Presurgical video-EEG monitoring with foramen ovale and epidural peg electrodes: a 25-year perspective

In selected patients with focal drug-resistant epilepsy (DRE), surgical removal of the epileptogenic zone is the most efficacious therapy. The aim of preoperative assessment is to delineate the tissue that needs to be removed, and to define the postsurgical benefit–risk ratio. In some cases, electrophysiological information derived from scalp video-EEG monitoring (VEM) is insufficient to reach this aim [1]. One limitation of scalp EEG is poor spatial localization of the seizure onset zone. Ictal events originating from deep or interhemispheric structures may not be visible on scalp EEG, thus providing no additional information, or may be misleading as the epileptic activity can appear bilaterally or contralaterally to the epileptogenic zone. This is particularly challenging in temporal lobe epilepsy, due to the tendency of seizures for prompt bilateral spread and due to low sensitivity of scalp EEG to electrical signals generated in mesial temporal structures [2]. Another limitation of scalp EEG is that physiological artifacts, e.g., by muscle activity or movement, may obscure ictal onset and interpretation of findings [1, 3].

Thus, non-invasive investigations at times must be supplemented by subsequent intracranial EEG (iEEG) diagnostics. The ILAE-defined indications for iEEG include to delineate the epileptogenic zone when prior non-invasive workup was inconclusive and to resolve divergence of non-invasive data pointing to several possible epileptogenic sources [4]. The most common iEEG diagnostics in use include stereotactic depth EEG (SEEG) and subdural electrodes (SDE). A main advantage of SDE is that it provides wide coverage of neocortical targets and that eloquent cortical structures can be defined by electrical stimulation; furthermore, this recording approach can also assist intraoperatively [4]. SEEG can be implanted bilaterally providing electrographic information from deep brain structures. These procedures, although highly informative, pose increased risks to the patients including a major complication rate of 9.6% for SDE implantation and 4.4% for SEEG [5]. Furthermore, studies have reported a detrimental cognitive effect caused by hippocampal depth electrodes implanted in the unaffected side [6, 7]. These complication rates may contribute to the fact that up to 50% of potential surgical candidates reject their treating physicians’ recommendations for implementation of iEEG investigations, thus excluding these patients from highly effective surgical therapy [8].

A diagnostic approach using foramen ovale (FO) and epidural peg electrodes (FOP) likely reduces risks to the patients. In this approach, electrodes are inserted through the FO and thus placed in close proximity to mesial temporal structures, while peg electrodes are inserted epidurally via burr holes in the skull over hemispheric regions. The electrophysiological rationale for utilizing FOP diagnostics is that FO electrodes can provide good sampling of mesial temporal structures and thus lateralization of the seizure onset zone [9, 10], while peg electrodes may give lateralizing and in some cases additional rough localizing information from the convexities. A limitation of this approach is that FO electrodes cannot sample deep-seated temporal structures, whereas peg electrodes are limited to superficial cortical regions. Compared to scalp VEM recording, FOP electrode recordings have reduced interference due to physiological artifacts including muscle, blinking, and movement.

Since first reported over 30 years ago [11, 12], several studies have described the clinical utility and safety of FOP VEM [9, 11, 13‐20]. These studies have reported positive results as to the clinical utility; however, they have mostly focused on FO electrodes in patients with mesial temporal lobe epilepsy. In this population, the most recent studies reported that a treatment decision for or against resection could be reached in 75–90% of cases [18, 19]. Studies examining combined FO and peg electrode recording are scarce, mainly describing experience from 1990s, a period different with regard to availability of modern day neuroimaging diagnostics including high resolution MRI [12, 21, 22]. Furthermore, the extent to which these methods can provide meaningful information in patients with extratemporal epilepsy is yet unexplored.

At the Epilepsy-Center Berlin-Brandenburg, a presurgical VEM approach utilizing FO and additional peg electrodes has been in regular use since the mid-1990s. Thus, after 25 years of experience, a unique cohort of patients has accumulated, allowing for reappraisal of the clinical utility of this method. This study aims to give a broad overview of the advantages and limitations of a presurgical FOP VEM approach.

In this study, we reviewed VEM findings of patients that underwent presurgical evaluation with FO and/or epidural peg electrodes at our center during the past 25 years. Our main findings are that in two-thirds of patients a clinical decision could be made directly following FOP VEM, including a recommendation for resection in one in five patients. Clinical decisions were facilitated by a significant increase in electrophysiological information on FOP VEM compared to previous noninvasive VEM. Furthermore, we identified predictors that are useful for correctly selecting patients for FOP diagnostics as well as providing prognostic postsurgical information. Finally, we report a low major complication rate during FOP diagnostics.

Previous studies have focused on the utility of FO recordings in the presurgical evaluation of patients with temporal lobe epilepsy [11, 13, 16, 18]. These included a study reporting that FO-based VEM resulted in treatment decisions in 90% of 42 adult patients with mesial temporal lobe epilepsy [19], and another study that found resection could be decided on in 45% of 38 children with previously suspected temporal lobe epilepsy [15]. Studies describing the utility of epidural peg electrode recording covering the convexities or a combined FO and peg electrode recording approach, however, are sparse. We have found no such studies of patients treated in the past 20 years, and studies describing experience in the 1990s are also limited [12, 14, 22, 24, 25]. In the largest series including 59 patients with combined peg and FO recordings, for 31% of patients, a resection was recommended, in 49% of patients additional invasive investigations were needed, and 15% of patients were disqualified from surgery [14]. Compared to these previous reports, our cohort is unique as it is larger, covers a longer time span, is composed of a more heterogenous group of DRE patients allowing for generalization of findings, and reports the use of combined FO and peg recordings. In our study, one major clinical benefit of FOP VEM was that resection could be recommended directly following the investigation among 20% of all patients and among 33% of those with temporal lobe epilepsy. In 47% of patients, resection was ruled out. The rate for resection recommendation was lower than that reported after invasive EEG investigations, with a recent comparative study finding that 78.6% of SDE patients and 66.5% of SEEG patients were likely to be resected [5]. However, postsurgical 1-year seizure freedom rate in our cohort was 53%, which is equivalent to that reported for surgery following invasive EEG diagnostics [5, 26]. Although FOP VEM lacks major advantages of invasive diagnostics, i.e., high spatial resolution, and specifically in the case of SEEG the ability to sample a multitude of deep-seated temporal structures, the equivalent post-surgical seizure freedom rate suggests that for a well-selected group of patients this diagnostic approach can provide data needed to resolve a clear hypothesis regarding the seizure onset zone, without the higher risks for complications associated with SDE or SEEG diagnostics.

The added electrophysiological value of FO electrodes compared to scalp for lateralization in patients with temporal lobe epilepsy has been previously described. A study examining simultaneous FO and scalp recordings of 314 seizures found that in over 40% of cases, seizures with apparent bilateral scalp onset start unilaterally on FO electrodes [9]. Another study reported that in 60% of temporal lobe epilepsy patients with bilateral or nonlateralizable scalp ictal onset, FO recordings were successful in lateralizing the epileptogenic source [17]. Our results also show significant improvement in identification of the ictal onset side and/or region on FOP VEM compared to noninvasive VEM. A clear side of ictal onset was identified in 68% of patients with previous nonlateralizable noninvasive VEM, including 30 patients (56%) that were subsequently recommended surgery. Similar to prior reports of scalp and FO VEM ictal discordance [9, 17], we found that in many cases where unilateral or independent bilateral ictal onsets are determined by noninvasive VEM, a more complex picture of the ictal onset is revealed during FOP VEM. Thus, in our study, 38% of patients with prior independent bilateral scalp ictal onsets had unilateral onset on FOP VEM, 14% of patients with prior scalp unilateral onset had bilateral ictal onset and 9% had contralateral unilateral ictal onset on FOP VEM. Although lateralization in itself is insufficient for surgical planning, as the seizure onset zone must be localized for successful resective therapy, it may provide important information for the clinician in resolving a hypothesis regarding the seizure onset zone (e.g., cases with divergent noninvasive findings). Scalp to FOP VEM discordance is possibly due to fast ictal propagation, a long latency between clinical ictal onset and electrophysiological changes on scalp EEG, or due to physiological artifacts. Interictally, we also found a significant addition of electrophysiological information, with only 2% of patients without interictal activity on FOP VEM compared to 19% on scalp VEM. Considering interictal activity was found to be a predictor of a recommendation for resection, we take this added information to be clinically valuable. Figure 2B and C demonstrates the advantage of improved sampling and reduced artifacts on FO and peg recordings in a patient with simultaneous scalp recordings.

Data derived from FO and peg electrodes served different, yet complementary roles. FO seizure onset was an independent predictor of resection recommendation and was the onset electrode in 86% of temporal lobe resections directly recommended following FOP VEM. However, in patients with clear ictal onset, peg electrodes provided additional information for excluding from surgery (4 FO and 12 peg onset, 34 both electrodes) or lateralization in patients requiring additional invasive (4 FO and 31 peg onset, 16 both) EEG investigation. In this group of patients, additional invasive diagnostics were aimed at defining eloquent cortex or further localization of the seizure onset zone. Of all 59 patients that required additional invasive testing, only in eight patients lateralization was unclear after FOP VEM, requiring additional bilateral iEEG. Thus, potentially, invasive EEG procedures in an uninvolved hemisphere aimed at epilepsy lateralization were avoided. Nevertheless, it is evident that the main advantage of FOP VEM was in decision-making of cases of temporal lobe epilepsy.

In examining trends over time of FOP VEM use, we found that in the early years of the study period, there were significantly higher rate of patients that required additional invasive diagnostics compared to the later years. One could hypothesize that this is due to advances in non-invasive investigations or experience accumulated with the use of this diagnostic approach in our center. Notably, 14 patients (23%) refused additional iEEG investigations following FOP VEM, possibly due to fatigue from procedures or due to discomfort caused by and fear toward the diagnostic process. It is possible that in these patients the opportunity for curative treatment was missed. Furthermore, one must keep in mind that those patients requiring additional invasive diagnostics following FOP VEM are exposed to more risks and incur higher medical costs, emphasizing the need for correct patient selection for this diagnostic approach.

To assess which patients are most appropriate for FOP VEM, we examined predictors of clear ictal onset and resection recommendation. Multivariate analysis revealed that temporal lobe epilepsy was an independent predictor for both these outcomes. Lesional imaging, although a predictor for identification of a clear ictal onset on FOP VEM, did not predict resection recommendation. Both ictal-onset electrode and interictal activity found on FOP VEM were, however, significant predictors of resection recommendation, emphasizing the benefit derived from additional FOP electrophysiological information for clinical decision-making.

On examination of the indications for FOP VEM, we found that an indication of resolving divergent non-invasive investigations was associated with both successful ictal onset determination and a resection recommendation. Considering that most divergent data were related to bilateral lesions on imaging or bilateral EEG ictal onsets, this suggests FOP VEM is more appropriate to answer a clear question of lateralization, rather than to find a seizure onset zone in nonlesional patients, emphasizing the need for a clear hypothesis prior to performing FOP VEM studies. However, clinicians should keep in mind the technical limitations of FOP diagnostics, specifically the fact that peg electrodes are unable to sample distant cortical areas, whereas FO electrodes are most appropriate for sampling electrophysiological signals from the mid- to posterior hippocampus, and may not be adequate for other deep-seated temporal lesions.

Providing patients with prognostic information is a central part of presurgical assessment. Previous studies have reported clinical, EEG and imaging factors associated with postsurgical outcome [27, 28], as well as a clinical algorithm for prediction of clinical prognosis [29]. We add to this literature by reporting our finding that FO electrode seizure onset was an independent predictor of seizure freedom. Notably, FO onset was not significantly associated with MTS pathology, and thus did not simply reflect the presence of this known predictor of postsurgical seizure freedom. [30] Possibly, onset on FO electrode represents early activation of mesial temporal areas of the seizure network. In any case, FO electrode onset is easily assessed during FOP VEM and could possibly be integrated in future seizure prediction models of patients undergoing FOP VEM.

We report a similar rate of major complications compared to previous studies on FO or peg electrode recordings [20], with two patients having experienced intracerebral bleeding. Notably, both patients were treated conservatively without neurologic sequalae. In contrast to previous studies reporting minor adverse events in 2.6 to 6.6 percent of patients [16, 18‐20], in our cohort, there was a much higher rate of 31% (of 136 patients with information available regarding complications). This difference is possibly attributed to the fact that we had a low threshold in reporting minor adverse events including transient sensory disturbances as well as non-CNS infections indirectly related to FOP procedures occurring during the VEM. Although in 44 patient files, there was no reference to complications, we find it unlikely that these patients had complications, as in this case, there would be additional documentation or a reference on discharge letters. Thus, the rate of minor complications in our total cohort may be lower than 31% but still higher than in other reports on FOP VEM. This high rate of minor complications should be discussed with the patient prior to FOP evaluations and thoroughly considered when deciding on the most appropriate diagnostic modality. Furthermore, clinicians should take in account that complications during FOP VEM may lead to patients being hesitant to undergo additional invasive investigations, if those are needed. Compared to SDE and SEEG investigations, we report a lower major complication risk. In a recent study of SDE and SEEG in epilepsy surgery, 9.6% of patients undergoing subdural electrode recordings and 4.4% of patients undergoing SEEG experienced a major complication. An additional meta-analysis has reported that SDE carry a risk of intracranial hemorrhage in 4% of patients, as well as CNS infections in 2.3% and increased intracranial pressure in 2.4% of patients [31]. In our cohort, patients that required additional SDE testing also had a similar major complication rate of 7.1%. The use of SEEG electrodes in temporal lobe epilepsy, although associated with lower complication rates compared to SDE [32], has been reported to carry an additional risk of detrimental cognitive effects due to implantation in the unaffected mesial temporal lobe [6, 7]. Avoiding this risk is a possible advantage of FOP VEM over SEEG recordings.

An important limitation of this study is that clear ictal onset was defined according to the treating physicians, and thus is prone to observer bias. On the other hand, physicians were highly experienced epileptologists, and over the two and a half decades covered by this study, the treating physicians involved were relatively constant, diminishing interrater variability and allowing for good longitudinal assessment. Another limitation comparing non-invasive and FOP VEM findings is that most patients underwent presurgical investigations at different times, allowing for differences in pharmacological treatment regimens which could affect both interictal and ictal epileptic activity. An additional limitation is that this study is retrospective, and thus standards of clinical documentation, seizure classification definitions, and imaging quality in the beginning of the study period were different compared to recent years. Furthermore, patient files in the early years of this study were less detailed, and for 44 patients no report on presence or absence of complications was found.

We report our extensive experience with presurgical assessment combining FO and epidural peg recordings in a heterogeneous population of patients with DRE. FOP VEM, in selected patients, adds clinically significant electrophysiological information leading to clear treatment decisions in two-thirds of patients, while maintaining a low complication rate. We hope our report of a favorable benefit–risk ratio of FOP VEM leads to more patients undergoing intracranial EEG investigations and subsequent resective surgery, to become seizure-free.