Acknowledgments: This research was supported by NIH R01-NS63039 (GW), European Regional Development Fund – Project FNUSA-ICRC (No. CZ.1.05/1.1.00/02.0123), and by the European Social Fund within the project Young Talent Incubator II (reg. no. CZ.1.07/2.3.00/20.0117).

Scalp EEG and intracranial EEG (icEEG) are critical technologies for localization of epileptogenic brain and guiding epilepsy surgery in patients with drug-resistant focal epilepsy and negative MRI. Scalp EEG is universally used to confirm the diagnosis of epilepsy and for developing a hypothesis for the localization of epileptogenic brain tissue. In order to localize the brain region(s) generating seizures, patients with negative MRI often have intracranial electrodes implanted based on the localization hypothesis developed using noninvasive studies (semiology, scalp EEG, and functional imaging). It can be argued that icEEG is the gold standard for localizing the seizure onset zone (SOZ) [1], but whether there is an electrophysiological biomarker for the epileptogenic zone (EZ) remains unanswered [2,3].

The EEG signature of epilepsy is the pathological electrical activity occurring during seizures (ictal), but it has long been recognized that there are also brief epileptiform transients between seizures (interictal), i.e., interictal epileptiform spikes and sharp waves (IES) [4]. The significance of IES in epileptogenesis and seizure generation has been widely debated, but IES have always played a clear and important role in clinical practice for localization of epileptogenic brain tissue. Despite the well-established clinical importance of icEEG and advances in digital electronics and computing that have revolutionized animal electrophysiology [5], clinical icEEG continues to primarily utilize narrow bandwidth (1–100 Hz) recordings from widely spaced (5–10 mm) macroelectrodes (> 1 mm²). A fundamental and unanswered challenge for clinical icEEG is the optimal spatial resolution and recording frequency bandwidth.

Table 12.1 The EEG spectrum of human brain electrical activity

The spatial organization of the human brain extends from neurons, submillimeter diameter cortical columns composed of thousands of neurons, to centimetre-scale lobar networks. The electrical activity generated by these neural assemblies range from direct current shifts to high-frequency oscillations (~0–1000 Hz). This remarkable range of human brain electrophysiology can now be probed with wide bandwidth acquisition systems and hybrid electrodes containing micro- and clinical macroelectrodes [6–8].

The nomenclature used to describe oscillations outside the standard clinical “Berger Bands” (1–25 Hz) is not firmly established, but includes ultraslow, slow, and high-frequency activity that are outside the common clinical bandwidth. In epileptic brain there are a range of pathological transients in addition to IES, including high-frequency oscillations (HFOs) [3,9], microseizures [10], focal ultraslow activity [7,11,12], and alterations in network synchrony and connectivity [13]. These interictal abnormalities are all promising biomarkers for mapping the spatial extent of epileptic brain, but reliably differentiating normal from pathological brain activity remains a fundamental challenge [14,15]. This is particularly true for HFOs, which are associated with normal activity like perceptual binding [14], memory encoding, and consolidation [16], sensory coding [17,18]\ and motor movements [9,19], but are sometimes pathological oscillations generated by epileptogenic brain tissue [3,20]. In the following sections, we discuss HFOs, microseizures, focal slow activity, and synchrony and the evidence that they are signatures, i.e. potential biomarkers of epileptogenic brain tissue.

Figure 12.1 Hybrid micro- and macro strip (A), depth (B), and grid electrodes (C). Microwire electrodes are embedded into the silastic substrate. Typical dimensions are 40µm diameter microelectrodes separated by 500 to 1000 µm. The hybrid electrodes increase the spatial coverage without increasing the number of clinical electrodes. When the embedded microwires are cut flush with the silastic substrate they should not increase the risk of irritating or injuring the underlying cortical tissue.

Interictal and ictal high-frequency oscillations (HFO): In addition to their role in normal brain function, high-frequency oscillations in the gamma, ripple, and fast ripple (FR: 250–1000 Hz) frequency ranges, are also increased in epileptogenic brain [3,14,20].

Analogous to the epileptic rats, both ripple and FR oscillations were first identified in humans from recordings in epileptogenic hippocampus [6]. Ripple and FR oscillations were found to be increased in slow-wave sleep compared to the waking period. In these early studies FR were identified in epileptogenic hippocampus and ripples were decreased in the epileptogenic hippocampus. These early animal studies were performed with microwire electrodes in chemotoxin-induced epileptic rats [21], and the human studies were primarily from patients with mesial temporal sclerosis [6]. This may explain why subsequent studies have found that ripple HFOs are also increased, similar to fast ripples in the SOZ [22,23]. Discrepancies could also originate from the fact that many more recent human studies have employed macroelectrode recordings rather than the microelectrodes used in the rat and early human studies. Multiple studies have subsequently clearly demonstrated that ripples and FR are reliably recorded using clinical macroelectrodes [23]. The fact that ripple HFOs are increased in the SOZ, rather than decreased, is consistent with reports from animals describing an increase in ripples prior to seizures [24], and supports the hypothesis that ripple-frequency HFOs are involved in the generation of seizures [25] and may serve as an accurate biomarker of epileptogenic brain tissue.

High-frequency oscillations at the onset of human seizures were initially described in icEEG recordings of patients undergoing evaluation for epilepsy surgery [26,27]. These early observations showed that focal seizures often begin with low-amplitude, high-frequency oscillations. The range of frequencies that have been reported vary, but are in the gamma, ripple, and fast ripple frequency range: 30–500 Hz [28], 40–120 Hz [27], 60–100 Hz [29], 70–90 Hz [30], 80–110 Hz [26], and 100–500 Hz [31]. Focal low-voltage fast oscillations (> 20 Hz–100 Hz) at seizure onset have been demonstrated to be associated with good epilepsy surgery outcome if the SOZ was completely resected [32,33]. Thus, there is now significant evidence that high-frequency oscillations are a functional signature of the epileptogenic brain tissue [3], and that they play a role in seizure generation [25]. Whether there are clinical advantages to recording icEEG at submillimeter spatial scale with microelectrodes in order to more efficiently sample HFO is not known.

Figure 12.2 Kurskal–Wallis applied to HFO (ripple/fast ripple), electrode type (micro/wiremacro), and brain regions of seizure onset zone (SOZ) and nonseizure onset zone (NSOZ). Box-plots and the results from posthoc analysis (***p < 0.002, **p < 0.01, *p < 0.05). The number of microwire ripple (R_m) and fast ripple (FR_m) HFO are increased in the SOZ compared to NSOZ. The number of macroelectrode ripple (R_M) and fast ripple (FR_M) HFO were increased in the SOZ compared to NSOZ, but less significantly for fast ripple.

(Adapted from Worrell et al. 2008 [23].)

Association of HFO with surgery outcomes: In a recent study investigating the spatial relation between HFOs and interictal epileptiform spiking in a group of lesional temporal and extratemporal patients, HFOs had a tighter correlation with the SOZ [34] and ripples and fast ripples HFOs were closely linked to the areas involved in seizure generation [35]. This study also showed that in focal cortical dysplasia, HFOs occurred in lesional areas that were not part of the SOZ, which might indicate that the potential epileptogenicity of these lesions is more widespread than the lesion visible on MRI.

In a subsequent study, Jacobs et al. correlated the resection of HFO-generating tissue with epilepsy surgery outcome and demonstrated that resection of tissue-generating ripple and FR HFOs was associated with a favorable outcome [22]. In a recent study by Haegelen et al. [36], removing HFO-generating tissue led to an improved surgical outcome in TLE group but not in a group of patients with ETLE. Further, Wu et al. [37] used intraoperative recordings to identify fast-ripple HFO (> 250 Hz) and found that all patients who had tissue with the HFO resected were seizure-free. In contrast, none of the patients (5/5) who did not have all fast ripple HFO tissue resected became seizure-free, including the one patient with negative MRI.

Scalp EEG recording of high-frequency oscillations

Studies using combined scalp and icEEG have previously reported that an epileptiform sharp wave can be detected on surface EEG if at least approximately 7 cm² of cortex is involved. Given that HFOs are often spatially limited, this suggests a significant challenge for the use of scalp EEG to detect HFOs. However, HFOs can be more widely distributed than initially thought [38,39]. A recent study using wide bandwidth EEG recordings from ten children with continuous spikes and waves during slow-wave sleep described HFOs on scalp EEG concurrent with interictal spikes, with peak HFO power that ranged from 97.7 to 140.6 Hz [40]. These data showed brief HFO events visible in the raw scalp EEG data associated with the EEG large-amplitude spike [40]. Similarly, in patients with focal epilepsy, gamma- and ripple-band activity has been described on scalp EEG, most often around the time of interictal spikes, but better than interictal in correlation with the SOZ [41].

It is important to note that noncerebral activity (artifact) can be especially difficult to separate from cerebral generators when analyzing band pass filtered data, as is often done when analyzing HFOs. For example, induced gamma-band activity on scalp EEG during presentation of a visual stimulus was shown to be artifact associated with microsaccade eye movements [42]. Saccades are accompanied by a spike potential of muscle origin, which have a broad high-frequency spectrum that could be misinterpreted as having a cerebral origin. Thus, some of the induced gamma-band power recorded from scalp EEG that was thought to be of cerebral origin is likely of muscular origin (eye movement related) [42].

Interictal and ictal DC shifts and slow-frequency oscillations: Although recognized in early animal cortical recordings [43], DC fluctuations and ultraslow frequency oscillations have received relatively little attention in human brain recordings [7,11]. The sustained DC and very slowly changing potentials (< 0.1 Hz) that characterize these activities likely have a cortical origin given that they can be recorded from small regions of neocortex that are devoid of inputs (i.e., neocortical slabs). Both neuronal and nonneuronal mechanisms are likely involved in generating these activities [44,45]. Ultraslow and slow oscillations synchronize higher-frequency oscillations and modulate cortical excitability, which may explain the increase in epileptiform discharges seen in slow-wave sleep [11]. Recent studies have demonstrated that ultraslow oscillations occurring during sleep can be focal [46].

Figure 12.3 Wide bandwidth recording from 6×6 subdural grid. Right icEEG recording from seizure onset region. At seizure onset there is a prominient slow wave with increased amplitude of activity in the high-frequency range (bottom).

(Courtesy of Matt Stead, MD, PhD.)

The majority of human studies used AC coupled amplifiers with sufficiently long time constants that record filtered DC shifts as slow-wave potentials. Focal ictal slow-wave potentials were identified in 85% of 89 seizure onset in six patients with neocortical epilepsy [12]. Bragin and colleagues reported 89% of hippocampal seizures with a high-frequency onset were associated with a coincident slow-wave potential [47]. There is evidence for the localizing value of interictal and ictal DC shifts, but the ultimate clinical utility remains to be determined [7].

Interictal microseizures: The ability to investigate in vivo human epileptogenic brain across a wide range of spatial and temporal scales yields remarkably rich data. One of the most dramatic findings is microdomain seizure-like events, apparent on microwires but not detected on adjacent clinical macroelectrodes [10,48]. These seizure-like events, or microseizures, are clinically silent electrographic events, detectable on microwires only. Microseizures have spectral features, morphology, and durations similar to electrographic seizures detected on macroelectrodes. Typical durations vary between 10–60 seconds (median of ~30 seconds), but can be over 100 seconds. Microseizures are subclinical, self-limiting, electrographical events that mostly remain isolated to a single microwire, but in some cases evolve directly into macroelectrode seizures.

Conclusions: In summary, the search for interictal biomarkers of epileptogenic brain remains an active area of research [2,3]. The existence of an interictal biomarker of the EZ would be a transformative technology for epilepsy surgery, since it would eliminate the reliance on chronic icEEG recording, and it would also help to delineate EZ in patients with negative MRI. To date there is substantial evidence to support the clinical usefulness of pathological interictal HFOs for localization of the EZ, and HFOs appear to be the best interictal biomarker of the EZ. The results are very encouraging; however, to date they are primarily from small retrospective studies in lesional or MRI-negative epilepsy, and few have directly compared IES, HFO, and other potential interictal biomarkers to localization provided by spontaneous seizures. In the future a prospective trial seeking class I evidence for the clinical utility of wide bandwidth interictal electrophysiology in lesional or MRI-negative focal epilepsy surgery should be possible.