Address correspondence and reprint requests to Dr. A. Pitkänen at AI Virtanen Institute for Molecular Sciences, University of Kuopio, PO Box 1627, FIN-70 211 Kuopio, Finland. E-mail: firstname.lastname@example.org
Summary: Purpose: This study examined the hypothesis that neurodegeneration continues after status epilepticus (SE) ends and that the severity of damage at the early phase of the epileptogenic process predicts the outcome of epilepsy in a long-term follow-up.
Methods: SE was induced in rats by electrical stimulation of the amygdala, and the progression of structural alterations was monitored with multiparametric magnetic resonance imaging (MRI). Absolute T2, T1ρ, and diffusion (Dav) images were acquired from amygdala, piriform cortex, thalamus, and hippocampus for ≤4.5 months after SE. Frequency and type of spontaneous seizures were monitored with video-electroencephalography recordings. Histologic damage was assessed from Nissl, Timm, and Fluoro-Jade B preparations at 8 months.
Results: At the acute phase (2 days after SE induction), quantitative MRI revealed increased T2, T1ρ, and Dav values in the primary focal area (amygdala), reflecting disturbed water homeostasis and possible early structural damage. Pathologic T2 and T1ρ were observed in mono- or polysynaptically connected regions, including the piriform cortex, midline thalamus, and hippocampus. The majority of acute MRI abnormalities were reversed by 9 days after SE. In later time points (>20 days after induction), both the T1ρ and diffusion MRI revealed secondarily affected areas, most predominantly in the amygdala and hippocampus. At this time, animals began to have spontaneous seizures. The initial pathology revealed by MRI had a low predictive value for the subsequent severity of epilepsy and tissue damage.
Conclusions: The results demonstrate progressive neurodegeneration after SE in the amygdala and the hippocampus and stress the need for continued administration of neuroprotectants in the treatment of SE even after electrographic seizure activity has ceased.
Status epilepticus (SE) is a clinical emergency with an incidence of ∼0.1%, resulting in estimated 180,000 episodes in the United States and 365,000 in Europe annually (1). SE is associated with a high risk of mortality (20–60%) as well as morbidity, including epileptogenesis (2) and cognitive decline (1). Data from humans and experimental models of SE suggest that both the risks of epilepsy and the severity of cognitive impairment are associated with brain damage caused by prolonged seizure activity (1).
Histologic studies in experimental models indicate that SE lasting for 30 to 40 min in rats (3,4) or 80 min in nonhuman primates (5) is long enough to initiate neurodegeneration. It is not, however, known for how long the neurodegenerative process advances after SE in different brain regions, and what is its temporal relation to epileptogenesis. Several studies in rats indicate widespread damage in cortical and subcortical regions a few days after either chemically or electrically induced SE (6,7). Recent histologic data from animal models suggest that SE-induced damage continues for several weeks (8). A growing number of follow-up studies in humans with magnetic resonance imaging (MRI) indicate that hippocampal atrophy progresses over weeks, months, or years after the first SE (for review, see Pitkänen et al., 1999). However, several caveats exists in the interpretation of the previous studies. Data from animal models are cross-sectional, and human studies are based on the follow-up of single patients; therefore it is difficult to assess the time-course of damage and its association with morbidity. These data, using surrogate markers to predict the phase of epileptogenesis after SE, are critically important for determining the window for optimal neuroprotective treatment.
Recent advances in quantitative MRI methods provide noninvasive tools for assessing the functional integrity (9), disturbed water homeostasis, and progressing structural damage over a period of time in experimental animals (10). Much of our understanding of the behavior of quantitative MRI parameters, such as relaxation times and diffusion, during evolving tissue damage comes from both experimental and clinical studies of stroke (11). These studies show a decrease in the diffusion coefficient of water in the hyperacute phase of cerebral ischemia, generally associated with depolarization of the cells due to energy failure, with subsequent changes in the intracellular-to-extracellular water ratio, as well as with decreased diffusivity in the cell interior and increased extracellular tortuosity (12). These initial changes might be reversible and, in many cases, might not correlate with the outcome of the tissue. During the evolution of damage, diffusion MRI normalizes, and in the chronic phase, it increases above normal levels. This MRI characteristic of the chronic stroke phase most likely reflects increased water content in the tissue and disruption of cellular structures. Changes in relaxation times are detected several hours after the induction of cerebral ischemia and thus are often associated with irreversible damage. Edema, magnetic susceptibility, and microenvironmental changes determine relaxation time–based contrast in damaging tissue. Recently, T1ρ MRI was reported to be sensitive to ischemia-induced changes in brain tissue, showing a high predictive value for the degree of developing neuronal damage (13). Because the disease processes leading to cell death after ischemia and other epileptogenic insults share some common features (14), multiparametric MRI might be useful to assess tissue status during evolving epileptogenesis.
We hypothesized that structural degeneration would progress for much longer than SE-associated seizure activity, and that the severity of damage at the early phase of the epileptogenic process predicts the severity of epilepsy in a long-term follow-up. We used a model of temporal lobe epilepsy (TLE) recently developed in our laboratory, in which SE triggered by electrical stimulation of the amygdala induces epileptogenesis, resulting in spontaneous seizures (15). Animals were examined by using quantitative MRI focusing on the brain areas associated with epileptogenesis (amygdala, hippocampus, piriform cortex, thalamus) sequentially until ≤4.5 months after inducing SE. Animals also were monitored with video-EEG for ∼8 months to determine the nature and frequency of spontaneous seizures. Histologic assessment was performed at the end of the study. Some of the MRI data were previously published as a preliminary report (8).
MATERIALS AND METHODS
Twenty adult male Harlan Sprague–Dawley rats weighing 333 ± 23 g on the day of surgery and 432 ± 103 g at day 153 after stimulation; and five adult male non–operated-on Wistar rats (437 ± 62 g; naive MRI controls) were used in the present study. The 20 Sprague–Dawley rats were divided into two groups; operated on and stimulated (n = 17) and operated on and not stimulated (n = 3). The rats were individually housed in a controlled environment (constant temperature, 22 ± 1°C; humidity, 50–60%; lights on 0700–1900). Animals had free access to food and water. All animal procedures were conducted in accordance with the guidelines set by the European Community Council Directives 86/609/EEC.
Implantation of amygdaloid and cortical electrodes
The rats were deeply anesthetized with sodium pentobarbital (PTB; 60 mg/kg, i.p.) and chloral hydrate (100 mg/kg i.p.) for implantation of the stimulation electrode. A bipolar removable stainless steel electrode (diameter, 0.127 mm; dorsoventral distance between the tips, 0.4 mm; Franco Corradi, Milano, Italy), surrounded by a plastic guide cannula above the skull, was implanted into the lateral nucleus of the left amygdala [3.6 mm posterior to bregma, 5.0 mm lateral to bregma, 6.5 mm ventral to the surface of the brain, according to the rat brain atlas of Paxinos and Watson (1986)]. The electrode setup was fixed to the skull with dental acrylate (Selectaplus CN; Dentsplay DeTrey GmbH, Dreieich, Germany).
To record the electrographic seizure activity, a home-made screw electrode was constructed from a plastic screw (Plastics One Inc., Roanoke, VA, U.S.A.) and a Teflon-coated platinum/iridium wire (0.005”; A-M Systems Inc., Carlsborg, WA, U.S.A.). The wire electrode supported by the plastic screw was implanted into the skull overlying the contralateral frontal cortex (coordinates: 3.0 mm anterior and 2.0 mm lateral to bregma). Two monopolar electrodes composed of a plastic screw and platinum/iridium wire were fixed to the skull symmetrically over the cerebellum and served as ground and reference electrodes, respectively.
Induction and video-EEG monitoring of SE
Stimulation paradigm and apparatus
Four weeks after surgery, a baseline EEG was recorded for ≥1 min from each rat to make sure that the EEG was recordable. SE was induced by stimulating the amygdala for 20 to 40 min (15). In brief, the stimulation consisted of a 100-ms train of 1-ms biphasic square-wave pulses (400 μA from peak to peak) delivered at 60 Hz every 0.5 s by using an A300 Pulsemaster Stimulator connected with two A360 Constant Current Stimulus Isolators, to which the animals were connected with a six-channel commutator (Plastics One Inc.) and shielded cables. Each rat was first stimulated continuously for 20 min. Thereafter, the stimulation was interrupted, and the behavioral and EEG seizure activity of the animal was observed for 60 s. If the behavior of the animals indicated the presence of epileptic activity (head nodding or limb clonus), observation was continued for ≤5 min. If an animal did not meet the criteria of clonic SE (continuous EEG epileptiform spiking and recurrent clonic seizures), stimulation was repeated, and behavior was checked again after 5 min. Once the criteria of SE were achieved, no further stimulation was given.
EEG was monitored continuously for 48 h via cortical electrodes by using the Nervus EEG Recording System (Taugagreining, Iceland) connected with an ISO-1032 amplifier (Braintronics, Almere, the Netherlands) SVT-S3000P Hitachi Time Lapse 168 VCR (Japan), and a Panasonic WV-CL350 Video Camera (Japan). The combined video-EEG monitoring system allowed simultaneous display of the EEG signal and video-image on the screen. The amygdaloid stimulation electrodes were removed under halothane anaesthesia 48 h after stimulation, and the electrode guide cannula was sealed with dental acrylate.
Of the 17 rats stimulated, four died after SE. The remaining 13 rats were included in the 6-month video-EEG and MRI analyses, from which five rats died while anesthetized during a MRI session. Complete 6-month MRI follow-up was available from eight epileptic rats. To control for the effect of the damage produced by the electrode tract on the MRI parameters, three electrode-implanted unstimulated controls were imaged in parallel with the stimulated rats at 73 ± 3 days and at 271 ± 1 days from the beginning of the study (after surgery). In addition, eight age-matched (238 days old, at the end of the study at the time of the last MRI imaging) Wistar rats were included in the study as MRI controls. Electrode implantation alone does not induce epileptogenesis (15).
Analysis of the severity and duration of SE
A conspicuous feature of EEG activity during SE was the occurrence of high-amplitude and frequency discharges (HAFDs), which were typically associated with behavioral seizures. An HAFD was defined as high-amplitude (>2 × baseline) and high-frequency (>8 Hz) discharge in the cortex that lasted for ≥5 s. To assess the severity of SE, the number of HAFDs in each animal was counted. The duration of SE was defined as the time interval between the first and last HAFD.
Video-EEG monitoring and analysis of the appearance of spontaneous seizures
To detect the occurrence of spontaneous seizures, rats were monitored continuously with a video-EEG system on five occasions. The first recording was performed at day 2 after SE, and thereafter at 28 days [continuous video-EEG monitoring (24 h/day) for 7 days], 42 days (for 11 days), 4 months (for 4 days), and 8 months (for 7 days) after SE (Fig. 1).
EEG signals were recorded with a Stellate EEG Monitor System (sampling rate, 200 Hz; high-pass filter, 1 Hz; and low-pass filter, 100 Hz) that was connected with two ISO-DAM 8 amplifiers (WPI, Sarasota, FL, U.S.A.) and to which the animals were connected with a six-channel commutator and shielded cables. This system allowed the animals to move freely without twisting the cables. Behavior was recorded by using a WV-BP312E Video Camera (Panasonic) that was positioned above the cages and connected with an SVT-S3000P Time Lapse 168 VCR (Sony) and PVC-145E Video Monitor (Sony). The video system was connected to the EEG recording system via a time-code generator (MUL; TIM Electrode Inc.). Type 955 Infra Red Light source (Videmech Ltd., Sandhurst, U.K.) was on at night to allow simultaneous video monitoring. A wide-angle lens allowed videomonitoring of up to eight to 10 animals. The manifestation of seizure activity was analyzed on DDS2-files and video tapes.
Each EEG file was analyzed manually by browsing the EEG on the computer screen. If an EEG seizure was observed, its behavioral severity was analyzed from the corresponding video-recording (see later). An EEG seizure was defined as a high-frequency (>5 Hz), high-amplitude (>2 × baseline) discharge in the contralateral frontal cortex that lasted for ≥5 s.
The severity of behavioral seizures was scored according to a slightly modified Racine's scale (16): score 0, electrographic seizure without any detectable motor manifestation; score 1, mouth and face clonus, head nodding; score 2, clonic jerks of one forelimb; score 3, bilateral forelimb clonus; score 4, forelimb clonus and rearing; score 5, forelimb clonus and rearing and falling.
MRI during epileptogenesis and epilepsy
Quantitative MRI measurements were performed after video-EEG recordings. The first MRI session was 2 days after SE, followed by 9, 23, 53, 84, 113 to 115, 152 to 154, and 174 to 176 days after SE (Fig. 1). All MRI measurements were performed in a horizontal 4.7-T magnet (Magnex Scientific Ltd, Abington, U.K.) equipped with actively shielded imaging gradients (Magnex) interfaced to a Varian UNITYINOVA console (Palo Alto, CA, U.S.A.). A linear birdcage volume coil (diameter, 40 mm) was used in transmit–receive mode. A coronal 1-mm slice (field of view, 35 mm; matrix size, 128 × 256) was set according to pilot images so that the center of the slice was 2.8 mm from bregma. T1ρ was quantified by using four variable-length (10–70 ms) adiabatic spin-lock (SL) pulses with B1 of 0.7 gauss followed by a crusher gradient, in front of a fast spin echo imaging sequence (TR = 2.5 s, echo spacing = 10 ms, 16 echoes/excitation, 4 averages). The on-resonance SL pulse consisted of a hyperbolic secant adiabatic half passage (AHP) followed uninterruptedly by the SL period and a second AHP pulse, which returned magnetization back to the z-axis. The trace of the diffusion tensor (Dav) images was obtained by using the method of Mori and van Zijl (17), incorporated in a spin-echo sequence (TR = 1.5 s, TE = 55 ms, b-values: 0, 470, 856 s/mm2). The method uses four pairs of bipolar gradients with diffusion time of 4.8 ms, yielding the trace images (Dav= 1/3TrD) in a single scan. T2 was quantified by using a multiecho sequence (TE = 20, 40, 60 ms, TR = 1.5 s, 4 averages/line).
Absolute T2, Dav, and T1ρ images were calculated on a pixel-by-pixel basis. Nine regions of interest (ROIs) were analyzed as follows: the piriform cortex bilaterally, the amygdala bilaterally, the hippocampus bilaterally, and the thalamus (Fig. 2). In addition, we analyzed the parietal cortex bilaterally as a control area. All values are expressed as mean ± standard deviation (SD). Tissue thickness or size of the anatomic area was determined from diffusion (TE = 55 ms, b = 856 s/mm2) images, by measuring either the width of a given region from a slice (amygdala) or outlining of the region (hippocampus), respectively.
The rats were transcardially perfused for histologic analysis 265 days after induction of SE. The animals were deeply anesthetized and perfused according to the Timm fixation protocol: 0.37% sulfide solution (30 ml/min) for 10 min followed by 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4 (30 ml/min), +4°C, for 10 min. The brain was removed from the skull and postfixed in buffered 4% paraformaldehyde for 4 h, and then cryoprotected in a solution containing 20% glycerol in 0.02 M potassium phosphate–buffered saline for 24 h. The brains were then blocked, frozen in dry ice, and stored at –70°C until cut. The brains were sectioned in the coronal plane (50 μm, 1-in-5 series) with a sliding microtome. The sections were stored in a cryoprotectant tissue-collecting solution (30% ethylene glycol, 25% glycerol in 0.05 M sodium phosphate buffer) at –20°C until processed. Adjacent series of sections were used for Nissl, Timm, and Fluoro-Jade staining.
To identify the cytoarchitectonic boundaries, the distribution and severity of neuronal damage, and the location of the stimulating electrode, the first series of 1-in-5 sections was stained for thionin. The severity of neuronal damage in the various temporal lobe structures was scored as follows: score 0, no damage; score 1, ≤50% neuronal loss; score 2, ≥50% neuronal loss.
Fluoro-Jade B staining
An adjacent series of sections was stained with Fluoro-Jade B to examine ongoing damage according to the protocol described by Schmued (18).
Synaptic reorganization (mossy fiber sprouting) was analyzed from sections stained with the Timm sulfide/silver method (19). For staining, all coronal sections (30 μm, 1-in-5 series) including the hippocampus were mounted on gelatin-coated slides and dried at +37°C. Staining was performed in the dark according to the following procedure: the working solution containing gum arabic (300 g/L), sodium citrate buffer (25.5 g/L citric acid monohydrate and 23.4 g/L sodium citrate), hydroquinone (16.9 g/L), and silver nitrate (84.5 mg/L) was poured into the staining dish. The sections were developed until an appropriate staining intensity was attained (60–75 min). The slides were then rinsed under tap water for 30 min and placed in 5% sodium thiosulfate solution for 12 min. Finally, sections were dehydrated through an ascending series of ethanol, cleared in xylene, and cover-slipped with DePeX mounting medium (BDH, Laboratory Supplies, Dorset, U.K.)
Mossy fiber sprouting was analyzed from the tip, mid, and crest portions of the granule cell layer of the septal hippocampus, which corresponds to the region of MRI analysis. The density of mossy fiber sprouting was scored according to Cavazos et al. (20): score 0, no granules, score 1, sparse granules in the supragranular region and in the inner molecular layer; score 2, granules evenly distributed throughout the supragranular region and the inner molecular layer; score 3, almost a continuous band of granules in the supragranular region and inner molecular layer; score 4, continuous band of granules in the supragranular region and the inner molecular layer; score 5, confluent and dense laminar band of granules that covers most of inner molecular layer, in addition to the supragranular region. For each section, mean sprouting scores in the tip and mid portions were calculated. Thereafter, mean Timm scores obtained in the tip, mid, and crest potions of all sections from the septal, dorsal-mid, or ventral-mid portions of the hippocampus were calculated separately and used in the statistical analysis.
Photomicrographs were taken with a Leica DM RB microscope equipped with a camera system. Low-power photomicrographs were taken with a Nikon 6 × 9-cm system.
Data were analyzed by using Excel and SPSS for Windows. Changes in seizure numbers between different animal groups [mild epilepsy (≤1 seizure/day) and severe epilepsy (>1 seizure/day)] were analyzed by using the Mann–Whitney U test. Changes in seizure frequency over time were analyzed by using the Pearson χ2 test. Changes in the proportions of seizure types between animal groups were evaluated by using the Mann–Whitney U test. A p value of <0.05 was considered statistically significant.
In total, 25 rats were included in the study: five normal controls without any treatment, three unstimulated controls with electrodes implanted in the amygdala, and 17 stimulated animals with electrodes implanted in the amygdala. Four of the 17 stimulated rats died within 24 to 48 h after induction of SE and were excluded from the final analysis. The progression of structural changes preceding the epilepsy diagnosis was followed with MRI in 10 of 13 rats (three rats died before spontaneous seizures were detected in video-EEG recordings; see Fig. 3). Of the 13 rats, eight survived the 8-month follow-up period and also were used for histologic analysis.
Characteristics of SE after electrical stimulation
Data were analyzed from all the animals that survived until epilepsy was diagnosed, based on EEG. The mean HAFD number during SE was 296 ± 145 (range, 151–441; median, 258.4; n = 10). The mean HAFD duration was 13.6 ± 4.2 s (range, 9.4–17.8 s; median, 12.9; n = 10). The mean sum of HAFD duration was 48 ± 19 min (range, 29–67 min; median, 43 min; n = 10). The mean duration of SE (time between the first and last HAFD) was 9 h 8 min ± 2 h 34 min (range, 6 h 34 min to 11 h 43 min; median, 10 h 12 min; n = 10).
No difference was found in the number or total duration of HAFDs between groups when the severity of SE was compared between animals with mild and severe epilepsy (see later).
Epileptogenesis and occurrence of spontaneous seizures
Analysis of spontaneous seizures in video-EEG recordings of the 10 rats that were examined with MRI at least until they developed spontaneous seizures indicated that six rats had spontaneous seizures at 28 days (performed between days 28 and 35) after SE, and all 10 rats had spontaneous seizures at 52 days (days 42–52).
At 28 days, the number of daily spontaneous seizures varied between 0.5 and 25.7 (mean, 13.1 ± 12.6; median, 7.9; n = 10) and at 52 days between 0.1 and 32 per day (mean, 16 ± 16; median, 7.1; n = 10). Fifty-two days after SE, four rats had mild disease (<1 seizure/day), and six had severe disease (≥1 seizure/day). One rat with mild disease, however, later developed several daily seizures and was therefore included in the group of animals with severe disease in the final analysis.
The mean score indicating the behavioral severity of spontaneous seizures was 3.9 ± 1.1 (range, 2.8–5.0; median, 4.2; n = 6) at 28 days and 2.9 ± 1.9 (range, 1.0–4.8; median, 3.5; n = 10) at 52 days.
In two rats (R-4; R-9), a clear change in seizure type was seen during the follow-up. During the first weeks, all spontaneous seizures were generalized, whereas at later times, most of seizures were partial. Both of these rats had a very high seizure frequency (17.1 and 19.5 seizures per day, respectively; see Fig. 3).
Multiparametric quantitative MRI
Quantitative MRI follow-up was performed in 10 rats. Two of the 10 rats died during the follow-up (one at 53 days and another at 84 days) and were not included in the relaxation time or diffusion results.
The normal T2 relaxation times measured from control animals were 55.3 ± 6.3 ms (piriform cortex), 54.0 ± 3.6 ms (amygdala), 44.0 ± 2.2 ms (thalamus), and 49.2 ± 3.0 ms (hippocampus). The corresponding T1ρ relaxation times and Dav values were 83.4 ± 3.6 ms and 83.0 ± 13.0 mm2/s (piriform cortex), 81.1 ± 2.9 ms and 84.5 ± 10.8 mm2/s (amygdala), 71.3 ± 2.2 ms and 78.2 ± 13.0 mm2/s (thalamus), 75.4 ± 2.0 ms and 78.8 ± 9.5 ×10-6 mm2/s (hippocampus), respectively. The relative changes after stimulation compared with values from control animals are given later.
The structures, both ipsilateral and contralateral relative to the stimulation site, including amygdala, piriform cortex, and thalamus, had a substantial (>26%) increase in both T2 and T1ρ 2 days after stimulation (p < 0.001; Fig. 4). At the first MRI time point, Dav was elevated in both ipsilateral and contralateral amygdalae (p < 0.05), but remained within the normal range in the other brain structures studied. In hippocampus, the only MRI abnormality was elevated T1ρ on the ipsilateral side (p < 0.05).
The MRI parameters, measured from brain regions 2 days after SE, returned to control levels, except for T1ρ (p < 0.05), which remained slightly elevated (114%± 2% of control) in contralateral piriform cortex until day 53 (see Fig. 4.). Assessment of the time-dependent changes of the MRI variables revealed a secondary response in several brain regions starting at days 23 and 53 after stimulation, which coincides with the beginning of the spontaneous seizures. In ipsilateral amygdala, Dav was elevated at days 23 and 84 (p < 0.05), whereas on the contralateral side, T1ρ was increased after day 53 (p < 0.05). In the ipsilateral hippocampus, T1ρ was increased 23 and 53 days after stimulation (p < 0.05), and at the latter time point, Dav was also abnormally high. In the thalamus, only MRI abnormality after acute phase was decreased T2 in the last time point (day 114, p < 0.01). In the piriform cortex, no MRI changes were detected after day 53, except a decrease in T2 at day 114 (p < 0.01). Because of severe atrophy in the piriform cortex, however, the last few data points might have been influenced by an increased partial voluming effect from CSF.
The absolute T2, T1ρ, or Dav MRI did not significantly correlate with epilepsy severity (seizure frequency, that is, number of seizures/day), amygdaloid or hippocampal neuronal damage, or mossy fiber sprouting, as assessed 8 months later.
Volume analysis by MRI
The thickness of the amygdala + piriform cortex on the stimulation side was decreased at all times after 9 days, relative to control values, determined from both control Sprague–Dawley and Wistar rats (no difference between the rat strains, data not shown). A more slowly progressing loss of tissue was seen in the contralateral amygdala (Fig. 5). Large animal-to-animal variation occurred in the volumes of hippocampi and parietal cortices during follow-up, and no significant changes were noted.
In all epileptic animals, the amygdala (lateral, basal, and accessory basal nuclei), the piriform cortex (layers II–III), and the thalamus (the mediodorsal nucleus) were severely (score >2.0) damaged at levels that corresponded to the MRI analysis. In the septal hippocampus, the granule cell layer was mildly damaged in one of eight rats. The hilus showed severe neuronal loss (score > 2.5) in four and mild (score < 0.5) in two animals. In two rats, no hilar damage was observed. The CA3 pyramidal cell layer was damaged in five of eight rats (score ≤2.0) and the CA1 pyramidal cell layer in two of eight rats.
A high correlation was found between neuronal cell damage and mossy fiber sprouting in the dentate gyrus, as assessed by histology 265 days after SE. MRI parameters did not correlate with hippocampal histology. The lack of correlation might be explained by the fact that damage was limited to selective subfields of the hippocampus, too-large ROIs, or too-low resolution for detection at these microscopic changes in vivo.
In one of eight epileptic animals, a large number of Fluoro-Jade B–positive cells were found with neuronal morphology in the granule cell layer as well as in the CA3a and CA3c regions (MRI-8). The labeled cells were located at the level of the hippocampus at which the septal and temporal ends become fused in coronal sections, corresponding to the caudal two thirds of the hippocampus (i.e., caudal to the slice imaged with MRI). In the remaining seven rats, labeling was seen in cells with a glial appearance, which were typically located in layer II of the piriform cortex, as well as in the ventral and dorsal intermediate subfields of the entorhinal cortex.
Mossy fiber sprouting
At the level of the septal hippocampus corresponding to the MRI analysis, the mean density of mossy fiber sprouting in the ipsilateral tip portion of the granule cell layer was 3.7 ± 1.4 (range, 1.8–5.0), in the mid portion 3.2 ± 1.9 (range, 1.0–5.0), and in the crest 3.4 ± 1.5 (range, 1.5–4.9). The mean sprouting (mean of tip, mid, and crest) was 3.5. Analysis of the data from each rat individually indicated that the density of mean mossy fiber sprouting was increased in all epileptic animals relative to controls (p < 0.05). Sprouting density was similar ipsilaterally and contralaterally. No difference was noted between the animals with rare or frequent seizures. The mossy fiber sprouting density correlated with the severity of hilar cell damage both ipsilaterally and contralaterally.
These results support the hypothesis that the neurodegeneration process continues long beyond SE-associated seizure activity. A low correlation was found, however, between the severity of damage as assessed by multiparametric MRI in the acute phase and the histologic outcome of the tissue in a long-term follow-up.
Dynamics and pathologic correlates of SE-induced MRI alterations
Pathologic tissue correlates of acute cell damage and MRI parameters are best understood in stroke. Therefore we compared the dynamics and cellular changes in SE-induced MRI alterations with those reported after acute stroke.
In the acute phase after SE, a widespread increase occurred in both T2 and T1ρ. This increase was followed by normalization, and within the next few weeks, by a secondary increase in Dav and T1ρ in several brain areas that are mono/polysynaptically connected to the stimulated amygdala. This same kind of reversal with a much shorter time frame is observed after transient ischemic attack, in which an initial diffusion decrease is followed by normalization during reperfusion and secondary damage. The reversal of MRI changes after SE was somewhat unexpected, however, because increased diffusion is generally observed at a relatively late irreversible phase in stroke. The increased T1ρ was reversible after SE. These data suggest that interpretation of the tissue status based on changes in MRI parameters cannot be directly adopted from stroke studies.
The increase in T2 in the amygdala, thalamus, and piriform cortex at 2 days after SE is consistent with previous reports of hyperintensity in T2-weighted MRI during bicuculline (21) or flurothyl-induced (22) prolonged seizures, as well as 1 h to 7 days after kainic acid (23) or 3 to 24 h after pilocarpine-induced SE (24). Interestingly, hippocampal T2 remained unchanged throughout the follow-up in the present study. In kainate and pilocarpine models, the hippocampal T2-weighted signal increase is detected 3 to 24 h after SE (25). The difference might be related to the milder overall damage of hippocampus in the present model compared with that in the two pharmacologic models of epilepsy. T2 predominantly reflects tissue water content, and it is likely that edema in the brain structures after prolonged SE is the key factor underlying the MRI observations, including T1ρ.
Post-SE edema, together with reorganization of the extracellular matrix and alterations in the intracellular organelles, are the most likely explanations for the elevated T1ρ in the subacute phase after SE. Whereas edema formation influences both relaxation times, T1ρ MRI is thought to be a more sensitive to factors related to the macromolecular pool, which might explain subacute changes in T1ρ.
Several previous studies reported reduced water diffusion either during (21) or within the first few hours after SE induction (25) in several brain areas, including the amygdala, piriform cortex, and thalamus. Findings are similar to the behavior of diffusion during acute stroke and are likely to be associated with a change in water compartmentalization. We did not image rats earlier than 2 days after SE, when Dav was already elevated in the stimulated amygdala but not in the mono/polysynaptically connected brain areas. The increased diffusion at this time point contrasts with reports on acute seizures (26) or spreading depression showing reduced diffusion (27). It is important to note that the diffusion increase is observed only in brain structures developing the most severe damage over time. It might be that during acute SE, these structures have an immediate energy failure, a tissue status resembling that observed several days after onset of cerebral ischemia in rat (28). Thus normalization of diffusion (or relaxation times) in the present model of epilepsy should not be interpreted to indicate a good prognosis, as the cellular and/or physicochemical processes leading to an altered water apparent diffusion constant might initiate tissue-damaging cascades, such as those reported in kainic acid (29) and pilocarpine (30) models of SE.
MRI reveals temporal patterns of long-lasting abnormality after SE in selected brain areas
A challenge for quantitative MRI is to reveal pathologic processes preceding disease onset and atrophy during epileptogenesis. In the present model, massive amygdaloid atrophy was associated with robust MRI changes within a few days after SE, that is, in the early phase of epileptogenesis. Conversely, hippocampal Dav was normal at the acute phase but elevated bilaterally ∼2 months after SE. Further, hippocampal T1ρ remained increased for ∼2 months. Recently Roch et al. (30) reported a progressive increase in the hippocampal T2 starting 2 days after pilocarpine-induced SE and continuing for 9 weeks. Despite differences in the models used, the changes in the hippocampal MRI were delayed and progressive. In the present model, pyramidal cell loss in the septal hippocampus was mild at the acute phase and continued to progress for ∼2 months (3), an observation consistent with the present MRI data. As the present study indicates, hippocampal degeneration can last for ≤8 months.
The differences in the timing of MRI abnormality in the amygdala and the hippocampus might reflect the different kinetics of SE-induced structural damage in these regions. The amygdala was the site for induction of SE and therefore becomes damaged bilaterally soon after induction of SE. Further, the various amygdala nuclei have a different sensitivity and rate of damage to SE-induced neurotoxicity (3,31). Therefore the finding of an elevated Dav in the amygdala even 2.5 months after SE was unexpected. As in the amygdala, measurement of the hippocampal thickness revealed progressive atrophy at the time period of the Dav increase (≤53 d). Interestingly, the hippocampal area then increased toward normal values, probably related to the constant circuitry reorganization that occurs during epileptogenesis, including axonal sprouting, gliosis, or neurogenesis (for review, see 31).
Damage after the appearance of spontaneous seizures
All animals expressed spontaneous seizures by day 53, and at this point, quantitative MRI revealed hippocampal T1ρ and Dav abnormalities. Two plausible explanations exist for the MRI changes in the hippocampus: first, MRI changes are associated with irreversible structural changes due to the damage processes in the tissue in a similar manner as in kainate (32) or pilocarpine (24) models. In addition, MRI abnormalities might be caused by partially reversible changes in water homeostasis due to recurrent spontaneous seizures. The increasing deviation of MRI parameters at later times and the weak correlation between MRI variables and histology suggest a significant role of the latter mechanism. Assessment of ongoing neuronal damage 8 months after SE showed Fluoro-Jade-B–positive cells in the CA3 of the ventral hippocampus in only one of eight rats, even though all animals had recurrent seizures. Therefore consistent with a previous pilot study (33), these data support the view that progressive damage is not associated with recurrent seizures.
Initial edema, as assessed by MRI, did not predict the severity of histologic damage or epilepsy
MRI is a tool for defining surrogate markers to predict the clinical outcome after brain-damaging insults. Here we addressed the issue of whether early MRI signal abnormalities predict the severity of epilepsy or histologic damage at the chronic phase. We attribute the early MRI abnormalities predominantly to edema formation in the tissue, with possible microstructural alterations. One key observation is that in the acute phase, the extent of this edema in the amygdala does not predict the long-term progression of epilepsy as assessed by seizure frequency. No correlation occurred between the MRI-detected edema and neuronal damage or mossy fiber sprouting at 8 months. Recently Roch et al. (30) reported that the T2 signal increase in the piriform and entorhinal cortices at 24 h predicted the occurrence of epilepsy after pilocarpine-induced SE in 21-day-old rats. They did not, however, correlate the severity of damage with epilepsy severity. In the present study, only the animals with a long-lasting SE were included in the analyses, and therefore an association of milder damage with the development of epilepsy was not assessed.
These data demonstrate the dynamic behavior of brain structure characteristic after induction of SE. The partial reversal of the MRI relaxation times in amygdala after SE indicates that the amygdala can recover from the initial insult, yet atrophy develops over time. Pathologic T1ρ and increased Dav at subsequent time points indicate progression of the damage. The weak correlation between MRI and the severity of disease progression is consistent with a lack of correlation with histologic studies and severity of epilepsy. Our data indicate that epileptogenesis involves the hippocampus later in the disease process. The MRI data suggest that cellular alterations in these structures are not parallel to the clinical manifestations of TLE, and MRI-detected tissue alterations during early epileptogenesis might not predict the severity of epilepsy. From the therapeutic point of view, these data stress the need for neuroprotective treatment in the treatment regimen of SE, even after discontinuation of the electrographic seizure activity.
Acknowledgment: This study was supported by grants from Sigrid Juselius Foundation, Academy of Finland, Centre of International Mobility, and the Research Foundation of Leiras.