Address correspondence and reprint requests to Dr. T. O'Brien at The Department of Medicine, Royal Melbourne Hospital, The University of Melbourne, Parkville, VIC, Australia 3050. E-mail: firstname.lastname@example.org
Present address of Dr. Tesiram: The Free Radical Biology and Aging Research Program, Oklahoma Medical Research Foundation, Oklahoma City, OK, U.S.A.
Summary: Purpose: The rat electrical amygdala kindling model is one of the most widely studied animal models of temporal lobe epilepsy (TLE); however, the processes underlying epileptogenesis in this model remain incompletely understood. Magnetic resonance imaging (MRI) is a powerful method to investigate epileptogenesis, allowing serial imaging of associated structural and functional changes in vivo. Here we report on the results of serial MRI acquisitions during epileptogenesis in this model.
Methods: Serial T2-weighted MR images were acquired before, during, and after the induction of kindling, to investigate the development and progression of imaging abnormalities.
Results: T2-weighted acquisitions demonstrated the development of regions of increased signal in the rostral ipsilateral regions of CA1 and dentate gyrus in kindled (five of seven) but not in control rats (p < 0.05). Quantification of the T2 signal demonstrated a significant increase in kindled animals when compared with controls, 2 weeks after kindling ceased, in the ipsilateral hippocampus and the hippocampal sub regions of CA1 and the dentate gyrus (p < 0.05). No significant difference was observed in hippocampal volumes between kindled or control animals at any of the times.
Conclusions: The results of this study validate a method for acquiring serial MRI during amygdala kindling and demonstrate the induction of T2 signal abnormalities in focal regions of the hippocampus. These regions may be important sites for the neurobiologic changes that contribute to epileptogenesis in this model.
The rat electrical amygdala kindling model is a well-studied and accepted animal model of acquired temporal lobe epilepsy (TLE) (1). Amygdala kindling is considered to be a model of both seizures and epileptogenesis and mimics many of the histopathologic characteristics of acquired TLE (2). One advantage for investigating mechanisms of epileptogenesis in this model over others of limbic epileptogenesis is that it lacks major cell loss in the hippocampus (3,4). For this reason, this model equates best with the subset of TLE patients with minimal or no hippocampal atrophy. The amygdala kindling model, therefore, offers the ability to study the pathophysiology of TLE with minimal impact from the confounding effects of cell loss. Additionally, the controlled and relatively slow induction of epileptogenesis provides appropriate time windows to sample serially the neurobiologic changes that accompany this process (5).
Magnetic resonance imaging (MRI) is a powerful diagnostic and research tool for in vivo studies in biology and medicine, including epilepsy. There are a wide array of MRI studies of small-animal models of epilepsy (6), but the majority of these have focused on the acute effect of seizures and early effects after status epilepticus (SE) (7,8), with only a few conducting long-term studies during the process of epileptogenesis (9,10). The post-SE models are all associated with widespread cell loss and tissue destruction in the limbic structures, and, in keeping with this, prominent changes are seen in these areas in MRI. These changes, however, may not necessarily be directly related to epileptogenesis. Increasing data suggest that cell loss may not be critical to the development of epilepsy. It is noteworthy that even when hippocampal cell loss is prevented in models of SE with the administration of neuroprotective agents, epileptogenesis still occurs (11). It also has been shown that apparent cell loss in the hippocampus occurs independent of the occurrence of spontaneous recurrent seizures in extended amygdala kindled rats (12).
No studies to date have used MRI to investigate the amygdala kindling model and thus investigated the induction of imaging changes in a model of TLE without major tissue damage. One reason for this is the practical difficulties associated with obtaining electrodes for kindling stimulations that are both MRI compatible and robust enough to allow multiple stimulations to be performed over a period of several weeks. The majority of electrodes used for electrical kindling have been constructed from steel, which disrupts field homogeneity, resulting in distortion of MR images.
The present study uses serial T2-weighted MR images acquired before, during, and after kindling to investigate the imaging changes during epileptogenesis in a model of TLE that is associated with minimal cell loss or tissue destruction.
Fourteen 14- to 16-week-old male Wistar rats were implanted with custom-made gold bipolar electrodes that had previously been shown to not affect the quality of MR images (personal observations). One week after surgery, seven animals received stimulations twice daily, at least 4 hours apart, 6 days per week (200-μA, 60-Hz, 1-ms bipolar square-wave pulses for 1 sec) for 4 weeks (total, 48 stimulations) according to previously established methods (1). Animals were not stimulated for at least 12 hours before imaging or on the day of imaging. Correct placement of the stimulating electrode in the amygdala was confirmed by MRI during the first imaging session (1 week after implantation). Seven control animals underwent identical surgeries, handling, and imaging, but were not electrically stimulated.
The study and all procedures were approved by the Animal Ethics Committees of The Ludwig Cancer Institute/Department of Surgery, Royal Melbourne Hospital and the Howard Florey Institute. All procedures were performed according to the guidelines set by the Australian Code of Practice for the Prevention of Cruelty to Animals.
MR image acquisition
MR images were acquired on a 4.7-T Bruker Biospec 47/30 Avance small-animal spectrometer (Ettlingen, Germany) by using a shielded-gradient set (Bruker Biospec) appropriate for rats. Radiofrequency (RF) pulse transmission and MR data acquisition were performed by using a 72-mm inner-diameter birdcage coil (Bruker Biospec) optimally tuned to the 1H frequency. T2-weighted axial structural images were obtained over 15 adjacent 1-mm-thick slices by using a fast spin-echo sequence (TA, 298 s; TR, 3.1 s; TE, 67.5 ms; MTX, 256 × 256; NA, 3; FOV, 6 cm; rare factor, 8). Animals were scanned under anesthesia in the prone position in a custom-built Plexiglas holder to ensure consistent positioning of the animal. Rats were anesthetized with 5% isoflurane in 1:1 air/oxygen and then maintained on 1.5–2.5% isoflurane for the remainder of the experiment. Images were collected by using Paravision 3.0 (Bruker Biospec).
Four sequential MR image acquisitions were acquired for each of the 14 animals, one every 2 weeks for 6 weeks: (a) 1 week after electrode implantation, before the first electrical stimulation, (b) 2 weeks later after 24 electrical stimulations, (c) 4 weeks later after the completion of the 48 electrical stimulations, and (d) 2 weeks after the final stimulation. Control and kindled animals underwent identical imaging procedures.
MR image analysis
A blinded reviewer qualitatively visually reviewed the MRI images. The reviewer was asked to determine whether focal increases occurred in T2 signal in the hippocampus for each image, and if so, to localize the hippocampal subregion involved. The images were initially viewed in the axial plane, but any regions with focal increased T2 signal were confirmed by review of the images reformatted into the horizontal and sagittal planes.
T2 signal intensity and hippocampal volumes were quantified with the aid of image-analysis software, Analyze (Mayo Foundation, Rochester, MN, U.S.A.) by a blinded operator. Regions of interest (ROIs) were drawn around left and right hippocampus and two hippocampal subregions (dentate gyrus and CA1 region) on each slice, based on regions defined in Paxinos and Watson (13) (Fig. 1A). Left-to-right ratios were calculated for each ROI. Total left, right, and left-to-right hippocampal volumes also were quantified.
Statistical analysis was performed by using the software package Statistica (Statsoft, Tulsa, OK, U.S.A.). The number of animals at each imaging session in whom qualitative visual analysis detected focal T2 signal change was assessed for a statistically significant difference between kindled and control animals by using the Fisher's exact test. Quantified T2 signal intensity in each hippocampus and respective subregions, and hippocampal volume measurements, were assessed for statistical significance between kindled and control animals by using analysis of variance (ANOVA) for repeated measures, with planned comparisons analysis used to compare the groups for differences at each of the imaging sessions. For imaging sessions in which a significant difference was found between kindled and control animals, a planned comparison analysis also was performed to assess statistical significance between individual image slices (i.e., rostral to caudal).
Qualitative assessment of T2-weighted MR images during amygdala kindling
Five of the seven kindled rats were found to show focal regions of increased T2 signal in the hippocampus ipsilateral to the site of stimulation on the final MR acquisition, 2 weeks after kindling (Fig. 1D). One of these rats showed bilateral changes, with the right hippocampus showing changes similar to those in the left. These areas of visually apparent increased signal were restricted to the most rostral regions of CA1 and dentate gyrus. The earliest increased T2 signal was seen was after 24 stimulations (one of seven); changes also were first seen in two further fully kindled animals at the third imaging acquisition (Fig. 1C). In these cases, the T2 signal changes intensified further on the subsequent MR imaging acquisitions (compare Fig. 1C and D). No changes were seen in the hippocampus in any of the control animals (p < 0.05, Fisher's exact test, 2 weeks after kindling). These results are summarized in Table 1.
Table 1. Summary of T2 signal changes, mean number of stimulations and class V seizures in control and kindled rats
Control Increased T2 signal
Kindled Increased T2 signal
Mean number of stimulations
Mean number of class V seizures
ap < 0.05, Fisher's exact test.
bOne rat died at session three, so only six rats were imaged in session four.
Quantitative assessment of hippocampal T2 signal and volume during amygdala kindling
Quantification of T2 signal supported the results of the qualitative visual analysis of the MR images at the final image acquisition (Fig. 2). T2 signal was significantly increased in the ipsilateral hippocampus, CA1, and dentate gyrus at the final imaging acquisition in kindled rats (5%, 5%, and 3% increase, respectively, when compared with control values, p < 0.05, ANOVA for repeated measures) (Fig. 2A–C). When analysis of the increase in T2 signal was performed for each of the individual slices to assess the rostral–caudal extent of the changes, all slices showed an increase in kindled when compared with control animals but it was significant only for the most rostral slice (p < 0.05, repeated measures ANOVA with planned comparisons) (Fig. 3).
No significant difference was observed between kindled and control animals for left or right volume measurements for the hippocampus, nor for the left-to-right volume ratios at any of the times investigated (p = 0.6, 0.2, and 0.3 at the final imaging session, repeated measures ANOVA with planned comparisons, respectively) (Fig. 2D).
Kindling induces T2-weighted signal change
This serial imaging study demonstrates the novel finding that amygdala kindling can induce increases in T2 signal intensity in focal regions of the hippocampus. These areas were restricted to the rostral subregions of CA1 and the dentate gyrus and were most prominent ipsilateral to the site of stimulation. The hippocampal changes in T2 signal were seen maximally 2 weeks after the cessation of kindling, but three of the seven animals showed apparent increases during kindling, which then further intensified over time. Quantification of T2 signal confirmed the findings of the qualitative visual assessment at the final time point, 2 weeks after the cessation of kindling, demonstrating a significant increase in T2 signal for the kindled versus control animal groups. However, the quantitative analysis did not demonstrate any significant difference between the groups at the earlier time points. This is likely because the signal changes at these earlier times detected on the visual qualitative analysis were present in only a minority of animals (one of seven at 2 weeks and three of seven at 4 weeks). The failure to detect signal changes in the other kindled animals at these times indicates that considerable interindividual variability occurs in their development. It also is possible that the changes in the “MRI negative” kindling animals may have been below the sensitivity of the imaging system to detect.
The observed significant changes in T2 signal were not accompanied by any changes in hippocampal volume. This increase in T2 signal in the hippocampus observed in response to kindling is consistent with previous studies in models of SE (7,9,14–16); however, these increases are more regionally restricted in the kindling model. These results provide the first evidence for imaging changes in the hippocampus in response to kindling, where relatively little tissue destruction occurs in contrast with the post-SE models. Furthermore, as the changes are seen in the hippocampus, remote from the site of the electrical stimulations (i.e., the amygdala), the changes likely result from an effect of secondary seizure spread on the hippocampus rather than a direct effect of implantation and simulation.
T2-weighted signal change: cell loss and seizures
Changes in T2-weighted images represent changes in free-water content within the brain; for this reason, any increase in signal is thought to reflect structural changes that disrupt water homeostasis, such as gliosis, edema, and neuronal loss. Previous studies have found that the majority of the histologic changes in the amygdala kindling model, although mild when compared with other models, are focused in the hippocampus. These changes include cell loss (17), synaptic reorganization (18), mossy fiber sprouting (19), and gliosis (20). It is possible that the increase in hippocampal T2 signal observed during the development of kindling in a subset of animals reflects the evolution of these structural changes in this region. The time course observed for the development of these changes does not correlate with those previously observed in histologic studies. The majority of the kindled animals showed an increase in T2 signal in the ipsilateral hippocampus 2 weeks after kindling, whereas most of the histologic effects of kindling become apparent after between 1 (gliosis, mossy fiber sprouting) and 4 weeks of kindling (cell loss) and have not been reported to intensify further after the cessation of kindling. This study did not investigate whether the MRI signal changes persisted long after the cessation of kindling, as animals were culled after their fourth imaging session 2 weeks later. A longer follow-up would be helpful in evaluating whether the changes represent a subacute transient response to kindling-induced cellular injury or a permanent change. Subsequent studies on another cohort of animals will examine this question.
Conflicting evidence exists as to the predictive value and correlation of MRI changes to long-term outcome in terms of histologic damage and seizure number in animal models of SE (10,14). It also has been demonstrated that T2 signal increases independent of cell loss in an animal model of febrile seizures (8). The current study demonstrated that T2 signal increases during kindling, independent of change in hippocampal volume. Previous work has demonstrated a correlation between reduced MRI hippocampal volume and pathologic cell loss in the hippocampus of TLE patients (21,22). However, conflicting evidence exists for the correlation between hippocampal volume and cell loss in animal models of TLE. Work done by Cavazos et al. (17) found that neuronal cell density is significantly reduced in the dentate hilus, CA1, CA2, and CA3 in the kindling model after at least three class V seizures without any change in total hippocampal volume. However, it has recently been demonstrated that in extended kindled rats, the observed change in neuronal cell density in the dentate hilus also is accompanied by an increase in volume in this structure, suggesting that the reduction in neuronal density is not caused by cell loss (12). This increase in volume may result from increased tissue water, which would be expected to be associated with an increase in T2 signal on MR, therefore providing a potential explanation for the findings in our study.
T2-weighted signal change: A marker for epileptogenesis?
The findings of this study provide evidence that the rostral regions of CA1 and the dentate gyrus of the hippocampus may be important areas in which the neurobiologic changes that contribute to epileptogenesis during kindling occur. Previous studies have shown both the presence and the progressive development of T2-signal change during the process of epileptogenesis in the more destructive post-SE models (7,9,14–16) However, other studies provided conflicting evidence for the role of these changes during epileptogenesis as a correlate for damage (8,10,14) and suggested that the T2-weighted signal changes may relate to processes other than histologic damage. It is possible that changes in water content and mobility may be markers for epileptogenic processes that merely contribute to histologic damage rather than act as a marker for the damage itself.
Although two animals within the group showed little or no change in T2-weighted signal, this is consistent with the findings of Roch et al. (14), who showed that after SE, T2-weighted images failed to change in some animals that became epileptic, whereas an increase in T2 relaxation times occurred in all animals that continued to have spontaneous seizures. A more sensitive method for detecting regions of epileptogenesis may need to involve multimodal image acquisitions (e.g., T2-relaxometry, diffusion, perfusion, spectroscopy) to account for any limitation in sensitivity associated with any one individual imaging parameter.
Although the electrical kindling model has been used in many studies to investigate the important neurobiologic processes in acquired limbic epileptogenesis, the mechanisms that result in the development of kindling remain poorly understood. Part of the reason for this may arise from the difficulty in determining the topographic sites important to epileptogenesis. These have proven to be difficult to assess by using traditional histologic and electrophysiologic methods (23–25). With such in vitro methods, it is difficult to determine the relative importance of the changes in these different regions in the process of epileptogenesis, within the same animal. MR imaging, in contrast, can provide three-dimensional images of the whole brain in vivo that can be acquired serially during the process of the induction of epileptogenesis. MRI therefore offers the unique potential to identify the topographic sites in which changes are developing in the brain during epileptogenesis in the living animal.
The results of this study should allow better topographic tissue targeting in future studies investigating the histologic, physiologic, and gene-expression changes important to epileptogenesis in the amygdala kindling model. In particular, tissue sampling directed on the basis of the sites identified on in vivo MR imaging may improve the specificity of gene-expression studies of epileptogenesis, which have produced conflicting and uninformative results, likely because of inclusion of tissue from functionally heterogeneous brain regions (26).
The current study reports for the first time the acquisition of serial MRI in the amygdala kindling model of TLE. The results demonstrate that amygdala kindling induces an increase in T2 signal focally in the rostral CA1 and dentate gyrus regions of the hippocampus, independent of any change in volume. These results demonstrate the induction of MRI abnormalities, as a result of repeated seizures and epileptogenesis, in a model with minimal cell damage. Changes in T2 signal in this model may represent critical sites for the neurobiologic processes resulting in acquired limbic epileptogenesis and should topographically guide tissue sampling for future in vitro studies.
Acknowledgment: This work was partially supported by a grant from the Friends of The Royal Melbourne Neurosciences and a Clinical Investigator Grant to Dr. O'Brien from The Sylvia and Charles Viertel Charitable Foundation. We thank Mr. Rink-Jan Lohman, The Department of Medicine, Royal Melbourne Hospital, for his technical assistance. We also thank Dr. David Howells, The Department of Medicine, The Austin Hospital, Melbourne, and Professor Rod Hicks and Mr. David Binns, The Centre for Molecular Imaging, The Peter MacCallum Cancer Institute, for their collaboration on this project.