Invasive seizure monitoring in the critically-Ill brain injury patient: Current practices and a review of the literature

Abstract

Seizures commonly occur in a variety of serious neurological illnesses, and lead to additional morbidity and worsened outcomes. Recently, it has become clear that not all seizures in the acute brain injury setting are evident on scalp EEG. To address this, we have developed a protocol for depth electrode placement in the neuro-intensive care unit for patients in whom the clinical suspicion of occult seizures is high. In the current manuscript, we review the literature on depth EEG monitoring for ictal events in critically-ill, unconscious patients, focusing on the incidence of seizures not detected with scalp EEG in various conditions. We critically discuss evidence in support of and against treating these events that are only detectable on depth recordings. We describe additional specific scenarios in which depth EEG recordings may be helpful, including for the detection of delayed cerebral ischemia following subarachnoid hemorrhage. We then describe current techniques for bedside electrode placement. Finally, we outline potential avenues for future investigations, including the use of depth electrodes to describe circuit abnormalities in acute brain injury.

1. Introduction

1.1. Seizures in acute brain injury

Seizures are common sequelae of a variety of neurological injuries, including trauma [1], infection [2], neoplasia [3], as well as intracerebral [4], subdural [5], and subarachnoid hemorrhage (SAH) [6]. It is now appreciated that many seizures in acutely ill, comatose patients are not clinically evident; nonconvulsive seizures have been observed in approximately 8% of all subarachnoid hemorrhage patients [7] and approximately 10% of comatose traumatic brain injury patients [8]. Because of this, continuous electroencephalographic monitoring (cEEG) has been proposed as the standard of care for unresponsive patients after acute brain injury (ABI) [9,10].

In ABI there is accumulating evidence supporting additional harm from uncontrolled seizure activity, including nonconvulsive status epilepticus [11], hippocampal atrophy ipsilateral to seizure onset [11], and nonconvulsive seizure burden following ABI, such as SAH, has been associated with worse functional and cognitive outcomes [12]. Seizures after TBI were associated with increases in intracranial pressure and metabolic crisis [8], and status epilepticus strongly correlated with mortality in the same report. Epilepsy (i.e., unprovoked seizures) commonly occurs after both TBI [13] and subarachnoid hemorrhage [14], although the contribution of early seizures to risk of epilepsy remains an open question.

1.2. Historical use of scalp and depth EEG for neuromonitoring

Seizure activity without obvious clinical manifestations, such as non-convulsive seizures, cannot be detected by direct observation, and thus necessitates the use of other neuromonitoring techniques. As early as the 1930s, EEG monitoring was used in epilepsy [15] and being evaluated as a method for detecting intracranial lesions [16]. Since that time, techniques with greater capacity for seizure detection than discrete monitoring with scalp EEG have emerged. In 1991, Hilz and colleagues described the first case reports of using continuous EEG (cEEG) to monitor patients in the neurological intensive care unit [17] (NICU), and subsequent work demonstrated cEEG could be used to detect non-convulsive seizures [18,19]and to prognosticate [20,21].Similarly, depth EEG, pioneered in 1961, offered another useful technique for seizure detection and significantly improved enumeration, lateralization, and localization of seizure foci compared to the use of clinical features and scalp EEG [22]. Since that time, depth EEG has been used increasingly for localization of seizure foci in refractory epilepsy [23].

The surgical epilepsy group at Yale advanced depth EEG technology by combining it with microdiaylsis [24]. Although their early work focused on adenosine and glutamate [25], a subsequent study characterized increases in local extracellular lactate which accompanied temporal lobe seizures [26]. Lactate concentrations increased on average by 90% following seizures, and persisted for 60–90 min after seizure onset. This suggested that seizures caused a period of regional metabolic distress later termned metabolic crisis, a condition that neurologists were beginning to appreciate as a contributor to poor outcomes in ABI patients [27]. This combination of microdialysis and depth EEG thus had both theoretical and practical appeal as a technique for monitoring in the NICU.

2. Discussion

2.1. Depth EEG placement in NICU patients

Depth EEG electrodes are typically placed in stuporous or comatose patients as part of a so-called “bundle” of other invasive monitoring modalities. At our institution, five modalities are typically placed via two cranial bolts. The larger bolt is an Integra three-port bolt (IM3.ST: Plainsboro, NJ) which has fittings for a Camino™ intracranial pressure monitor, a Licox™ brain-tissue oxygenation monitor, and a microdialysis catheter. The depth electrode is placed through a Codman (Synthes-Depuy Codman Neuro: Raynham, MA). dual-port bolt (the other port is used for Hemedex Bowman Perfusion Monitor™ [Hemedex Inc.: Cambridge, MA] blood flow monitoring). Of note, many similar bolt options are available from several companies; some Hemedex bolts, in particular, have the advantage of MRI compatibility and some offer up to four ports. General techniques for multimodality monitoring have been recently reviewed [28]. The focus of the current manuscript will be on depth electrode placement and recordings, while other monitoring techniques including subdural strip electrodes will not be discussed.

Prior to placement, the patient's family is asked about medications that interfere with hemostasis, as well as a history of easy bleeding or bruising, and routine blood tests are ordered for platelet count and coagulation parameters. The imaging is then reviewed by the primary team and the neurosurgeon, and an appropriate place for electrode placement is identified. Typically this is near Kocher's point, but may be somewhat more rostral, if craniotomy or ventriculostomy incisions need to be avoided. Pathology typically dictates the side of the procedure; all things being equal, we prefer right-sided procedures.

Reference and ground electrodes are placed on the patient. The operative site is marked and shaved, and the patient is prepped and draped in the usual sterile fashion. Cranial Access Kit such as that by Codman™ are available including circular drapes that adhere to the patient's skin. For the depth electrode, skin is incised, and a small self-retaining retractor is placed on the skull. The coronal suture may be identified and used as a landmark, if desired. The drill bit accompanying the Codman™ bolt kit is fitted to the drill in the Cranial Access Kit. At this point, the surgeon may place the electrode freehand, or if desired under electrophysiological guidance. If he or she plans to place it freehand, the electrode is placed inside the Codman™ dual-lumen bolt assembly and measured to a depth of 2.5 or 3 cm from the tip of the bolt fitting, depending on preoperative imaging (Fig. 1). We use a 2-0 silk tie to mark the appropriate depth. If placement under electrophysiological guidance is desired, this step is not needed. Using a freehand technique, the skull is drilled. An 18-gauge needle is used to create a circumferential durotomy. The bolt fitting is placed, and the electrode is inserted to depth. If using electrophysiological guidance, the end of the electrode is attached to a Cabrio connector (Ad-Tech Medical™: Racine, WI). The electrode is then passed through the bolt, and potentials for all eight channels are monitored. The electrode is optimally placed when signal is detected on all channels. The bolt fittings are tightened, and a strain relief loop is created using a Tegaderm bandage (3M™: Two Harbors, MN) to prevent inadvertent removal. The wound is dressed with Xeroform gauze (Covidien™: Jersey City, NJ).

Fig. 1.

Typical bedside placement of a depth electrode. Panel A: Patient with diffuse SAH has an EVD, a brain tissue oxygenation monitor, a depth electrode (red arrow), and several other probes placed in the right frontal lobe. Panel B: Coronal image reveals an eight-contact depth electrode in the frontal cortex.

Optimal placement of the depth has not been determined in the present published literature. In general, we have attempted to place the electrode in perilesional tissue, with a strong bias towards the area around Kocher's point, but it is not clear that this is the best location. Moreover, freehand techniques we utilize are not optimized for precise placement; notably, frameless stereotactic techniques have been used for depth electrode placement in the past [29] and may be useful in this setting as well.

2.2. Safety

Bedside depth electrode placement appears to have a safety profile similar to that of other intracranial monitoring probes. One series of sixty-one patients undergoing multimodality monitoring reported malfunction or dislodgement as the most common complication (43%), while infection and hemorrhage were considerably less frequent (5% and 3%, respectively) [30]. The three patients with infections each had a CSF culture that grew a different organism (Candida albicans, Escheria coli, and Acinetobacter), and all three had their intracranial hardware removed. The most serious complication was a large intracerebral hemorrhage in one patient; however, the patient was considered at very high risk as he was taking aspirin and clopidogrel at admission, and the treatment plan was discussed extensively with the patient's family who desired an aggressive treatment paradigm. The overall hemorrhage rate is notably in line with hemorrhage rates reported for Camino intracranial pressure monitors (2–4%). However, larger, multicenter studies will be needed to confirm these findings.

2.3. Findings on depth versus surface EEG

There is evidence that depth EEG has superior sensitivity to cEEG for non-convulsive seizures. A study examining fourteen patients with brain injury who were subsequently implanted with an eight-contact depth electrode demonstrated that depth EEG detected seizures in eight patients who demonstrated no ictal activity on cEEG, and detected seizures in two additional patients whose seizures showed only intermittent corresponding ictal activity on surface electrodes [31]. In a subsequent study of 48 comatose subarachnoid hemorrhage patients, researchers found that eighteen patients (38%) showed seizure activity on intracortical EEG, while only four (8%) showed seizure activity on continuous scalp EEG [32] (Fig. 2, left panel). Recently, this finding was reproduced in traumatc brain injury patients; in a cohort of 34 patients, 21 had seizures or periodic discharges, and nine of these had ictal phenomena observable only on the depth electrode [33]. Depth EEG recordings also have the advantage (over scalp electrodes) of high signal-to-noise ratio and sensitivity to gamma band oscillations [34].

Fig. 2.

Depth electrodes detect seizures invisible to scalp electrodes. Top left panel: Compressed spectral array (CSA) of EEG data recorded from electrodes ipsilateral to the depth (top row), contralateral electrodes (middle row), and the depth electrode (bottom row). CSA reveals a depth-only seizure associated with an increase in broadband amplitude in scalp contacts, without clear ictal activity. Bottom left panel: There is evidence of hyperemia with an increased pbtO2 and CPP,as well as a spike in ICP. Top right panel: CSA of EEG data reflecting a seizure detectable on both scalp and depth contacts. Bottom right panel: pbtO2 collapses, and there is a spike in ICP, brain temperature, and cerebral blood flow.

In acute brain injury, nonconvulsive seizures may result in metabolic distress and intracranial pressure changes which may be detectable using invasive neuromonitoring [32,35–39]. During a seizure, the brain metabolism may increase briefly [32], brain tissue partial pressure of oxygen in the brain's extracellular fluid drops [32,35], and the intracranial pressure increases [36,39,38] (Fig. 2, bottom right panel). The brain endures a “metabolic crisis,” characterized by an elevated lactate/pyruvate ratio and decreased extracellular glucose; findings thought to indicate increased glucose consumption, reduced oxidative metabolism, and an impaired redox state [37].

Emerging evidence suggests seizures isolated to the depth electrode may be associated with a variety of deleterious findings on multimodality monitoring [31–33]. In subarachnoid hemorrhage, depth EEG seizures are associated with increases in heart rate and blood pressure, and there was a trend towards increased ICP and CPP (Fig.2,bottom left panel) [32]. Some evidence suggests that in patients acute brain injury regional cerebral blood flow increases only minutes after the onset of seizure possibly due to the fact that vasoreactivity is impaired following acute brain injury [32]. In trauma, depth seizures were associated with local metabolic derangement, as indicated by increases in the lactate-pyruvate ratio and decreases in glucose [33]. These associations do not prove unequivocally that isolated findings on depth EEG themselves indicate a deleterious process; however, depth only seizures are more strongly associated with poor outcome than surface seizures. The current does suggest that additional study is warranted.

2.4. Developing applications

As described above, depth EEG is used for seizure detection in tandem with other neuromonitoring modalities, including scalp EEG. Much of neurocritical care is focused on detection and minimizing the effects of secondary complications of ABI. Depth EEG may have a role for early detection of secondary neurological complications commonly seen in NICU patients, although the available data is sparse. In particular, its heightened signal-to-noise ratio may be an advantage in the detection of ischemia. In one series, researchers reported two patients who experienced secondary neurological complications (one patient with SAH complicated by widespread cerebral infarct in the setting of sepsis, and one patient with a large ischemic infarct complicated by hemorrhagic conversion), both of whom showed isolated changes on depth EEG between two and six hours before detection on other invasive neuromonitoring devices, and eight hours prior to clinical detection by physical exam [31]. Similarly, in a report of five patients with poor-grade subarachnoid hemorrhage who underwent intracortical EEG placement, researchers reported that alpha-delta ratio calculated every20 s was able to correctly predict the presence or absence of vasospasm in all five patients [40]. This may offer the added opportunity of automated and reliable detection of ischemia from vasospasm in patients with subarachnoid hemorrhage. Thus, while further studies with larger cohorts are necessary, the evidence for depth EEG as a tool for early detection of secondary neurological complications is encouraging.

The interpretation of physiological signals recorded from invasive brain monitoring is challenging at times [41]. One opportunity to more comprehensively gain insights into underlying brain physiology and evolving pathophysiology is a more comprehensively assessment of brain physiology. Intracortical EEG recordings may offer a major step in this direction. For example a drop in partial brain oxygenation may suggest both ischemia or a seizure. The EEG recording may show depressed or discontinuous background activity in the former and seizure activity inn the later. In other words intracortical EEG recordings may further support interpretation of other invasive brain monitoring signals.

3. Future outlook

Bedside depth electrode placement appears to be safe and effective for the detection of occult seizures in unconscious patients in the NICU. It is technically straightforward, and requires no specialized equipment, beyond the electrode itself, EEG monitoring equipment, and some basic neurosurgical supplies. However, the technique itself is ripe for improvement, which includes both the technical aspects of electrode placement as well as the analysis of the data. Targeted placement of the electrode using stereotactic neuronavigation may allow for recordings of specific brain regions. Depth electrodes have been used in patients undergoing evaluation for epilepsy surgery to describe brain networks required in a number of cognitive functions, including memory [42], reward-based decision making [43], and consciousness [44]. In particular, high gamma band EEG activity (>70 Hz) is thought to reflect local action potentials [45]; this suggests that depth electrodes in the ICU could potentially be used to study frontal brain regions compromised in unconscious patient with acute brain injury.

4. Conclusion

Depth electrodes allow detection of seizures not visible on scalp recordings. Further research is needed to confirm the clinical relevance of depth-only seizures. Additionally, isolated cases suggest that depth electrodes may aid in the detection of evolving ischemia. In the near term, we depth EEG has the potential to be integrated as a tool in the neurointensivist's armamentarium.