Correlates of Spreading Depolarization, Spreading Depression, and Negative Ultraslow Potential in Epidural Versus Subdural Electrocorticography

Abstract

Spreading depolarizations (SDs) are characterized by near-complete breakdown of the transmembrane ion gradients, neuronal oedema and activity loss (=depression). The SD extreme in ischemic tissue, termed 'terminal SD,' shows prolonged depolarization, in addition to a slow baseline variation called 'negative ultraslow potential' (NUP). The NUP is the largest bioelectrical signal ever recorded from the human brain and is thought to reflect the progressive recruitment of neurons into death in the wake of SD. However, it is unclear whether the NUP is a field potential or results from contaminating sensitivities of platinum electrodes. In contrast to Ag/AgCl-based electrodes in animals, platinum/iridium electrodes are the gold standard for intracranial direct current (DC) recordings in humans. Here, we investigated the full continuum including short-lasting SDs under normoxia, long-lasting SDs under systemic hypoxia, and terminal SD under severe global ischemia using platinum/iridium electrodes in rats to better understand their recording characteristics. Sensitivities for detecting SDs or NUPs were 100% for both electrode types. Nonetheless, the platinum/iridium-recorded NUP was 10 times smaller in rats than humans. The SD continuum was then further investigated by comparing subdural platinum/iridium and epidural titanium peg electrodes in patients. In seven patients with either aneurysmal subarachnoid hemorrhage or malignant hemispheric stroke, two epidural peg electrodes were placed 10 mm from a subdural strip. We found that 31/67 SDs (46%) on the subdural strip were also detected epidurally. SDs that had longer negative DC shifts and spread more widely across the subdural strip were more likely to be observed in epidural recordings. One patient displayed an SD-initiated NUP while undergoing brain death despite continued circulatory function. The NUP's amplitude was -150 mV subdurally and -67 mV epidurally. This suggests that the human NUP is a bioelectrical field potential rather than an artifact of electrode sensitivity to other factors, since the dura separates the epidural from the subdural compartment and the epidural microenvironment was unlikely changed, given that ventilation, arterial pressure and peripheral oxygen saturation remained constant during the NUP. Our data provide further evidence for the clinical value of invasive electrocorticographic monitoring, highlighting important possibilities as well as limitations of less invasive recording techniques.

Keywords: cerebral ischemia; hypoxia; malignant hemispheric stroke; spreading depression; subarachnoid hemorrhage.

FIGURE 1

Comparison of normoxic SD with hypoxic SD preceded by non-spreading depression of spontaneous activity. (A) Experimental setup in rats. LDF, laser-doppler flow probe; [K⁺]_o and DC, K⁺-selective glass-microelectrode; Pt/Ir electrodes, platinum/iridium plate electrodes; KCl droplet, SD was triggered with a droplet containing KCl at 2M. Panel (B) shows the first half of an experiment with 4 KCl-induced SDs followed by the first spontaneous SD induced by hypoxia, and panel (C) shows the boxed portion of (B) with an expanded time scale. KCl application is marked with red dots.Traces 1–3: The full-band ECoG signal (DC/AC-ECoG) contains information on both the negative DC shift that identifies SD (Dreier, 2011) and the SD-induced depression of activity (negative is up for ECoG recordings shown in all figures). Note the differences in amplitude and shape of the DC shift between the gold standard recording technique in animals, the intracortical glass-microelectrode, and the Pt/Ir electrodes on the brain surface, gold standard in humans (cf. Table 1). For example, the amplitude ratio of the initial DC negativity divided by the subsequent DC positivity was significantly larger for the glass-microelectrode than for the Pt/Ir electrodes in normal tissue [Ag/AgCl: 9.1 (IQR = 8.0, 17.0); caudal Pt/Ir: 1.2 (IQR = 1.0, 1.7); rostral Pt/Ir: 1.0 (IQR = 0.7, 1.3), P ≤ 0.05, n = 7, Kruskal–Wallis One-Way Analysis of Variance on Ranks with post hoc Dunn test]. Also note the spread of the SD in rostral (Pt/Ir-electrode 2)-caudal (Pt/Ir-electrode 1) direction. Trace 4: The hallmark of SD is the abrupt, near-complete breakdown of the transcellular ion concentration gradients. Here, the large SD-associated increase of [K⁺]_o is seen. Note the significant prolongation under hypoxia. Traces 5–7: The spontaneous activity is assessed in the higher frequency band between 0.5 and 45 Hz. Note that the SD under hypoxia is preceded by non-spreading depression of the spontaneous activity [blue and green asterisks in (B,C)]. Non-spreading depression precludes that the subsequent SD can induce spreading depression of the activity. Interestingly, the depression of activity was reversible after the SD despite continued hypoxia. Trace 8: MAP fell under hypoxia. Trace 9: SD-induced hyperemia under normoxia and hypoxia, respectively. Note the marked reactive hyperemia following the recovery from SD under hypoxia. Note also that the Pt/Ir electrodes in contrast to the glass microelectrode displayed a marked negative DC reverberation following the recovery from SD under hypoxia (red asterisks), which may hinder correct assessment of the duration of the negative DC shift of SD. The fourth SD under normoxia showed much smaller amplitude than the three previous SDs at both the Pt/Ir-electrode 2 and the Ag/AgCl-based intracortical microelectrode. Such amplitude reductions are sometimes observed in clusters of SDs and could, for example, result from a proportion of parenchymal cells that are still refractory after the previous SD.

FIGURE 2

Spontaneous SD during hypoxia compared with terminal SD in death. Traces are the same as in Figure 1. The Pt/Ir plate electrodes were compared with the K⁺-selective glass microelectrode over the full SD continuum from SD in well-nourished tissue (0–80 min shown in Figure 1) to SD in hypoxic tissue to SD during global cerebral ischemia (terminal SD) (81–150 min shown in Figure 1 and 150–300 min in Figure 2). Systemic hypoxia was induced by partial replacement of O₂ with N₂ in the inspired air. The negative DC shift is necessary and sufficient for identification of SD, and the duration of the negativity is a measure of the metabolic and excitotoxic burden imposed on tissue by SD (Dreier et al., 2017). Accordingly, it can be seen that both negative DC shift and increase of [K⁺]_o were markedly longer under hypoxia than under normoxia, and they were terminal after CA. Note that the Pt/Ir electrodes in contrast to the glass microelectrode displayed a marked negative DC reverberation following the recovery from SD under hypoxia (red asterisks), which may hinder correct assessment of the duration of the negative DC shift of SD. It is also interesting in this animal that terminal SD began spontaneously under hypoxia prior to the spontaneous CA and induced spreading depression of activity. Note the spread of the terminal SD in rostral (Pt/Ir-electrode 2)-caudal (Pt/Ir-electrode 1) direction. Panel (B) shows the same data from (A) on an expanded time scale (275–290 min). The terminal SD-induced spreading depression of the activity is shown in (B) using the integral of power of the spontaneous activity.

FIGURE 3

Terminal SD during death in the rat brain. Traces are similar to Figure 1, 2. A drop of KCl triggered the SD under hypoxia in this animal. Note that the hypoxic SD induced spreading depression of activity (cf. x) in contrast to the hypoxic SD in Figure 1 which was preceded by non-spreading depression. Negative DC shift and [K⁺]_o were only mildly prolonged indicating that hypoxia was less severe during this experiment compared to the one displayed in Figure 1, 2. Accordingly, SD induced a relatively normal spreading hyperemia (SH). Subsequently, the fraction of inspired oxygen (FiO₂) was further reduced, and a spontaneous, slowly progressive respiratory arrest (cf. end-expiratory pCO₂) eventually led to non-spreading depression of spontaneous activity (cf. asterisks). Thirty-five seconds after isoelectricity was established a spontaneous terminal SD occurred first at the rostral Pt/Ir electrode (rostral-caudal delay: 63 s) and triggered a marked spreading ischemia (SI) with rapid drop in rCBF from 112 to 1%. The SI was detected by the LDF probe 36 s after the SD had first appeared at the rostral Pt/Ir electrode. Circulatory arrest (CA) with a rapid drop in MAP occurred only 247 s after the onset of SI.

FIGURE 4

Terminal SD in the wake of circulatory arrest. Sudden circulatory arrest was induced in this rat by injection of 10 ml of air into the heart via the femoral vein. MAP fell rapidly from 99 to 6 mmHg and respiratory fluctuations of expiratory pCO₂ ceased within 18 s; non-spreading depression caused isoelectricity within 26 s (asterisks) and terminal SD started after 98 s. Terminal SD showed an initial and a late negative DC component similar to human recordings (Dreier et al., 2018b). Whereas the initial one is the actual SD component, the late one is termed the NUP.

FIGURE 5

Epidural and subdural recordings of single SDs. (A) The 31 epidurally identified SDs involved more subdural electrodes than the 36 SDs that could not be identified using epidural recordings. (B) When the SDs were pooled this was statistically significant (Mann–Whitney Rank Sum Test, P ≤ 0.001). The whiskers (error bars) above and below the box indicate the 90th and 10th percentiles. The red reference line marks the mean. (C) SDs are observed as a negative DC shift propagating between neighboring electrodes (cf. top traces showing the DC/AC-ECoG (0–45 Hz) at the different electrodes, negative up). Red portions indicate potentials associated with SDs. SD-induced spreading depression is observed in the AC-ECoG recordings (0.5–45 Hz) as a rapidly developing reduction in the amplitudes of spontaneous activity that spreads together with SD between adjacent recording sites. The squared spontaneous activity is called AC-ECoG power (cf. Power AC-ECoG-traces from the different electrodes). In patient 1, 10 non-clustered SDs were recorded using subdural electrodes. Six SDs only occurred at electrode 6 and 4 also involved electrode 5. The left panel shows an example of one of these SDs. The red asterisk marks the SD-induced depression of activity at electrode 6. The light green arrow points at the SD-induced tissue hypoxia (Dreier et al., 2009; Bosche et al., 2010; Winkler et al., 2017). Different from the subdural recordings, the epidural ones did not allow clear identification of the SD, though there could be a correlate. By contrast, in patient 4 (middle panel), 14 of 16 SDs were not only seen subdurally but also epidurally. A non-clustered SD is displayed. The preceding SD occurred 12.5 h before and the subsequent one 25.1 h after this SD. Note that the SD involved all six subdural electrodes. At the epidural electrodes, not only the negative DC shift but also the SD-induced depression was visible. The red asterisks mark the SD-induced spreading depression of activity at subdural electrodes 2–5 and the two epidural electrodes. In patient 6 (right panel), the subdural DC/AC-ECoG of 3 SDs was available to the analysis. None of these showed an unambiguous correlate in the epidural DC/AC-ECoG. One of these is displayed here as an example.

FIGURE 6

Epidural and subdural recordings of clustered SDs. (A) Top six traces show the raw DC/AC-ECoG recordings from each subdural electrode (negative up). Red portions indicate potentials associated with SDs. SD-induced spreading depressions are assessed in the power of the AC-ECoG (traces 7–12). Below, the epidural recordings are shown. In the left panel, clustered SDs in patient 4 are displayed. Between an MRI on Day 2 and a CT on Day 9, the patient developed a new infarct in the vicinity of the recording strip. The SDs involved all subdural and epidural electrodes. The red asterisks mark the SD-induced spreading depressions. The third SD is an isoelectric one. In patient 7 (right panel), clustered SDs on Day 8 are shown. The electrodes were located close to a delayed cerebral infarct as shown in Figure 8. The SDs were observed both subdurally and epidurally in contrast to the SD-induced spreading depressions which were only seen at the subdural electrodes. (B) Amplitudes and durations significantly correlated between subdural and epidural negative DC shifts indicating SDs.

FIGURE 7

Progress to brain death in patient 7. (A) The six top traces show a series of several transient SDs followed by terminal SD. The terminal SD consists of the initial SD component followed by the NUP (Dreier et al., 2018b). Traces 7–12 display the SD-induced spreading depression that persisted through death (observed as a reduction in the amplitudes of the AC-ECoG power propagating from electrode to electrode). SDs and NUP are smaller in the epidural than the subdural DC/AC-ECoG recordings. However, in particular the NUP is clearly visible also in the epidural recordings (traces 13 and 14). Also the terminal SD-induced depression of activity is seen at the epidural recordings (traces 15 and 16). Traces 17 and 18 display MAP and arterial pressure fluctuations. Regions of Box 1 outlined by the dark gray lines and of Box 2 outlined by the light gray lines are shown in detail in Figure 8 which additionally gives the continuous ICP recordings. (B) The occurrence of SDs, SD-induced depression periods and isoelectric SDs in patient 7 over the whole clinical course. (C) When the patient underwent circulatory arrest 20 h after brain death the DC potential at all subdural and epidural electrodes showed a slow positive drift (positive down). The asterisks mark the moment when the arterial pressure monitoring was switched off.

FIGURE 8

Progress to brain death in patient 7. (A) A subset of the same traces in Figure 7 is shown in greater detail. In addition, ICP is given. As outlined by Box 1 in Figure 7, the initial time period around the terminal SD is displayed. The SDs are marked in red. The second SD is a terminal one that starts at electrode 4 from where it spreads to electrode 5. The terminal nature of this event is indicated by the persistence of the negative DC shift. This negative potential develops further through time as a NUP, which is thought to reflect the progressive development of tissue necrosis. The red arrows mark the spreading depression of activity (middle traces). Whereas the spontaneous activity transiently recovered at subdural electrodes 1–3, the activity depression persisted through death at subdural electrodes 4–6 after the first SD. The continuous physiological variables in the bottom traces remained stable during the terminal SD. The asterisk marks the moment when the EVD was left open to counteract the ICP increase. (B) As outlined by Box 2 in Figure 7, the time period following the one in (A) is shown. The terminal SD spread further to subdural electrode 3 to 2 to 1. The activity depression at the subdural electrodes was eventually followed by activity depression at the epidural electrodes which persisted through death. Note that the NUP was superimposed by additional SDs of decreasing amplitude. In the right lower corner, three neighboring slices of the last CT scan are given, showing the subdural electrodes anterior to the delayed infarct in the MCA territory.