High-resolution Electroencephalography Provides New Insights into Epilepsy

Perturbations of the electrical oscillations of the cerebral cortex of epileptics were first demonstrated by Berger,¹ soon after he discovered the existence of recordable brain electrical activity in humans in July 1924. Although interested in electroencephalography (EEG) not as a medical tool but rather as a physical measure of mental processes, Berger’s writings provided the initial descriptions of many of the patterns important to the current practice of clinical EEG: the alpha rhythm and alpha blocking with eye opening, sleep patterns, disorganization of the EEG following anoxia, three per second discharges with absence seizures, and post-ictal slowing.
Advances in medicine are often driven by technological developments. The Eithoven string galvanometer allowed Berger to discover the human EEG. Vacuum tube amplification permitted clinical EEG to flourish, remaining the principal brain imaging technique until the 1970s, when it ceded that role to computerized X-ray tomography. Still, EEG remains the principal clinical tool for the diagnosis of epilepsy. During the 1980s, improved amplifiers and video technology, plus advances in digital data acquisition and storage, made it possible for video EEG monitoring and intracranial recording to become standard fare for the evaluation and treatment of certain patients with epilepsy.
Currently, microelectrodes suitable for human use, along with higher-capacity data acquisition and storage systems, are enabling technologies for refining, by orders of magnitude, the temporal and spatial resolution of brain electrical recordings in patients with epilepsy. Early results suggest that data obtained from microelectrodes and microelectrode arrays implanted in the seizure foci of epilepsy patients are likely to inform, in very significant ways, our understanding of the patho-physiological disturbance present in the epileptogenic cortex, as well as the process of seizure generation. These results hold the promise of leading to novel treatments for medically refractory epilepsy, including improvements in surgical brain resection and implantable devices designed to warn patients of impending seizures, or even to stop them from becoming clinically manifest by targeted delivery of drugs or electrical stimulation to affected brain areas.^2,3
Recorded through the intact skull, the EEG reflects summated synchronous fluctuations of membrane potentials, primarily of the apical dendrites of millions of neurons spread over several square centimeters.⁴ Brain signals passing through the scalp and skull are blurred and attenuated, especially in the higher frequencies.⁵ A large area of the brain’s surface must be active at any one time to generate EEG signals that are recordable at the scalp. Indeed, brain signals involving less than about 6cm² of cortex may be invisible on scalp EEG.⁶ Furthermore, EEG signals from brain areas such as the inferior frontal lobe are difficult to record from surface electrodes; scalp-recorded EEG may detect only propagated activity from distant sources.⁷ Nonetheless, for routine diagnostic purposes scalp recordings are the most practical option and are generally sufficient.
However, for epilepsy patients requiring surgical treatment the data provided by scalp recordings often prove insufficient to adequately localize the part of the brain responsible for generating seizures.⁸ In these cases, the electrical effects of the skull are eliminated by surgically implanting electrodes into the subdural space or into the substance of the brain. Intracranial electrode recordings help to more accurately target tissue for surgical removal by defining the locations of interictal and seizure activity at a resolution of 0.5–1cm—the typical interelectrode spacing used in clinical subdural and depth arrays. Intracranial EEG also extends the scope of brain signals that can be monitored. Scalp EEG recording is best suited to detecting oscillations up to about 40Hz, a consequence of volume conduction and the electrical properties of the coverings of the brain.⁹ With intracranial EEG, higher frequencies in the gamma range (25–~150Hz) can be discerned. These signals often provide important clues as to the location of the epileptogenic zone, and are often prominent at seizure onset.^10–14
Using microwire electrodes implanted into the mesial temporal structures of patients undergoing surgery for temporal lobe epilepsy, a team at the University of California, Los Angeles (UCLA) discovered ‘fast ripples’— brief and highly spatially restricted oscillations in the 250–600Hz range.¹⁵ Fast ripples have been found to be most prominent in the location of seizure onset in human mesial temporal lobe epilepsy,^15–18 as well as in animal models of epilepsy.^19–21 Similar oscillations have been detected in both mesial temporal structures and the neocortex^14,18,22 in recordings of high temporal resolution from standard subdural electrodes. As signals in this frequency range can represent a mixture of membrane potentials and action potentials, there is an ongoing debate about the physiology underlying fast ripples. Fast ripples correlate strongly with bursts of action potentials from very small populations of neurons and may be seen in conjunction with epileptic sharp wave discharges (see Figure 1).^23,24
It is unclear whether these oscillations are the result of fast-spiking, tightly interconnected interneurons, augmented by currents cycling through gap junctions,²⁵ or clusters of neurons firing out of phase.²⁴ Despite their strong spatial correlation with the epileptogenic zone, the role that fast ripples play in epilepsy is not known. Trevelyan and colleagues have proposed that they may be related to an inhibitory feedback mechanism, serving to dampen the spread of ictal discharges and constituting a natural defense against seizures.²⁶ Using a 4mm-square penetrating array of 96 1mm-long microelectrodes with recording tips 3–5 microns in diameter implanted into the epileptogenic cortex, we recorded electrical patterns resembling seizures restricted to a small number or even a single microelectrode (see Figure 2).²⁷ These tiny discharges are lost in recordings from standard subdural electrodes, as these electrodes effectively average the electrical signals from the area of cortex that they cover. Since the array’s interelectrode spacing of 400 microns corresponds roughly to the scale of cortical macrocolumns— architectural elements ranging from 300 to 500 microns in diameter²⁸— it is possible that these ‘microseizures’ are limited to one or several macrocolumns. Similar discharges have also been observed by researchers at Mayo Clinic using 40-micron-diameter microwire surface electrodes with 1mm spacing embedded in subdural grids.²⁹
The significance of microseizures and their potential role in the generation of clinical seizures are matters of speculation. Although they are most frequently seen within the epileptogenic zone, the majority of microseizures that we have observed have had not a clear relationship to larger, clinically evident seizures. Most microseizures appeared to start and stop on their own. In a few cases in both our series and the Mayo Clinic study, however, microseizures were observed to be involved in seizure initiation and evolution. Figure 3 shows the recruitment of cortex, sampled by the microelectrode array, into an ongoing seizure beginning with a series of propagated ictal discharges that appear to incite local microseizure activity at a small number of microelectrode sites. This secondary microseizure rapidly spreads to adjacent electrodes, then to the entire array, and is later picked up by the nearby subdural intracranial EEG.
Any formulation of the respective roles of fast ripples and microseizures in the process of seizure development is strictly speculative. While it appears that they may both be signatures of epileptogenic cortex, sharing the common property of having tiny, isolated generating areas, they appear to represent distinct phenomena. We only rarely observed fast ripples and microseizures to be temporally coincident. Furthermore, they are generally found at different locations within the cortex sampled by the microelectode array, sometimes adjacent to one another and sometimes completely separate (see Figure 4). Still, we find it tempting to speculate that perhaps clinical seizures occur when microseizures, beginning in tiny cortical domains surrounded by non-epileptogenic tissue but possibly functionally interconnected, spread to involve adjacent regions; fast ripples reflect a defensive system that acts to contain their spread. ■