Simultaneous multilobar electrocorticography (mEcoG) and scalp electroencephalography (scalp EEG) during intracranial vascular surgery: A new approach in neuromonitoring

Clinical Neurophysiology 116 (2005) 2734–2740 www.elsevier.com/locate/clinph

Simultaneous multilobar electrocorticography (mEcoG) and scalp electroencephalography (scalp EEG) during intracranial vascular surgery: A new approach in neuromonitoring Damien Debatissea,*, Etienne Pralonga, Amir R. Dehdashtib, Luca Reglib a

Unite´ Neurochirurgicale de Neuromonitoring (UNN), Centre Hospitalier Universitaire Vaudois (CHUV), 1011 Lausanne, Switzerland b De´partement de Neurochirurgie, Centre Hospitalier Universitaire Vaudois (CHUV), 1011 Lausanne, Switzerland Accepted 12 August 2005 Available online 26 October 2005

Abstract Objective: Intraoperative neuromonitoring for intracranial vascular surgery is primarily aimed at detecting early ischemic changes to prevent subsequent infarction. Despite various neurophysiological approaches detection of early and focal ischemic changes remains difficult. This study explores the feasibility and sensitivity of intraoperative monitoring using surface EEG (scalp EEG) and multilobar Electrocorticography (mEcoG) recording during intracranial vascular procedures. Methods: About 21 recordings were acquired in 20 patients undergoing craniotomies for intracranial aneurysms (17), superficial temporalmiddle cerebral artery bypass (twice in the same patient) and arteriovenous malformation (2). The recording of scalp EEG (needle electrodes) and EcoG was performed (cupules electrodes) during all of the surgery. Signal was visually analyzed online and using spectral analysis software offline. Results: Good recordings were obtained in all cases, without adding any procedural morbidity. The most common abnormalities on mEcoG were high frequency waves (23–37 Hz; HF-b3), which appeared just after vascular occlusion and were occasionally followed by slow waves or burst suppression pattern. This focal pattern was seen in a majority of cases (20/21) on the mEcoG, but only in 4 out of 21 cases on the EEG. Conclusions: Multi-lobe (mEcoG) recording is feasible during craniotomies and detects earlier and more EEG pattern variation than surface EEG monitoring during intracranial vascular manipulations. The authors discuss the high sensitivity of this technique to ischemic changes. Significance: By detecting earlier and more focal changes than scalp EEG, mEcoG may favor during surgery an increase in interactive strategies and reduction of deleterious event. q 2005 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. Keywords: Intraoperative monitoring; Aneurysm; Vascular malformation; Electrocorticography (EcoG); EEG

1. Introduction Multimodality cerebral monitoring aims at identifying events liable to cause brain damage, and is mainly used in the operating room and intensive care unit (ICU) (Florence et al., 2004; Guerit, 1999; Lopez et al., 1999; Martin et al., 2002; Nuwer et al., 1999). Its primary goals * Corresponding author. Department of Neurosurgery (NCH-UNN), Centre Hospitalier Universitaire Vaudois, Lausanne 1011, Switzerland. Tel.: C41 21 314 2552; fax: C41 21 314 2595. E-mail address: [email protected] (D. Debatisse).

are: (1) the detection of brain dysfunction at a reversible stage in order to prevent irreversible lesions, (2) the indirect evaluation of metabolic disturbances and of the level of anesthesia, and (3) the understanding of neurological events occurring intra-operatively, in order to improve outcome. The history of neuromonitoring in the ICU dates back to the EEG discovery in the late 1920 s when it was considered a unique tool to evaluate the state of consciousness in comatose patients. Evoked potentials (EPs) were progressively introduced at the end of the 1970 s, both in the ICU and the operating theater.

1388-2457/$30.00 q 2005 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.clinph.2005.08.011

D. Debatisse et al. / Clinical Neurophysiology 116 (2005) 2734–2740

Temporary arterial occlusions are common during intracranial vascular surgery. Real-time detection of cerebral ischemia by monitoring could draw the attention of the surgeon to the inadequacy of collateral blood flow during the occlusion time. Monitoring techniques used clinically in anterior circulation aneurysm surgery include somatosensory evoked potentials (SSEPs), EEG, and rarely monolobar EcoG (using grid or strip electrodes). The value of SSEPs has been investigated in a large number of patients and responses were found to change in roughly 20% of surgical procedures. Median nerve SSEPs monitor the middle cerebral artery territory ignoring other locations, and have therefore a significant incidence of false negative results (Martin et al., 2002; Min et al., 2001; Parenti et al., 1996). Lower-limb SSEPs have been reported to be sensitive to ischemia in the anterior cerebral artery territory (Guerit, 1999). Variations in the EEG are less well studied with 5–40% of patients showing unspecific changes strongly dependent on the surgical exposure (extension of skin flap and craniotomy) and the level of anesthesia (Nuwer, 1993; Vespa et al., 1997). In contrast to SSEPs, EcoG can monitor different vascular territories depending on the positioning of the electrodes (Guerit, 1998). The purpose of this study was to detect early EEG pattern changes during temporary arterial occlusion in surgery. Therefore, we analyzed (1) the feasibility and (2) the sensitivity of continuous EcoG recording by multilobar electrodes placed in the operative field during intracranial vascular surgery and compared it to bilateral surface EEG.

2735

Following the classical pterional craniotomy and dural opening (Yasargil et al., 1987), three subdural mEcoG cupula electrodes were positioned on the frontal, temporal, and parietal lobes (Fig. 1). Cupules electrodes were chosen for their ease of use. Electrode placement did not require modification of the surgical technique for the craniotomy. Similarly, the electrodes did not influence the intra-dural surgical procedure. Positioning of the three mEcoG electrodes (frontal, temporal, parietal), added on average 5 min to the procedure (Fig. 2). To reduce electrical artifacts, the wires were stapled to the operative field and a new ground electrode (G1) shifted to the temporal muscle. Online monitoring was performed with a total of 7 leads: 4 for the scalp EEG (two ipsilateral and two contralateral), and 3 for the mEcoG. This study focuses on the early detection of changes noted on scalp EEG and mEcoG recordings during temporary clipping of intra-cranial arteries. After 5 min of flow reestablishment, EEG pattern normalization was evaluated (Fig. 3). Offline Fast Fourrier (FFT) signal analysis was performed using the ASAw software (ANT, Germany) (Schiff et al., 1994; Zanow and Peters, 1995). This report is a feasibility study, therefore surgical decisionmaking was independent and no interaction strategy was designed.

2. Material and methods Intra-operative monitoring with EEG and multilobar EcoG (mECoG) was performed in 21 recordings using a Micromedw digital EEG system (Micromed, Italy) with 32 channels. The signal was sampled at 256 Hz and visualized using a bandpass filter between 0.3 and 100 Hz. Surface EEG gain was 50 or 100 mV/cm and mEcoG was 200 or 400 mV Ground (G1) and reference electrode (G2) were positioned in A1/A2. All patients had a preoperative baseline scalp EEG in the operating theater under the same anesthetic conditions. During surgery, surface needle electrodes were positioned in Fp1/Fp2 and CP3/CP4. Signal was visualized in monopolar and bipolar montage (Fp1–CP3, Fp2–CP4, Fp1–Fp2, CP3–CP4), allowing monitoring of anesthesia and of intraoperative baseline EEG bilaterally. Induction of anesthesia was performed using propofol to reduce cerebral volume. After craniotomy and the opening of the dura-mater, anesthesia was shifted to sevoflurane in order to reach a constant background EEG and to reduce pharmacologically induced EEG high frequencies and bursts (Huotari et al., 2004). During temporary clipping, systemic arterial hypertension was induced (increase by 20% of the mean arterial pressure; at least 100 mmHg).

Fig. 1. Typical pterional craniotomy illustrating the positioning of mEcoG leads. Before anesthesia, surface needle electrodes were positioned in Fp1/Fp2 and CP3/CP4 allowing monitoring of anesthesia pharmaceutical effect and of the scalp EEG intraoperative baseline. Following the craniotomy and dural opening, three subdural mEcoG cupules electrodes were positioned on the frontal (F), temporal (T), and parietal (P) lobes, and one ground on the muscle (not seen). mEcoG placement did not require modification of the surgical technique for the craniotomy and were not a constraint during the procedure. Positioning of the 3 mEcoG electrodes added on average 5 min to the procedure.

2736

D. Debatisse et al. / Clinical Neurophysiology 116 (2005) 2734–2740

Fp1-G2 FP2-G2 C3-G2 C4-G2 Fp1-C3 Fp2-C4

1 EcoG Frontal-G2 2 EcoG Temporal-G2 3 EcoG Parietal-G2 1-2 2-3 1-3

LFF=0.3 HFF=100

1 Sec.

Scalp EEG (Gain 50µV) and mEcoG (200µV) = Baseline

Fig. 2. Baseline before transitory clipping. The first six derivations correspond to monopolar and bipolar recordings of the scalp EEG (LFFZ0.3 Hz, HFFZ 100 Hz and gain 50 mV). The last six derivations correspond to monopolar and bipolar recordings of the cortical (mEcoG) activity (HFFZ0.3 Hz, LFFZ 100 Hz and gain 200 mV). We observed the presence of beta2 (18–20 Hz) activity on the mEcoG.

3. Results Twenty patients were monitored for temporary clipping during 21 craniotomies for intracranial vascular lesions: 17 had aneurysms, two had arterio-venous malformations (AVM), and one had internal carotid artery occlusion and underwent two separate procedures for extracranial– intracranial bypass surgery. Twelve patients (60%) presented with intracranial hemorrhage. Mean age was 49C11 years; there were 13 females and 7 males (Table 1). Positioning of the subdural electrodes was possible in all patients (Fig. 1). In the 12 patients presenting with intracranial hemorrhage, intracranial hypertension was either controlled by CSF drainage or hematoma evacuation, before subdural electrode placement. In all patients scalp EEG and mEcoG signals were recorded (Fig. 2), allowing for adequate on- and offline analyses of the EEG pattern. All patients, except 4, underwent one or more episodes of intracranial arterial temporary clipping. Temporary clipping duration ranged from 1 to 80 min (mean 29C25 min; median 14 min). Changes noted in mEcoG recordings following arterial occlusion (17 recordings in 16 patients) were characterized by early (less than 2 min) high frequency waves (HF-b3; 23–37 Hz) (Fig. 4). HF-b3 were the only

observed abnormalities in 2/17 monitorings (12%). In 15/17 (88%) HF-b3 were followed by delta-theta waves (2–6 Hz, Fig. 2), 8 of which were followed by temporary focal ‘burst suppression pattern’ (Table 1). HF-b3 and delta–theta waves were never detected on the scalp EEG (0/17) (Table 1). Only two patients presented a burst suppression pattern simultaneously on both the mEcoG and scalp EEG recordings (Table 1). mEcoG changes persisted in 7/17 monitorings 5 min after clip removal. In two of them, the primary lesion was possibly responsible for an increased susceptibility to ischemia (one ruptured AVM and one middle cerebral artery (MCA) aneurysm both with large intraparenchymal hematomas). For the 5 other monitorings with persistent changes: in 3 patients temporary clipping, although of short duration (2–8 min), was applied to the internal carotid artery distal to the post-communicating artery (C1 segment), and in two patients temporary clipping was longer than 60 min (one anterior cerebral artery (A1) and one middle cerebral artery (M1)). Among the 4 patients without intraoperative temporary arterial interruption none presented new changes of the EEG recording during surgery. Two showed, however, abnormal recordings (HF-b3, theta waves and burst suppression pattern on mEcoG and burst suppression pattern on Scalp

D. Debatisse et al. / Clinical Neurophysiology 116 (2005) 2734–2740

2737

Fp1-G2 FP2-G2 C3-G2 C4-G2 Fp1-C3

Fp2-C4 1 EcoG Frontal-G2 2 EcoG Temporal-G2 3 EcoG Parietal-G2 1-2 2-3 1-3 LFF=0.3 HFF=100 1 Sec.

Scalp EEG (Gain 50 µV) and mEcoG (200 µV) = End of Surgery

Fig. 3. Scalp EEG and mEcoG recorded at the end of surgery after the definitive clip. The first six derivations correspond to monopolar and bipolar recordings of the scalp EEG (LFFZ0.3 Hz, HFFZ100 Hz and gain 50 mV). The last six derivations correspond to monopolar and bipolar recordings of the cortical (mEcoG) activity (HFFZ0.3 Hz, LFFZ100 Hz and gain 200 mV). Surgery’s end corresponds to a return to baselines scalp EEG and mEcoG.

EEG) during the whole period of monitoring. One patient was severely diabetic and had an acute subarachnoid hemorrhage, and the other presented with severe deficits secondary to a ruptured AVM.

4. Discussion The interpretation of intraoperative neurophysiological results during intracranial vascular surgery is delicate and complex. Despite the limited size of this series, our preliminary results reveal two major elements suggesting the potential value of this neuromonitoring technique. The first element is the feasibility of continuous multilobar EcoG monitoring during intracranial vascular surgery. The positioning of multiple subdural corticographic electrodes (temporal, frontal and parietal) was always feasible, including in patients with acutely ruptured aneurysms. In patients with symptomatic intracranial hypertension due to large intraparenchymal hematomas, evacuation of the clot has to be performed before electrode placement. mEcoG did not necessitate an increase of the craniotomy size compared to the standard pterional

exposure described earlier (Yasargil, 1984). The time spent to position the electrodes was less than 5 min. The second element shown by this study is the high sensitivity of mEcoG compared to scalp EEG to changes of the monitored modality during arterial occlusion. The sensitivity of neuromonitoring varied greatly depending on the recording technique, with an overall sensitivity of 100% for mEcoG (17/17) and of only 12% for EEG (2/17) in this study. The high sensitivity of mEcoG can be attributed to several factors: the multilobar recording technique, that allows to monitor the lobes by mEcoG (temporal, parietal, and frontal) corresponding to the vascular territories at risk; the cortical location of the electrodes closer to the threatened parenchyma; and the elimination of bone and scalp filter. Scalp EEG monitoring, conversely, is bound by the extension of the scalp and bone flap. The earliest changes noted was HF-b3 appearing as soon as 2 min after arterial occlusion, followed by delta waves, and eventually burst suppression pattern with prolonged occlusion. More precisely, evolution of the neurophysiological abnormalities from early ‘high frequency waves’ (HFb3) to delta–theta waves was observed in 88% of temporary occlusions (15/17), with further progression to burst

2738

Table 1 Summary of all the investigated patients characteristics Sex

Age

Lesion

Ruptured

Temp. clip

Duration (min)

Pre´clamp. scalp EEG

Pre´clamp. EcoG patho

CLIP:HF mEcoG

Theta/ Delta mEcoG

BURST mEcoG

Scalp EEG

Clip remov mEcoG

Clip remov scalp EEG

1 2 3 4

M M F F

21 35 36 38

AcoA MCA MCA PcoA g

C K C K

1* 0 10 5

K K K K

K K K K

C K C C

C K K C

K K K C

K K C K

K K K C

K K K K

5

M

40

C

45

C

C

C

C

C

C

C

K

6 7 8 9 10 11 12

F F F M F F M

41 44 46 47 49 50 50

C C K K C C C

M1 A1 No M1 No M1 No

18 60 0 4 0 60 0

K K K K C K K

C K K K C C C

C C C C C C C

C C C K C C C

C C K K C C K

K K K K C K C

K K K K C C C

K K K K C K C

13 14

F F

52 54

AVM: subcallosal MCA AcoA MCA g MCA AcoA M1 AVM: temp MCA PcoA

A2 No M1 ICA postPcom Both A2

K C

7 5

K K

K K

C C

C C

K K

K K

K K

K K

15

F

54

M1 ICA postPcom A2

30

K

K

C

C

K

K

K

K

16a

M

58

16b

M

58

17 18

F F

60 60

19

F

61

20

M

70

Pericallosal Anas. STAMCA Anas. ECAMCA AcoA 2MCA, PcoA ICA bifurcation AchA

C K

STAMCA

67

C

C

C

C

K

K

K

K

K

STAMCA

64

K

K

C

C

C

C

C

C

C K

A1 M2

80 1*

K K

K K

C C

C C

K K

K K

C K

K K

K

ICA postPcom ICA postPcom

8

K

K

C

C

K

K

C

K

1

C

C

C

C

C

K

C

K

C

D. Debatisse et al. / Clinical Neurophysiology 116 (2005) 2734–2740

Patients

D. Debatisse et al. / Clinical Neurophysiology 116 (2005) 2734–2740

2739

Fp1-G2 FP2-G2 C3-G2 C4-G2 Fp1-C3 Fp2-C4 1 EcoG Frontal-G2 2 EcoG Temporal-G2 3 EcoG Parietal-G2 1-2 2-3 1-3 LFF=0.3 HFF=100 1 Sec

Scalp EEG (Gain 50 µV) and mEcoG (200 µV) = Temporary Clip

Fig. 4. Early changes induced by temporary clipping on the EcoG and sEEG. The first six derivations correspond to monopolar and bipolar recordings of the scalp EEG (LFFZ0.3 Hz, HFFZ100 Hz and gain 50 mV). The last six derivations correspond to monopolar and bipolar recordings of the cortical (mEcoG) activity (HFFZ0.3 Hz, LFFZ100 Hz and gain 200 mV). Two main features emerge from these traces recorded during temporary clipping: (a) an increase of the delta–theta activity on the parietal mEcoG and (b) high frequencies (HF-b3) with 24–40 Hz on the frontal and temporal mEcoG. No change was observed on the scalp EEG.

suppression pattern in 41% (7/17). mEcoG was able to detect one, two or all three of these stages, while Scalp EEG detected only the final stage in 4/22 monitorings. This finding suggests, that mEcoG is not only more sensitive, but possibly also more specific in the detection of early intraoperative ischemia compared to Scalp EEG. Interestingly, persistent mEcoG changes tended to correlated with duration (60 min and more) or site of the temporary occlusion (post-communicating segment of the internal carotid artery). Finally, the high sensitivity of mEcoG to HF-b3 may be of limited use for clinical neuromonitoring and decision making during hypoperfusion. However, we hypothesize that the combination of HF-b3 followed by theta–delta as seen on mEcoG only may indicate a threatening threshold. A more detailed analysis correlating the monitoring data with patient outcome is necessary to appreciate the specificity, the positive and negative predictive value of mEcoG, but is beyond the scope of this preliminary study that was designed without an interactive strategy during surgery. The HF-b3 waves were characterized by ‘spindle-like pattern’ in a range of frequency between 23 and 37 Hz

that have not been described in humans before, to best of our knowledge. The origin of the HF-b3 waves may reflect the transition between an aerobic and anaerobic state of metabolism. Ischemia in vitro induces a biphasic neuronal response consisting, first, of cell hyperpolarization due to potassium channel activation (Fujimura et al., 1997), rapidly followed in minutes by cell depolarization due to the fall of ATP-dependent ionic transporters (i.e. the Na/K-ATPase) (Muller and Somjen, 2000). Interestingly, cell spiking and ‘20–40 Hz network oscillation’ accompany anoxic depolarization. These oscillations depend on an intact GABAergic and glutamatergic transmission and can be induced in normoxic conditions by an increase in extracellular potassium concentration in the presence of adenosine A1 agonist (Dzhala et al., 2001). A similar biphasic response was also observed in vivo in rats (Xie et al., 1995). Based on the above, we hypothesize that the HF-b3 waves observed minutes after clamping in our patients may represent the electrophysiological correlates of the anoxia-induced oscillation. One could then assume, that as long as HF-b3 waves are the only observed changes, the vascular territory supports

2740

D. Debatisse et al. / Clinical Neurophysiology 116 (2005) 2734–2740

the temporary reduction of blood flow, while on the contrary the territory is suffering, when the HF-b3 waves develop into slow delta waves and burst suppression patterns, classically described as the neurophysiological correlate of anaerobic metabolism. As shown in our study, intra-operative scalp EEG alone is not sensitive enough. We believe that combined approaches are necessary to improve the predictive value of neuromonitoring. mEcoG, scalp EEG, SSEP, measurements of flow, and metabolic markers combined in a multimodality monitoring scheme may form the lines of future research for intra-operative neuromonitoring.

5. Conclusion Multilobar electrocorticography simultaneously monitoring temporal, parietal and frontal lobes is easy to perform and is more sensitive than scalp EEG in detecting lobar neurophysiological pattern changes during temporary arterial occlusion. In our hands, the poor sensitivity of scalp EEG to neurophysiological changes during temporary arterial occlusion precludes its use as a valuable tool for monitoring intracranial vascular surgery. However, the usefulness of each mEcoG pattern observed during vascular occlusion need to be evaluated further, before one can substantiate its clinical value. In the future, electrophysiological neuromonitoring of cerebrovascular diseases should combine multimodality monitoring scheme.

Acknowledgements This study was supported by a grant to L.R. from “Fondation Leenaards”, Switzerland.

References Dzhala V, Khalilov I, Ben-Ari Y, Khazipov R. Neuronal mechanisms of the anoxia-induced network oscillations in the rat hippocampus in vitro. J Physiol 2001;536:521–31. Florence G, Guerit JM, Gueguen B. Electroencephalography (EEG) and somatosensory evoked potentials (SEP) to prevent cerebral ischaemia in the operating room. Neurophysiol Clin 2004;34:17–32.

Fujimura N, Tanaka E, Yamamoto S, Shigemori M, Higashi H. Contribution of ATP-sensitive potassium channels to hypoxic hyperpolarization in rat hippocampal CA1 neurons in vitro. J Neurophysiol 1997;77:378–85. Guerit JM. Neuromonitoring in the operating room: why, when and how to monitor? Electroencephalogr Clin Neurophysiol 1998;106:1–21. Guerit JM. Medical technology assessment EEG and evoked potentials in the intensive care unit. Neurophysiol Clin 1999;29:301–17. Huotari AM, Koskinen M, Suominen K, Alahuhta S, Remes R, Hartikainen KM, Evoked EEG, Jantti V. Evoked EEG patterns during burst suppression with propofol. Br J Anaesth 2004;92:18–24. Lopez JR, Chang SD, Steinberg GK. The use of electrophysiological monitoring in the intraoperative management of intracranial aneurysms. J Neurol Neurosurg Psychiatry 1999;66:189–96. Martin CJ, Sinson G, Patterson T, Zager EL, Stecker MM. Sensitivity of scalp EEG, cortical EGG, and somatosensory evoked responses during surgery for intracranial aneurysms. Surg Neurol 2002;58:317–20 [discussion 320–21]. Min KT, Kim JH, Shin YS, Kwon SY, Nam YT. The monitoring of somatosensory evoked potentials and neurologic complications in aneurysm surgery. Yonsei Med J 2001;42(2):227–32. Muller M, Somjen GG. Na(C) and K(C) concentrations, extra- and intracellular voltages, and the effect of TTX in hypoxic rat hippocampal slices. J Neurophysiol 2000;83:735–45. Nuwer MR, Comi G, Emerson R, Fuglsang-Frederiksen A, Guerit JM, Hinrichs H, Ikeda A, Luccas FJ, Rappelsberger P. IFCN standards for digital recording of clinical EEG. The international federation of clinical neurophysiology. Electroencephalogr Clin Neurophysiol Suppl 1999;52:11–14. Nuwer MR. Intraoperative electroencephalography. J Clin Neurophysiol 1993;10(4):437–44. Parenti G, Marconi F, Fiori L. Electrophysiological (EEG–SSEP) monitoring during middle cerebral aneurysm surgery. J Neurosurg Sci 1996;40(3–4):195–205. Schiff SJ, Aldroubi A, Unser M, Sato S. Fast wavelet transformation of EEG. Electroencephalogr Clin Neurophysiol 1994;91:442–55. Vespa PM, Nuwer MR, Juhasz C, Alexander M, Martin N, Becker DP. Early detection of vasospasm after subarachnoid hemorrhage using continuous quantitative EEG. Electroencephalogr Clin Neurophysiol 1997;103:607–15. Xie Y, Zacharias E, Hoff P, Tegtmeier F. Ion channel involvement in anoxic depolarization induced by cardiac arrest in rat brain. J Cereb Blood Flow Metab 1995;15:587–94. Yasargil MG. Micorsurgical anatomy of the basal cisterns and vessels of the brain, diagnostic studies, general operative techniques and pathological considerations of the intracranial aneurysms. Stuttgart: Thieme; 1984. p. 215–227. Yasargil MG, Reichman MV, Kubik S. Preservation of the frontotemporal branch of the facial nerve using the interfascial temporalis flap for pterional craniotomy. J Neurosurg 1987;67(3):463–6. Zanow F, Peters MJ. Individually shaped volume conductor models of the head in EEG source localisation. Med Biol Eng Comput 1995;33: 582–8.