CHRONIC AMBULATORY EEG WITH IMPLANTED ELECTRODES
BENJAMIN N. BLOND, MD and LAWRENCE J. HIRSCH, MD
The gold standard for the definitive diagnosis, characterization, and localization of epilepsy remains inpatient monitoring. However, as discussed throughout this text, ambulatory EEG provides several advantages over standard inpatient monitoring, including longer term observation than is practical in an inpatient stay and an ability to monitor patients in their home environment on their normal medication regimen with exposure to normal triggers. These benefits come at the cost of increased artifact from inability to continually adjust electrodes and often from lack of continuous video monitoring. A separate major advantage of inpatient monitoring is the ability to perform intracranial EEG, whether subdural (electrocorticography [ECoG]), intraparenchymal via depth electrodes (including stereo-EEG), or a combination. This intracranial monitoring substantially increases spatial resolution and dramatically improves the signal-to-noise ratio, largely eliminating artifact from the recordings. New fully implanted devices now exist, with more being developed, which allow for chronic intracranial EEG in the ambulatory setting. The current and future potential implications of chronic ambulatory intracranial EEG (CAIEEG) are the focus of this chapter.
ECoG refers to the recording of electrical activity from the surface of the cerebral cortex. The main indication for ECoG has been for the surgical treatment of refractory epilepsy when other tests to identify the seizure focus are discordant or inconclusive, when there is no MRI abnormality (except select medial temporal cases), when the seizure onset zone abuts eloquent cortex (including many lesional cases), and when there is dual pathology (eg, hippocampal sclerosis plus a lesion) (1). Although some cortical mapping and identification of the irritative (“spiking”) zone can be done via brief intraoperative ECoG, implantation of electrodes in order to monitor over 1 to 2 weeks is usually required in order to identify the seizure onset zone and its relationship to eloquent cortex. Complete removal of the seizure onset zone is associated with a greater chance of seizure freedom, even after accounting for lesion resection (2).
Because ECoG is recording from the surface of the brain, it is able to detect high-frequency activity that scalp recording cannot detect due to the intervening skull, muscle, and other soft tissues. High-frequency oscillations (HFOs; ripples: 80–250 Hz and fast ripples: 250–600 Hz) may help to localize epileptogenic tissue and may be more localizing than traditional interictal epileptiform discharges (3,4).
ECoG is not without risk. Complications of implanted intracranial electrodes, and associated extracranial wires, amplifiers, and other equipment, including raised intracranial pressure (ICP), subdural hematoma, and infection occur in about 9% of patients and are mostly transient with permanent deficits in less than 2% and rare mortality (5). Risks are higher with greater numbers of implanted electrodes, larger subdural grids, and perirolandic location. Most significant for the current discussion, risks are also increased with increased length of monitoring, at least in part due to infection risk (6).
CHRONIC AMBULATORY INTRACRANIAL EEG
As described in the previous section, standard intracranial EEG provides many advantages in regards to sensitivity and localization of epileptogenic foci. However, there are significant deficiencies. In addition to the morbidity from an invasive procedure, there are limitations to the length of time such recording can take place, due to the increasing risk of complications, especially infections. This short duration, typically 1 to 2 weeks for most patients, comes with inherent limitations, as this represents an artificial period subject to potential confounders of the acute surgical procedure, medication withdrawal, and inpatient stay. Scalp ambulatory EEG, as discussed in the preceding chapters, provides comparative advantages in capturing patients without these acute confounders. CAIEEG could theoretically provide profound insights with longitudinal assessment of an individual patient’s seizures, but the quickly accruing artifact as a result of the lack of continuous electrode maintenance limits such investigations to typically a 1- to 3-day period.
Fully implantable devices are now available for recording CAIEEG. This has the potential to be truly revolutionary, as it provides the advantages of long-term ambulatory monitoring while eliminating the main limitation of artifact, and concurrently provides the advantages of spatial resolution of intracranial EEG without the confounders present in the acute setting. Such devices may allow seizure prediction and warning, which would improve patient safety and quality of life, as well as allow responsive treatment for seizure prevention, such as with electrical stimulation (currently available via the responsive neurostimulation [RNS] device), cooling, medication administration, or combinations of these. Electrocorticographic signal analysis is also useful for brain–computer interfaces.
Prior research has demonstrated that “open-loop” stimulation systems (stimulate on a preset schedule regardless of brain activity, etc.), such as vagus nerve stimulation and anterior thalamus deep brain stimulation, can modestly reduce seizure frequency, demonstrating improved seizure control over time, with approximately 50% of subjects achieving at least 50% reduction in seizure frequency after several years (7). A closed-loop system would theoretically be superior, as providing stimulation only when necessary to abort seizures should be able to reduce adverse effects, minimize battery use, and expand the possible locations where stimulation could be safely administered. New intracranial devices have been developed that monitor continuous intracranial EEG for just such a purpose. Seizure detection programs can be tailored to individual patients based on prior EEG data for optimal accuracy and can be adjusted as needed. Once a seizure is detected, the device can apply electrical stimulation in order to disrupt seizure propagation, with the hope of preventing evolution into a clinical event. One of these devices, the RNS® System (NeuroPace, Mountain View, California) was Food and Drug Administration (FDA) approved in late 2013 for the treatment of focal onset epilepsy based on the results of a randomized double blind trial demonstrating decrease in seizure frequency with use of responsive stimulation compared with sham stimulation (8). This device provides responsive cortical stimulation via a cranially implanted programmable neurostimulator connected to one or two recording and stimulating depth or subdural cortical strip leads, each of which contains four electrodes (Figure 9.1). These strips or depth electrodes are surgically placed in the brain in the predetermined seizure focus. The neurostimulator continually senses intracranial EEG activity and is programmed by the physician to detect abnormal activity and then provide stimulation (Figure 9.2 provides an example of this). The physician adjusts detection and stimulation parameters for each patient to optimize detection and control of seizures. The device is capable of storing about 6 minutes of four channels of intracranial EEG data for later review (or 12 minutes of two channels), plus diagnostic and therapeutic data for the past couple hundred of detections and stimulations (without the full EEG tracings); the exact storage can be predetermined by the clinician. Detection is based on amplitude, frequency, and other pattern features on two different channels and can be combined. Patients can also swipe a magnet over the location of the implanted device to leave a marker and to store a sample at that time, and a preceding time interval, also programmable by the clinician. The stored intracranial EEG recordings are downloaded from the device to a dedicated laptop (where there is nearly unlimited storage capacity), and the data are uploaded to a secure website where the treating MD can review it (and try out new detection parameters if desired). At our center, we encourage patients to download their intracranial EEG recordings daily to maximize the data available for interpretation and to keep clinical diaries to correlate their reporting with the device recordings. The most recent follow-up study of the efficacy of the RNS system examined patient data from an average of 5 years with the device. The initial reductions in seizure frequencies of around 50% were sustained and even improved to about 60%, and the responder rates were also about 60% at these time points (9). Quality of life, as measured by the QOLIE-89, was significantly improved (9). For a review of this subject, see Fisher and Velasco (7).
Source: From www.neuropace.com/manuals/RNS_System_User_Manual.pdf, with permission.
As with any new technology, prior to consideration of potential advantages, it is important to attempt to systematically validate that the diagnostics are reliable and accurate. This was formally assessed in a recent study by Quigg et al (10). Five reviewers formed five pairs that reviewed 7,221 ECoG recordings from 128 subjects. Interrater reliability for identification of seizures was variable based on patients. Half of the patients had agreement rates of 94% or greater. It was the lowest quartile with agreement rates of less than 75%, which weighted down the overall average to a moderate agreement rate of 79%. Reasons for disagreement in these subjects included (a) low amplitude rhythmic activity with limited spatial or temporal evolution until the end of the discharge; (b) discharges with clear evolving activity that were less than 10 seconds because ictal onsets, and thus, the overall duration, were ambiguous; and (c) quasi-periodic discharges during which runs became more organized with frequencies near 2 Hz. These challenging diagnostic situations are not unique to CAIEEG. Much of this work has been completed in analysis of critical care EEG. This process of standardizing terminology (even if chosen arbitrarily) has substantially increased interrater reliability compared with prior terminologies (11). It will be important to continue to develop rigorous terminology to improve consensus and allow research on highly epileptiform intracranial EEG patterns of varying durations.
ADVANTAGES OF CAIEEG
Elimination of Acute Confounding Variables
A potential advantage of CAIEEG, akin to other forms of ambulatory EEG, is the ability to remove the confounders of the acute hospital environment. Seizures captured in the acute setting may be partially provoked and therefore not typical. An often cited benefit of ambulatory EEG is the ability to detect seizures when patients are on their home medications. In order to capture seizures during the short duration of stay necessitated by inpatient monitoring admissions, rapid withdrawal of antiepileptic medications is common practice. This can lead to adverse effects including a higher incidence of secondarily generalized seizures (12). There has been some literature to suggest that seizures provoked by rapid withdrawal from antiepileptic medications may not always be typical of baseline seizures and may be falsely localizing (13). Engel and Crandall reported a case of a patient who had left temporal onset of seizures during a rapid medication taper, which were of atypical semiology. The patient was ultimately found to have an infiltrating right temporal meningioma and was seizure free after right temporal lobectomy, supporting the contention that the atypical events could be ascribed to medication withdrawal and were not reflective of the patient’s underlying epilepsy (13). However, although frequently cited as a potential issue, this appears to be uncommon. Studies examining this phenomenon have indeed recorded atypical seizures, but these have also occurred while on therapeutic levels of antiepileptic medications prior to withdrawal in patients who were shown to have multifocal disease or in patients who were later found to have multifocal disease (14). There is no clear evidence that these atypical events lead to significant morbidity or impair final medical decision making (14–16). There is some concern that electrode implantation may itself serve as a provoking factor for seizures and could be an alternate reason for atypical seizures in the acute monitoring setting (14).
Another chief advantage of recording in a subject’s home environment is the ability to capture normal sleep patterns. Epilepsy and sleep are highly interrelated, with both epileptiform activity and clinical seizures being strongly related to certain circadian patterns (17). However, the inpatient environment can be highly disruptive to sleep. Some early work using chronic ECoG from an RNS device over a short time frame in an epilepsy monitoring unit suggested that CAIEEG devices are capable of detecting circadian variations, demonstrating peaks in intracranial EEG energy and complexity at 0530 and 1500 (18). A recent study greatly expanded on this finding by using CAIEEG to study circadian associations longitudinally by following 65 subjects over an average of 676 days in their home environment with presumably baseline sleep patterns (19). Significant circadian rhythmicity of epileptiform activity was seen in 63 out of 65 subjects, and this was seen in all lobes, hemispheres, and for both neocortical and hippocampal foci. The main circadian peaks were at 2300 and 0500, and there were lesser peaks in the early afternoon. These results further support the importance of the relationship between sleep and epilepsy and may imply an important role for CAIEEG for diagnosing or localizing epilepsy during these normal circadian rhythms. Further, understanding the circadian peaks of epileptiform activity may be important for future seizure prediction technology. This may also be used to guide the optimal timing of antiepileptic medication administration or scheduled neurostimulation in order to provide maximal protection during highest risk periods for seizures.
Length of Monitoring
The main advantage of CAIEEG is the capability for long-term monitoring. As mentioned earlier, inpatient intracranial EEG is typically limited to approximately 2 weeks due to the risks of complications and other practical considerations. This results in a population of patients for whom inpatient monitoring is not sufficient to adequately localize a seizure focus. CAIEEG provides an opportunity to obtain substantially more data over months or years, with the hope this may provide more accurate and definitive localization in some of these patients. Long-term monitoring has demonstrated the potential fallibility of shorter periods of observation. An analysis of a case implanted with RNS looked at 54 seizures occurring over 2 years in a patient with bitemporal onsets (20). Although the lateralization was not significantly different for the total counts over the 2-year period, the localization and lateralization of seizures did significantly change over time (Figure 9.3). Note that the patient had predominantly left-sided seizures for the first 5 months of monitoring and then had predominantly right-sided seizures including a 6-month period with only right-sided seizures. This suggests that a short time period of observation could be misleading within an individual patient, due to the dynamic nature of seizures.
The specific question of how CAIEEG can be applied in a surgical evaluation for refractory epilepsy has been analyzed in some detail in a retrospective analysis of 82 subjects with medically refractory bitemporal lobe epilepsy with focal onset seizures participating in the randomized double blind trial of RNS and who were implanted with bilateral temporal lobe electrodes (21). Electrographic seizures were defined as episodes of low-voltage fast activity or rhythmic sharp activity, distinct from background, evolving and lasting longer than 25 seconds (shorter episodes were usually not stored). Out of this group, 71 subjects were presumed to have bilateral onset of seizures. The remaining 11 subjects were presumed to have unilateral seizures but had bitemporal electrodes placed due to additional data suggestive of contralateral pathology, such as bilateral hippocampal atrophy or mesial temporal sclerosis, an intracarotid amobarbital (Wada) test indicating that the contralateral temporal lobe did not adequately support memory (five subjects), a prior contralateral temporal lobectomy (two subjects), or discordant EEG and PET lateralization (one subject).
Source: Reproduced and legend modified from Ref. (20). Smart O, Rolston JD, Epstein CM, Gross RE. Hippocampal seizure-onset laterality can change over long timescales: a same-patient observation over 500 days. Epilepsy Behav Case Rep. 2013;1:56–61.
In the 71 subjects who were presumed to have bilateral seizures, 62 (87%) demonstrated bilateral seizures on CAIEEG. The first contralateral seizure was recorded during the first week of CAIEEG in 38.7% (24–62) and during the second week in 17.7% (11–62). However, first contralateral seizure was not detected until the third week in 6.5% (4–62), during the fourth week in 9.7% (6–62), and after the fourth week in 27.4% (17–62), which demonstrates that approximately 40% of the patients would not have had such information available during a standard 2-week inpatient monitoring stay. Even more surprisingly, 9 of the 71 “bitemporal” subjects displayed only unilateral seizures over an average of 5 years of CAIEEG.
In the 11 subjects with only unilateral seizure onsets documented previously (but implanted bilaterally), CAIEEG showed that 7 (64%) of the 11 subjects had independent bilateral electrographic mesial temporal lobe (MTL) seizures; average time to record the first contralateral electrographic seizure was 72.4 days (median 35 days; range 7–330 days). Four of the 11 subjects had only unilateral seizures recorded by CAIEEG, with an average duration of 3.9 years (median 4.1 years; range 0.4–7.0 years) of recording. Figure 9.4 displays a graph of all 69 patients who were ultimately found to have bilateral seizure onsets and the time required for detection.