www.elsevier.com/locate/ynimg NeuroImage 27 (2005) 715 – 724
BOLD signal increase preceeds EEG spike activity—a dynamic penicillin induced focal epilepsy in deep anesthesia Minna Ma¨kiranta,a,b,* Jyrki Ruohonen,a,f Kalervo Suominen,b Jaakko Niinima¨ki,a Eila Sonkaja¨rvi,c Vesa Kiviniemi,a Tapio Seppa¨nen,d Seppo Alahuhta,c Ville Ja¨ntti,e and Osmo Tervonen a a
Department of Diagnostic Radiology, University of Oulu, Kajaanintie 50, 90220 Oulu, Finland Department of Clinical Neurophysiology, Oulu University Hospital, Kajaanintie 50, 90220 Oulu, Finland c Department of Anesthesiology, University of Oulu, Kajaanintie 50, 90220 Oulu, Finland d Department of Electrical and Information Engineering, University of Oulu, Oulu, Finland e Ragnar Granit Institute, Tampere University of Technology, and Department of Clinical Neurophysiology, Tampere University Hospital, Tampere, Finland f NMR Research Group, Department of Physical Sciences, University of Oulu, P.O. Box 3000, 90014 Oulu, Finland b
Received 19 May 2004; revised 13 May 2005; accepted 14 May 2005 Available online 11 July 2005
In 40 – 60% of cases with interictal activity in EEG, fMRI cannot locate any focus or foci with simultaneous EEG/fMRI. In experimental focal epilepsy, a priori knowledge exists of the location of the epileptogenic area. This study aimed to develop and to test an experimental focal epilepsy model, which includes dynamic induction of epileptic activity, simultaneous EEG/fMRI, and deep anesthesia. Reported results are from seven pigs (23 T 2 kg) studied under isoflurane anesthesia (1.2 – 1.6 MAC, burst-suppression EEG) and muscle relaxant. Hypo- and hypercapnia were tested in one pig. Penicillin (6000 IU) was injected via a plastic catheter (inserted into the somatosensory cortex) during fMRI (GRE-EPI, TE = 40 ms, 300 ms/two slices, acquisition delay 1700 ms) in 1.5 T (N = 6). Epileptic spikes between acquisition artifacts were reviewed and EEG total power calculated. Cross-correlation between voxel time series and three model functions resembling induced spike activity were tested. Activation map averages were calculated. Development of penicillin induced focal epileptic activity was associated with linear increase and saturation up to approximately 10 – 20%, in BOLD activation map average. Its initial linear increase reached 2.5 – 10% at the appearance of the first distinguished spike in ipsilateral EEG in all six animals. Correlated voxels were located mainly in the vicinity of the penicillin injection site and midline, but few in the thalamus. In conclusion, development of focal epileptic activity can be detected as a BOLD signal change, even preceding the spike activity in scalp EEG. This experimental model contains potential for development and testing different localization methods and revealing the
* Corresponding author. Department of Diagnostic Radiology, University of Oulu, Kajaanintie 50, 90220 Oulu, Finland. Fax: +358 8 3152112. E-mail address:
[email protected]. (M. Ma¨kiranta). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2005.05.025
characteristic time sequence of epileptic activity with fMRI during deep anesthesia. D 2005 Elsevier Inc. All rights reserved. Keywords: Penicillin; Epilepsy; Anesthesia
Introduction The important indication for the clinical application of combined simultaneous electroencephalography and functional magnetic resonance imaging (EEG/fMRI) is to find the exact location of the epileptic focus in medically refractory epilepsy. However, in 40 – 60% of cases with interictal activity in EEG, fMRI cannot locate any focus or foci (Krakow et al., 1999; AlAsmi et al., 2003). There is a need to obtain more understanding about the coupling between interictal spike activity in EEG and the blood-oxygen-level-dependent (BOLD) contrast changes localizing the primary focus or foci. Such brain regions sometimes remain active although the activity of the rest of the brain is suppressed with anesthetics to treat status epilepticus or during epilepsy surgery (Fiol et al., 1993; Ja¨ntti et al., 1994; Prasad et al., 2001; Igartua et al., 1999). Deep anesthesia may provide a mean to limit confounding baseline activity in the brain, and also inhibit the spread of epileptic activity. Otherwise, the impact of focal epilepsy can be widespread on the brain function (Witte and Bruehl, 1999). In experimental focal epilepsy a priori knowledge exists where the true epileptogenic area is located. This enables characterization of the surrounding BOLD changes around the true primary focus region, comparison with EEG changes and further development of fMRI localization techniques. A few previous
716
M. Ma¨kiranta et al. / NeuroImage 27 (2005) 715 – 724
studies utilizing experimental epilepsy models with simultaneous EEG/fMRI exist (Opdam et al., 2002; Van Camp et al., 2003). Dynamic induction of experimental epilepsy during fMRI could give further insight about the characteristic time sequence of fMRI changes in epilepsy. Epileptogenic agents (e.g., bicuculline, picrotoxin and penicillin) are commonly used to induce a focal epileptic lesion. Penicillin G is a GABAA receptor blocker, impairing GABA-dependent inhibition and widely used in focal epilepsy models (Avanzini and Franceschetti, 2003). Opdam et al. (2002) have demonstrated the first fMRI results during focal penicillin induced seizure activity in electrocorticogram. However, no dynamic studies were found from the literature with simultaneous EEG/fMRI. The goal of this study was to develop and to test an experimental focal epilepsy model, which combines dynamic induction of epileptic activity, simultaneous EEG/fMRI and deep anesthesia as a relevant model for status epilepticus management or epilepsy surgery. Our objective was to study the temporal and spatial relationship between the BOLD signal and EEG in the dynamically induced experimental epileptic zone.
Materials and methods The animal model and experimental procedure Thirteen female piglets (2 to 3 months, 20 – 33 kg) were studied altogether. First, pilot experiments with six animals were performed to adjust anesthesia, suitable catheter, penicillin dose, filling, catheter fixation, dynamic vs. non-dynamic epilepsy induction and application of EEG/MRI. The results reported here include data from the last six animals (2 to 3 months, 20 – 24 kg) and from one animal with a hypo/hypercapnia experiment during fMRI. The pilot experiment with one pig showed that midazolam, used for general anesthesia as a treatment for status epilepticus in children (Igartua et al., 1999), was not sufficient for general anesthesia in piglets, even in massive doses. The level of anesthesia is well controlled with a volatile anesthetic, especially when EEG burst-suppression pattern is the end-point (e.g., Mustola et al., 2003). In piglets, isoflurane induces a clear burst-suppression EEG pattern. Isoflurane has been shown to be the optimal anesthetic in fMRI with dogs (Willis et al., 2001). Another previous fMRI study reported diminished or scarce BOLD response during deep isoflurane anesthesia in rats (Tenney et al., 2003). To explore the fMRI signal change in deep isoflurane anesthesia, one experiment was undertaken with hypocapnia and hypercapnia tests, without EEG (Fig. 1). Animals were fasted 12 h prior to the induction of the anesthesia. Premedication consisted of i.m. midazolam (Dormicum\, 1.5 mg/kg) and ketamine (Ketalar\, 15 mg/kg). A venous cannula was inserted into the ear of a pig. Intubation was facilitated with i.v. administration of thiopental (Pentothal\, 25 mg/ml). The animals were normoventilated (7 – 8 L/min, 18 rpm) with 40% oxygen in air. Anesthesia was maintained with isoflurane at end-tidal concentrations (ET%) of 1.4 – 1.8%, the EEG burst-suppression pattern as an end-point. Muscular paralysis was obtained with repeated doses of pancuronium bromide (Pavulon, 4 mg/h). Relaxation was not induced before the preferred anesthesia level was obtained. Prior to the preparation required for catheter insertion, local anesthesia was applied (lidocaine and adrenaline, Xylocain\ adrenalin, 10 mg/ml + 5 Ag/ml) together with fentanyl
boluses 50 Ag i.v. Anesthesia throughout the study period was supervised by a senior anesthesiologist (E.S.). An intra-arterial cannula for mean arterial pressure (MAP) measurement was inserted into the femoral artery of pig, successful in three of six animals. Failures occurred owing to the small and contractible arteries of the pig. MAP, heart rate, end-tidal isoflurane and carbon dioxide (CO2) concentrations were monitored (S/5i Compact Anesthesia Monitor, Datex-Ohmeda, Helsinki, Finland) and recorded with 5-s time resolution during MRI. One-minute data from the beginning of the experiment (baseline) and from the end of the experiment were averaged. Plastic epidural catheter (Portex\ epidural minipack, 19-gauge) for penicillin injection was carefully prefilled to avoid air bubbles with 0.9% NaCl (tip) and benzylpenicillin sodium (Geepenil\) in 0.9% NaCl. It was inserted into the somatosensory cortex (Craner and Ray, 1991) at a target depth of 5 mm below the dura mater, through a cranial hole on the left side anteriorly to the coronal suture. The catheter was fixed securely with tissue glue. In pilot studies, air arriving through the catheter raised serious problems in the image quality. However, only in one case (Pig F), air caused a major signal intensity drop at the injection site. This animal (Pig F) also had a post-mortem penicillin injection. To dynamically follow development of epileptic activity, a bolus of penicillin (6000 IU) was injected during simultaneous fMRI and EEG recording. The injected volume was 0.11 ml due to the catheter volume. The baseline was 1 min (N = 3) or 3 min (N = 3). At the end of the MRI experiment, the pigs were euthanized with an overdose of pentobarbital (Mebunat\). The protocol of these experiments had the approval of the local Research Animal Care and Use Committee. EEG Digital EEG equipment, EEG-electrodes and cables compatible with the MRI environment (Scan\, SynAmps\, NeuroScan\, MagLink\, NeuroScan, El Paso, TX) were used with the following amplifier set-up parameters: DC-recording, bandwidth 0 – 200 Hz, gain 150, sampling frequency 1000 Hz, range 37 mV, accuracy 0.559 AV. Two EEG electrodes were attached with tissue glue onto the surface of the pig skull, over the right (contralateral to lesion) and the left (ipsilateral to lesion) side posteriorly to the coronal suture. The reference and ground electrodes were attached to the most frontal area, around the frontal sinuses. EEG was continuously recorded throughout the MRI session. In the acquisition used delayed fMRI (1 fMRI image set in 300 ms and acquisition delay 1700 ms), EEG was readable during a 1700ms delay. A display filter
