Description of technique

The multiple hippocampal transection (MHT) technique consists of a series of blunt transections perpendicular to the long axis of the hippocampus. MHT was first described by Shimizu et al. in 2006 and draws from the rationale of multiple subpial transections (MSTs) . The objective of the procedure is to transect the longitudinal fibers running parallel to the long axis of the hippocampus which are thought to be responsible for the synchronization of epileptic activity while preserving the transverse circuits of the hippocampus which are critical to memory function . Indeed, intraoperative electrocorticography (ECoG) demonstrates a reduction in epileptiform spiking synchrony following hippocampal transections lending credibility to this hypothesized mechanism .

Several approaches to the hippocampus have been described in the literature. Several groups have reported approaching the hippocampus via the temporal stem with the initial description of the technique utilizing a several-centimeter corticotomy at the superior temporal gyrus (STG) . Alternatively, groups have reported utilizing a transsylvian approach, reasoning that it reduces the risk of iatrogenic neurologic deficits . A transsylvian approach, however, can be complicated by variations in vascular anatomy, and it has been suggested that a patient’s particular vascular anatomy may be used to decide between STG corticotomy and a transsylvian approach to the temporal stem . Others have reported accessing the hippocampus via lateral neocortical approaches. These approaches require violation of potentially healthy brain tissue but may be less technically challenging and may provide greater exposure of the hippocampus while sparing the temporal stem fibers . Open surgical approaches, however, may independently contribute to memory decline. Indeed, laser amygdalohippocampectomy has been associated with superior cognitive outcomes versus open resection . Further, imaging studies indicate that disruption of extra-mesial temporal lobe white matter tracts—such as uncinate fasciculus and entorhinal cortex superficial white matter tracts—correlates with poorer postoperative memory outcomes . Certainly, care must be taken when approaching the MTL to minimize the disruption of critical white matter tracts within the extra-mesial temporal lobe to mitigate the risk of postoperative memory decline.

After sharply opening the alveus, and being careful to preserve the fimbria, transection of the hippocampus is accomplished using custom-made circular and ovoid ring transectors . The original description of MHT reports the use of a 2-mm ring transector to transect the superficial pyramidal cell layer, a 4-mm ring transector to transect the endofolium, dentate gyrus, and CA1 regions, and a 4 mm by 2 mm ovoid ring transector at the posterior aspect of the hippocampus . Similar instruments have been used successfully in pediatric patients .

The number of transections required to achieve good clinical seizure control remains an issue of some debate. It was initially thought that additional transections should be made until spiking activity across the entirety of the hippocampus was eliminated . Subsequent work differentiates between “Type B” spikes and “Type H” waves which have maximum amplitudes over the body and head of the hippocampus respectively. Type B spikes are most relevant to seizure outcomes . Additional transections aimed at completely eliminating spiking activity may therefore be avoided, provided Type B spikes have been controlled. Further reduction in the number of transections may be possible by first transecting regions of the hippocampus with the greatest amplitude of spiking activity . While the literature regarding pediatric MHT is limited, there are pediatric cases reported with Engel class 1 outcomes without complete elimination of hippocampal spiking, though a reduction in amplitude was seen on intraoperative ECoG .

MHT has been reported in combination with both MST and/or resection of additional temporal lobe structures in several case series. Shimizu et al. initially performed amygdalectomy only if amygdalar spiking was seen on intraoperative ECoG . Patil and Andrews reported the use of MHT in conjunction with parahippocampal gyrus MST with or without amygdalectomy . A more aggressive technique has been investigated wherein the temporal tip, amygdala, and temporal neocortex lateral to the fusiform gyrus were resected in conjunction with MHT. This series compared seizure and neuropsychiatric outcomes prior to the addition of these resections, demonstrating no significant benefit nor significant morbidity associated with the additional resections .

Indications

MHT was initially reserved for patients with medically refractory epilepsy (MRE) with normal hippocampi on magnetic resonance imaging and preserved preoperative memory function as these patients are at greatest risk of postoperative memory dysfunction following more well-established destructive techniques such as anterior temporal lobectomy (ATL), selective amygdalohippocampectomy (SAH), or stereotactic laser ablation of the hippocampus and amygdala . With time, MHT was attempted in patients with dominant hippocampal sclerosis (HS) with preserved function with good outcomes . It was initially speculated that MHT could be useful in patients with bilateral hippocampal seizure onset zones (SOZs) . This has not been reported in the English-language literature to date, presumably due to the risk of significant morbidity if both hippocampi are damaged by such a procedure. Taken together, it appears that MHT provides a nonresective treatment option for patients with dominant hippocampal SOZ and preserved associated memory function.

Contraindications

HS was initially considered a contraindication for MHT given concern regarding the ability to safely transect sclerotic tissue without damaging nearby structures.63 Further, such patients were considered to be at lower risk of functional memory decline making ATL or SAH the treatment of choice . However, successful MHT has been reported in a patient with HS and preserved associated memory function . MHT is therefore a reasonable option in patients with a meaningful degree of preserved memory function attributable to the epileptogenic hippocampus. Patients without retained memory function are not candidates for MHT as hippocampal resection is better supported by long-term data. Resection may be considered preferable in young children who have not yet established hemispheric dominance as the remaining hippocampus is expected to take on the dominant role . The precise age at which this relocation of function can no longer occur is not well known though hemispheric dominance is thought to be established by approximately 3 years of age .

Advantages

The major advantage of MHT is the preservation of memory function postoperatively when significant memory deficits would be expected with hippocampal resection or ablation . In some cases, improvement in memory function has indeed been reported . The possibility of preserving memory function while providing good seizure control in patients with preserved memory despite dominant hippocampal epileptogenicity makes MHT favorable compared to hippocampal resection or ablation in such patients.

Disadvantages

Adoption of MHT is presumably hindered by a lack of direct comparison to hippocampal resection. Further, while MHT is not resective, it does result in an irreversible lesion of the treated hippocampus . As discussed above, the surgical approach itself may also be associated with varying degrees of functional memory morbidity. MHT appears to be most effective in patients with a clearly defined hippocampal SOZ ; unfortunately it may not be possible to positively identify such a focus without invasive monitoring which adds an additional barrier to confidently identifying patients who may benefit from MHT.

Outcomes

Overall, MHT has demonstrated good albeit variable rates of seizure control. In case series of 10 or more, seizure freedom rates range from 17% to 93% . In the same reports, Engel’s class 1 outcome rates ranged from 29% to 93% . The majority of cases reported in the literature to date are in adult patients with dominant temporal lobe epilepsy (TLE). Individual patient data has been reported for five pediatric patients. At the last published follow-up, four of five pediatric patients remained seizure-free and one patient had an Engel class 3 outcome . While conclusions regarding efficacy in the pediatric population are limited by a lack of evidence, the extant adult literature suggests MHT may be a promising approach to treating mesial temporal lobe epilepsy (MTLE) while preserving hippocampal function.

Neuropsychiatric outcomes have been reported for 93 unique patients across eight case series . In a 24-patient series, Usami et al. report a transient decrease across multiple memory domains at the cohort level at 1-month postop which returned to baseline at the last visit (at least 5-year follow-up) . Transient neurologic deficits including memory dysfunction, word finding, and diplopia have been reported in other case series as well, all of which were resolved by the last follow-up . Only five patients in the published literature are reported to have persistent memory dysfunction, accounting for roughly 5% of reported cases . Four patients are reported to have improved memory function postoperatively and putatively due to reduced seizure burden . Neuropsychiatric results are individually reported in only four pediatric patients. All four pediatric patients had stable memory function across multiple domains with two showing improved mood symptoms and one showing improved visual memory . One pediatric patient did show a decline in confrontational naming . No operative complications have been reported to date in the pediatric and adult literature . Overall, MHT appears to be a safe and effective treatment for epilepsy arising from an otherwise functional hippocampus, allowing for the preservation of memory function while demonstrating good efficacy in reducing seizure burden.

Description of techniques

Responsive neurostimulation (RNS) is a closed-loop system which focally stimulates gray matter via subdural or depth electrodes in response to the detection of epileptiform discharges. It was initially thought that this burst of stimulation aborted epileptic seizures by disrupting synchronized activity . Long-term follow-up studies, however, demonstrated an increased efficacy over time suggesting a chronic remodeling process contributes to RNS seizure reduction . Further work implicates the indirect effects of RNS via the plasticity of epileptic networks as the primary mechanism by which RNS reduces seizure burden .

The NeuroPace RNS System is currently the only closed-loop ECoG and stimulation system commercially available. The RNS System consists of a cranially implanted pulse generator which can be connected to two subdural or depth electrodes. Each electrode has four contacts which provide continuous ECoG monitoring. A programable detection algorithm in the pulse generator identifies epileptiform activity and delivers stimulation via the attached electrodes to epileptogenic zones .

Implantation of the RNS system requires first identifying the treatment target(s) and then the appropriate electrode coverage. The invasive recording is typically employed for localization (90% in a recent netaanalysis) with stereotactic electroencephalography used more often than subdural strips, subdural grids, or depth electrode recording . After localization, depth electrodes and/or subdural strip electrodes are implanted. Depth electrodes are available in 30 and 44 cm lengths with the longer being intended for targets contralateral to the stimulator implant. Contacts are spaced at 10 or 3.5 mm apart with the former being preferred for a trans-occipital approach to the hippocampus and the latter being preferred for small discrete targets . The trajectory of depth electrodes should be along the long axis of the target region . Subdural strip electrodes are also available and have been used in conjunction with depth electrodes . Subdural electrodes can be placed via burr holes and are used over cortical tissue. During the initial adult clinical trials, as many as four electrodes were sometimes implanted, though only two may be active at a time as the stimulator has only two ports . The pulse generator is typically implanted in the frontoparietal region as the skull curvature here best matches the curvature of the device . “C” or “U” shaped incisions are reported in the literature with the critical point being that there is no incision overlying the electrodes so that any revision surgery using the same incision will not damage the electrodes . Placing an RNS system in a child requires special considerations due to the smaller skull which can limit possibilities for placement due to the sharper curvatures and thinner skull. This may make the metal tray (ferrule) that holds the stimulator sit proud leading to stress on the overlying scalp or undesirable contour irregularity.

After implantation, a variety of parameters can be adjusted in the outpatient setting, including modifying the detection algorithm and modulating the frequency, current, pulse width, and burst duration of the stimulator . The system applies biphasic, charge-balanced stimulation to maintain charge equilibrium . The system also has the ability to deliver four additional, independently programmable stimulations if epileptiform activity is not ameliorated by the initial stimulation. The typical patient in adult trials of RNS for focal seizures (i.e., not limited to MTLE) had 600–2000 detection and stimulation events per day, typically accounting for less than six minutes of stimulation each day. This yielded a reported battery life of approximately 8.4 years in this cohort . Due to the continuous recording, the RNS can serve as a diagnostic tool in patients with multiple SOZs as it can reveal that one target is responsible for the majority of seizures which can ultimately have therapeutic implications (e.g., bilateral hippocampal epilepsy).

Deep brain stimulation (DBS) has been used across various neurologic disorders since its inception. The use of DBS for epilepsy has been discussed in the literature for several decades and is supported by Class I evidence provided by the Stimulation of the Anterior Nucleus of the Thalamus for Epilepsy (SANTE) trial . The DBS system consists of a generator implanted superficial to the pectoralis major connected to cranial depth electrodes via subcutaneously tunneled wires . Depth electrodes are placed using standard stereotaxy techniques . The effect of DBS is generally not considered to be lesion-mediated though there is likely a microlesional component in at least some cases . Stimulating targets such as the anterior nucleus of the thalamus (ANT) and centromedian thalamus (CMT) at high frequency are thought to functionally inhibit these targets, potentially interrupting spiking synchronization across the epileptogenic network . At a network/circuit level, stimulating the ANT is also thought to modify the limbic circuit of Papez, reducing the epileptogenicity of the circuit . Stimulation of the CMT, which interacts with the ascending reticular system and has broad (predominantly frontal motor and premotor) cortical projections, is thought to interrupt synchronized spiking and has been shown to abort generalized seizures . Accordingly, ANT is the favored deep nucleus target for MTLE while CMT is considered most effective in the treatment of generalized and multifocal epilepsy . Hippocampal stimulation using DBS systems has been reported in MTLE patients as well, potentially facilitating similar circuit-level remodeling as has been proposed as the mechanism for chronic RNS efficacy . Herein, however, we will focus principally on the stimulation of deep thalamic nuclei when discussing DBS.

Indications

Neuropace’s RNS System was approved for use in adults with focal-onset MRE in 2013 . The use of RNS in pediatric patients remains off-label and is therefore reserved for palliative treatment of focal-onset MRE when patients are not candidates for resective surgery .

DBS of the ANT has been approved for the treatment of adults with MRE following the SANTE trial . DBS is not currently approved for pediatric epilepsy though there is an FDA-Humanitarian Device Exemption for pediatric dystonias in children over 7 years of age . The use of ANT DBS for pediatric epilepsy therefore remains off-label and is reserved for palliative treatment of MRE in patients who are not candidates for resective surgery .

Contraindications

RNS has not demonstrated efficacy in primary generalized epilepsy. The system allows for only two electrodes to be active at once, thereby limiting its potential in treating multifocal epilepsy with three or more foci. Both DBS and RNS may only be used as off-label palliative therapies for MRE in patients who are not good resection candidates. Neurostimulation treatment of epilepsy in pediatric patients is therefore contraindicated in patients who have not adequately trialed two or more AEDs or are candidates for more well-established resective, ablative or disconnection approaches if the epileptogenic zone is noneloquent.

Advantages

The principle advantage of RNS compared to other surgical approaches for treating focal-onset epilepsy is that it does not damage the brain and thus can be used to treat eloquent brain tissue while maintaining function . Another major diagnostic and scientific advantage of RNS is the collection of long-term ambulatory ECoG data. Indeed, on a retrospective review of patients with presumed bilateral MTLE who underwent bilateral MTL RNS, 16% eventually underwent MTL resection of the more active seizure-generating MTL or were determined to have a unilateral epileptogenic zone which was then resected. At three-month follow-up, all patients had at least a 50% reduction (median 100% reduction) in disabling seizures, and all patients found to have unilateral onset by RNS ambulatory ECoG were seizure-free at the last follow-up following resection of the pathologic MTL . Extensions of the initial trials of the RNS stimulator in adults have also shown utility in using RNS ECoG data to objectively evaluate the efficacy of AED changes postimplantation .

While there is a dearth of evidence regarding the use of DBS for pediatric epilepsy, the safety of the DBS system in pediatric patients has been well-established by its use in pediatric dystonia treatment . Additionally, if the principal mechanism of RNS is network remodeling, DBS may achieve the same result with a more rapid onset of therapeutic effect .

Disadvantages

There is very little literature regarding the application of RNS and DBS in pediatric patients. Accordingly, there is no FDA indication for the use of neurostimulation in pediatric patients. While these approaches are not meant to introduce new lesions, they do risk damage to surrounding parenchyma and vasculature during implantation . Increased rates of depression and memory impairment symptoms were reported in adults undergoing bilateral ANT DBS in the three-month blinded phase of the SANTE trial though no evidence of neurobehavioral deterioration was captured at long-term follow-up with formal neuropsychological testing . From a surgical perspective, there is a risk of infection associated with the implantation of hardware which may require additional future operations. Finally, older DBS models and current RNS systems do not have transcutaneously rechargeable batteries, necessitating future surgeries to maintain therapeutic effect.

Outcomes

The use of RNS for focal-onset MRE in adult patients is supported by Class I evidence provided by the RNS System in Epilepsy Study Group . This multiinstitution, double-blinded randomized controlled trial demonstrated an initial reduction in seizure burden of 37.9% in the treatment group compared to 17.3% in the sham group (implanted system which was not active during the blinded period). There was no difference between mesial temporal onset and other SOZs . Nine-year follow-up on the same cohort found a median seizure frequency reduction of 75%, indicating a delayed increase in benefit from RNS treatment . Long-term (mean 6.1 years) open-label follow-up of a subset of this cohort with MTLE also showed improved efficacy over time with a mean seizure reduction of 70% at the last visit. In this subset, 29% had experienced at least one 6-month period of seizure freedom and 15% had experienced at least one 12-month period of seizure freedom . There was no difference in primary outcome associated with mesial temporal sclerosis, bilateral (vs unilateral) onset, nor prior resection within this long-term MTLE subgroup . A case series ( n =10) from an independent research group applying RNS to the parahippocampal white matter or temporal stem for refractory MTLE demonstrated a 44% reduction in seizure frequency at 3 years follow-up .

The studies which have established RNS as an approved treatment for MRE in adult patients notably exclude pediatric patients. Evidence in the literature regarding the safety and efficacy of RNS in pediatric patients is limited to case reports and case series . These cases have recently been summarized in both a systematic review ( n =46) and a meta-analysis ( n =49) which report overall responder rates of 73.2% and 79.5%, respectively with no reported postoperative neurologic deficits . Overall, seizure reduction rates and responder rates were similar when subgroup analysis compared patients under 12 years old and patients 12–18 years old . Subgroup analysis of temporal lobe onset patients identified seven cases with a mean seizure frequency reduction of 50% . Since these summative studies were published, two independent larger case series have been published . One series ( n =20, age 8–21 years old) found 57% of patients had meaningful improvement with RNS while another series reported an 85% responder rate ( n =22, age 6–22 years old) . These series include patients with various seizure etiologies and foci. Four patients in one series are identified as having mesial temporal lobe onset with two having Engel class IIIA outcomes and two having Engel IVA outcomes . Taken together, the pediatric RNS literature provides evidence that RNS can safely be used in pediatric patients and does demonstrate some degree of efficacy. However, given the limitations and high risk of bias inherent to case reports and case series, multiinstitution randomized controlled trials would be ideal to determine the true efficacy of RNS in pediatric patients. Further, there are very few examples of pediatric MTLE patients undergoing RNS implantation making it difficult to conclude the efficacy of RNS in this unique subset of epilepsy patients.

Class I evidence for DBS in the treatment of adult MRE comes from the SANTE Trial which included adult patients with various SOZs, though the majority (60%) of patients in this study had TLE . The SANTE Trial demonstrated a 44% reduction in seizure frequency with stimulation of the ANT in TLE patients. Similarly to the RNS literature, long-term follow-up of the TLE subgroup in the SANTE trial showed improved efficacy with a 76% reduction in seizure frequency at 5-year follow-up .

Evidence for DBS use in pediatric epilepsy is limited to case reports to date. Stimulation of the ANT with DBS has been reported in five pediatric patients, ranging in age from 14 to 17 years old . One patient (17 years old) underwent ANT DBS for refractory status epilepticus with immediate resolution upon activation of the DBS . Two patients underwent ANT DBS for poorly localized multifocal epilepsy with 50% reduction in seizure frequency at 11 months follow-up and 93% seizure reduction at 2-year follow-ups . Another patient underwent ANT DBS for bilateral frontal seizures with no discernable effect . Finally, one patient underwent ANT DBS for frontotemporal epilepsy with >60% reduction at 12-month follow-up . While these cases demonstrate that DBS of the ANT in pediatric patients is feasible, there is no direct evidence for its use in pediatric MTLE patients.

Neurostimulation broadly has not been associated with neuropsychiatric decline across stimulation devices and targets, and neurostimulation—likely via improved seizure control—has indeed been associated with improved cognition and memory in some cases . The most common adverse event in the initial adult trial of RNS for refractory focal onset epilepsy was headaches, reported by 10% of the study population postimplantation . Other neurologic adverse events included dysesthesia in 6.3% of patients, increased tonic-clonic seizures in 4.7%, subjective memory decline in 4.2%, and depressive symptoms in 3.1% of patients within the first year of follow-up . Notably, this analysis included symptoms with unclear association with the RNS system. Indeed, 50% of patients in the study had a history of depression prior to implantation . There was a decreased prevalence of moderate depression and suicidality in the treatment group (vs sham) during the blinded period and there were no overall adverse effects on mood or suicidality on long-term follow-up . In the blinded phase of the SANTE trial, an increased prevalence of subjective memory impairment was reported in the active group (13.0% vs 1.8% in the control group) . Despite reports of subjective memory dysfunction, objective memory measures showed improvement in verbal memory and verbal learning in MTLE patients . On long-term follow-up, the SANTE group found no significant worsening of depressive symptoms and no significant cognitive decline in ANT DBS patients . An independent group investigating ANT DBS for MRE found improved verbal fluency and delayed verbal memory and an otherwise stable neuropsychiatric profile at least 1-year postimplantation . Sleep disruption has been reported in association with ANT stimulation period in a nine-patient case series and the effect size was correlated with stimulation voltage .

Notable surgical complications of neurostimulation include intracranial hemorrhage, implant-site infections, and skin erosion overlying device implants. In the large RCT trial of RNS, 3.1% of patients had postoperative intracranial hemorrhage—three epidural hemorrhages requiring surgical evacuation, two nonoperative intraparenchymal hemorrhages, and one subdural hemorrhage . Subdural hemorrhages outside of the postoperative period were seen in 1.6% of patients in this study and were associated with seizure-related head trauma . Implant/incisional site infections were seen in 5.2% of patients in the RNS trial (3.1% in the first year), with four of those 10 patients requiring device explant . The most common surgical complication was implanted site pain which was predominantly in the first four weeks postimplantation . Follow-up reports from this cohort report a rate of 0.03 device-related adverse events per implant year and a 3.7% implant site infection rate per implantation procedure . In the SANTE trial, paresthesias were reported in 18.2% of patients, including participants from both the treatment and sham groups . Other adverse events included implant site pain in 10.0% of patients and implant site infections in 9.1% of patients, predominantly in the first year postimplant . Importantly when considering RNS or DBS use in pediatric patients, skin erosions overlying device components are likely more common in pediatric patients .