided in the included studies and lack of patient-level data.We are hopeful that predictors of seizure freedom after central lobule epilepsy surgery will be better elucidated in subsequent work.Fourth, we were not able to directly compare resective and ablative procedures due to substantial discrepancy in sample size (n = 400 vs n = 6).While the results might be promising, ablative procedures, especially LITT, were more likely to be offered to patients with lesions that did not overlap with functional cortex.Furthermore, ablation does not allow for intraoperative mapping of eloquent brain areas and more tailored eradication of the epileptogenic zone.As such, we are in need of more data on the safety and efficacy of ablation before it can be readily offered to these patients.

Conclusions
In this review, we have highlighted the safety and efficacy profile of resection, ablation, and stimulation for patients with medication-refractory central lobule epilepsy.We are hopeful that this review will be useful to providers and patients when tailoring decision-making for this challenging patient population.

FIG. 2 .
FIG. 2. Flowchart of article selection process according to the PRISMA guidelines.+/-= with or without.

TABLE 1 . Summary of baseline characteristics and postoperative outcomes on patients undergoing open surgical approaches for rolandic epilepsy
* Values in parentheses indicate the number of patients with available data.

TABLE 3 . Summary of baseline characteristics and postoperative outcomes on patients undergoing stimulation procedures for rolandic epilepsy
58eq = frequency; NA = values could not be calculated due to either absent information or nonapplicable data points.Values represent the number of patients/total patients with available data (%) unless stated otherwise.*Values in parentheses indicate the number of patients with available data.†The median was used as a surrogate for the mean seizure frequency reduction in the study by Jobst et al.58

TABLE 4 . Summary of postoperative seizure outcomes and neurological deficit rates for each treatment approach
* Forty-five patients had available data for seizure freedom and 24 patients had available data for postoperative neurological deficit.† Four patients had available data on postoperative neurological deficit.