Can structural or functional changes following traumatic brain injury in the rat predict epileptic outcome?

Authors


Summary

Purpose

Posttraumatic epilepsy (PTE) occurs in a proportion of traumatic brain injury (TBI) cases, significantly compounding the disability, and risk of injury and death for sufferers. To date, predictive biomarkers for PTE have not been identified. This study used the lateral fluid percussion injury (LFPI) rat model of TBI to investigate whether structural, functional, and behavioral changes post-TBI relate to the later development of PTE.

Methods

Adult male Wistar rats underwent LFPI or sham injury. Serial magnetic resonance (MR) and positron emission tomography (PET) imaging, and behavioral analyses were performed over 6 months postinjury. Rats were then implanted with recording electrodes and monitored for two consecutive weeks using video–electroencephalography (EEG) to assess for PTE. Of the LFPI rats, 52% (n = 12) displayed spontaneous recurring seizures and/or epileptic discharges on the video-EEG recordings.

Key Findings

MRI volumetric and signal analysis of changes in cortex, hippocampus, thalamus, and amygdala, 18F-fluorodeoxyglucose (FDG)–PET analysis of metabolic function, and behavioral analysis of cognitive and emotional changes, at 1 week, and 1, 3, and 6 months post-LFPI, all failed to identify significant differences on univariate analysis between the epileptic and nonepileptic groups. However, hippocampal surface shape analysis using large-deformation high-dimensional mapping identified significant changes in the ipsilateral hippocampus at 1 week postinjury relative to baseline that differed between rats that would go onto become epileptic versus those who did not. Furthermore, a multivariate logistic regression model that incorporated the 1 week, and 1 and 3 month 18F-FDG PET parameters from the ipsilateral hippocampus was able to correctly predict the epileptic outcome in all of the LFPI cases. As such, these subtle changes in the ipsilateral hippocampus at acute phases after LFPI may be related to PTE and require further examination.

Significance

These findings suggest that PTE may be independent of major structural, functional, and behavioral changes induced by TBI, and suggest that more subtle abnormalities are likely involved. However, there are limitations associated with studying acquired epilepsies in animal models that must be considered when interpreting these results, in particular the failure to detect differences between the groups may be related to the limitations of properly identifying/separating the epileptic and nonepileptic animals into the correct group.

Traumatic brain injuries (TBIs) are induced by biomechanical forces to the brain, and represent a global health concern and socioeconomic problem (Maas et al., 2008). Among the chronic disabilities associated with TBI, posttraumatic epilepsy (PTE) occurs in approximately 25% of patients with TBI (Asikainen et al., 1999; Garza & Lowenstein, 2006), and accounts for up to 20% of symptomatic epilepsy cases (Annegers et al., 1998; Garza & Lowenstein, 2006). Although TBI can induce numerous neurobiologic changes in the brain that could potentially result in PTE (D'Ambrosio & Perucca, 2004; Kharatishvili & Pitkänen, 2010; Li et al., 2011), the causal mechanisms of PTE have yet to be identified (Garza & Lowenstein, 2006; Kharatishvili & Pitkänen, 2010). Consequently, there are currently no reliable biomarkers for PTE (Dichter, 2009; Dash et al., 2010; Mishra et al., 2011; Pitkänen et al., 2011), or effective pharmaceutical treatments known to prevent its onset (Beghi, 2003; Temkin 2009; Loane & Faden, 2010). Although anticonvulsants might suppress seizures at acute stages after TBI, they do not prevent or suppress the chronic spontaneous seizures that reflect the culmination of the epileptogenic process (Beghi, 2003; Kharatishvili & Pitkänen, 2010). There is a clear need to better understand the changes evoked by TBI that may contribute to PTE (Garza & Lowenstein, 2006; Temkin, 2009). Given the limitations involved in addressing this issue in the clinical population, the use of an animal model to study PTE may be beneficial (Pitkänen et al., 2011).

The lateral fluid percussion injury (LFPI) is one of the most commonly used experimental models of TBI, and is capable of inducing physiologic, pathologic, and behavioral changes in rodents consistent with those occurring in the clinical condition (Thompson et al., 2005; Jones et al., 2008a; Liu et al., 2010). For example, previous works from our laboratory have found progressive structural, functional, and behavioral changes in rats given LFPI (Jones et al., 2008a; Liu et al., 2010). LFPI has also been shown to be a valid model of PTE, as the proportion of rats given LFPI that develop PTE is similar to the frequency reported in patients with TBI, and LFPI induces a number of cerebral changes thought to be associated with PTE (Thompson et al., 2005; Kharatishvili & Pitkänen, 2010; Pitkänen et al., 2011). Past LFPI studies have reported initial evidence that imaging techniques, such as magnetic resonance imaging (MRI), may be able to detect TBI-induced changes that predict the onset of PTE (Kharatishvili & Pitkänen, 2010; Pitkänen et al., 2011). However, these studies are limited to a small number of imaging modalities (Pitkänen et al., 2011) and may have confounding factors (Kharatishvili et al., 2007). In light of these issues and the need to better understand TBI-induced changes that may underlie or predict PTE, the current study investigated whether structural or functional abnormalities, as assessed by serial in vivo T2-weighted MRI and 18F-fluorodeoxyglucose (FDG)–positron emission tomography (PET) imaging over 6 months postinjury, were associated with the occurrence of PTE in rats that had severe LFPI. Because neurologic symptoms, such as emotional disturbances, are known to occur in patients with epilepsy and might be associated with PTE (Beyenburg et al., 2005), behavioral analyses were also completed.

Materials and Methods

Subjects

Subjects were 70 male Wistar rats obtained from an inbred colony at the Royal Melbourne Hospital Biological Research Facility. Rats were 8–12 weeks of age at the time of LFPI, were housed individually under a 12 h light/dark cycle, and were given access to food and water ad libitum for the duration of the experiment. All experimental procedures were approved by the Melbourne Health Animal Ethics Committees (AEC#0705687). We have previously published data obtained from these rats (Jones et al., 2008a; Liu et al., 2010).

Experimental groups

Rats were assigned to receive either a sham injury (n = 25) or an LFPI (n = 45). Of the LFPI rats, 11 (24%) died immediately post-LFPI, one died during electrode surgery, and 10 others were excluded from the current study due to compromised video–electroencephalography (EEG) recordings. Therefore, a total of 23 rats given LFPI were included in this study. Based on video-EEG analysis for measures of PTE (as described below), the LFPI rats were further divided into epileptic (n = 12) and nonepileptic (n = 11) groups. Here we compared only the epileptic and nonepileptic groups. For results comparing sham-injury and LFPI groups (see Jones et al., 2008a and Liu et al., 2010).

Lateral fluid percussion injury (LFPI)

LFPI and sham-injury procedures were based on a standard protocol as described previously and used by our group (McIntosh et al., 1989; Thompson et al., 2005; Jones et al., 2008a). Briefly, under anesthesia a 5-mm craniotomy, positioned 4-mm lateral and 4-mm posterior to bregma, was performed to create a circular window exposing the intact dura mater of the brain. A modified female Luer-Lock cap was secured over the craniotomy window by dental acrylic. The rat was then removed from anesthesia and attached to the fluid percussion device via the modified female Luer-Lock cap. Once the rat responded to a toe pinch, a severe-intensity (3.2–3.5 atmospheres) fluid pulse of silicone oil generated by the fluid percussion device was delivered to the brain via the modified female Luer-Lock cap. Rats were resuscitated with pure oxygen postinjury if required. Upon resumption of spontaneous breathing, and return to pre-LFPI levels of heart rate and oxygenation status, the dental acrylic caps were removed and the wound sutured closed. Sham-injury rats underwent the same procedures as LFPI rats, with the exception that the fluid pulse was not given.

Acute neuromotor assessment

As described previously, acute neurologic injury was assessed in all rats on the day before injury, and on each day for 3 days after injury, using a composite neuromotor score (McIntosh et al., 1989; Jones et al., 2008a). Briefly, neuromotor assessment included ability to traverse a flat wooden beam, Rotarod, forelimb flexion when the rat is elevated by its tail, reflex in response to a loud startle, and the light escape task. Each task was judged on an ordinal score of 0 (pass) or 1 (fail), except the forelimb flexion, which has an ordinal graded severity (maximum deficit score of 2 per limb).

Neuroimaging

Acquisition

T2-weighted MRI scans were performed at baseline, and both MRI and 18F-FDG-PET scans were performed at 1 week, and 1, 3, and 6 months post-LFPI. The details of MRI and 18F-FDG-PET scanning procedures and protocols used in this study have been described previously (Liu et al., 2010). Briefly, T2-weighted small-animal MRI data were acquired using a 4.7-T 47/30 Advance small-animal spectrometer with PARAVISION 3.0 (Bruker Biospec, Ettlingen, Germany), with a shield-gradient set appropriate for rats (Bouilleret et al., 2009). PET images were acquired using a Mosaic Animal PET scanner (Philips, Amsterdam, The Netherlands), and completed 1–3 days before the corresponding MRI scans. Rats were injected intraperitoneally (i.p.) with 37–74 MBq (1–2 mCi) of 18F-FDG at 30 min before scanning. Rats were then anesthetized with 2% isoflurane (1:1 oxygen to air) and placed onto an acrylic platform for a 30-min acquisition. Corrections were applied for dead-time loss, decay, and activity injected (standardized uptake value [kBq/mL]).

MRI volumetric analysis

Two rats from the epileptic group were excluded from all imaging analysis due to incomplete scans. All imaging analysis procedures followed those described previously (Liu et al., 2010). Briefly, T2-weighted MRI volumes of selected brain regions were quantified with manually drawn regions of interest (ROIs) using Analyze (Analyze; Mayo Foundation, Rochester, MN, U.S.A.). Ten ROIs, including the cortex, hippocampus, thalamus, amygdala, and lateral ventricles from both hemispheres, were drawn as described previously (Bouilleret et al., 2009; see Fig. 1). ROIs were drawn on consecutive axial MRI slices by an investigator blinded to experimental conditions. Only slices containing hippocampus were analyzed. Before ROIs were drawn, the MR image set was coregistered to a template MR image with the ROIs predefined. The ROIs were then redrawn for the target MR image on the basis of this template.

Figure 1.

(A) Regions of interest (ROIs) used in imaging analyses. White outlines illustrate cortex (Ctx), hippocampus (Hc), thalamus (Tha), and amygdala (Ag) ROIs used for MRI and PET analyses. The circles within each ROI were used for intensity analysis. (B, C) Representative coronal thionin-stained histologic images of LFPI (B) and sham-injury lesions (C; scale bar = 2 mm; see Jones et al., 2008a for additional details).

MRI intensity analysis

As described previously (Onyszchuk et al., 2009), analysis was performed with Matlab (MathWorks, Natick, MA, U.S.A.) to measure mean intensities in ROIs (cortex, hippocampus, thalamus, and amygdala from both hemispheres) drawn on coronal MRI slices by an investigator blinded to experimental conditions (see Fig. 1). Only slices containing hippocampus were analyzed. ROI intensities were normalized to muscle tissue intensity.

Large-deformation high-dimensional mapping of hippocampal morphometry

Large-deformation high-dimensional mapping (HDM-LD) of hippocampal morphometry was also completed as described previously (Hogan et al., 2004; Liu et al., 2010). HDM-LD utilizes computer-assisted shape recognition to identify patterns within MRI data (Bouilleret et al., 2009; Hogan et al., 2009). Computational anatomic techniques produce three-dimensional surface representations of the hippocampus with resolution at a subvoxel level, enabling visualization of details of hippocampal surface anatomy (Gardner & Hogan, 2005). Changes in morphology can be detected even when total volume changes are not significantly different (Csernansky et al., 1998; Posener et al., 2003; Hogan et al., 2004, 2008). Here we generated coordinates for hippocampal surface transformation for each image using a common template. Hippocampal surfaces were first corrected for global size differences by registering the corresponding brain volume to an average rat atlas and obtaining global scaling estimates in the three spatial directions (Snyder, 1996). After scaling, the surfaces were registered to a common hippocampal surface template. Average hippocampal deformations for the baseline and 1 week scans for both groups were then computed. Vertex-level differences within and between groups were entered as a dependent variable in analysis of variance (ANOVA) and post hoc unpaired t-tests, corrected for multiple comparisons with a cluster-constrained empiric noise estimation model.

PET analysis

For PET ROI analysis, each 18F-FDG-PET image was coregistered to its corresponding MRI scan, on which the 10 ROIs were defined. The mean activity was measured for all 10 ROIs, and mean value of activity for the ROIs was calculated. For PET statistical parametric mapping (SPM) analysis of PET data we used SPM (SPM5; Wellcome Trust Centre for Neuroimaging, University College London, London, United Kingdom). For PET preprocessing, the PET subvolumes encompassing the brain were extracted and resampled to a 0.11718 mm3 voxel size into the three-dimensional space of a reference MR image using trilinear interpolation (Analyze). We chose one sham-control PET scan with a high degree of symmetry, alignment, and freedom from artifacts as a reference PET image. We then created a mean image for use as the PET template by realigning six other individual PET scans using a rigid transformation and trilinear interpolation to the reference PET image (Casteels et al., 2006). To account for brain size variations over time, all individual PET images were spatially normalized to the PET template using the affine spatial normalization option in SPM and a trilinear interpolation. The voxel size of the normalized and resampled PET images was 0.1 mm3. Flexible factorial analysis with three independent factors was applied: treatment (sham or LFPI), time (1 week and 1, 3, or 6 months), and subject (rat number), with an interaction between factors one and two to perform the analysis. The PET images were proportionally normalized using the cerebellum mean value derived from the manual analysis. Contrasts assessing differences between epileptic and nonepileptic at each time point were derived and tested at a p < 0.01.

Behavioral assessment

As described previously (Jones et al., 2008a; Liu et al., 2010), all rats underwent well-validated assessments of anxiety-like behavior (elevated-plus maze and open-field test; Prut & Belzung, 2003; Carobrez & Bertoglio, 2005), learning and memory (water maze; Jones et al., 2009), and depression-like behavior (sucrose preference test and forced swim test; Porsolt et al., 1977; Willner et al., 1987). All of the tests, except the forced swim test, were performed at 1, 3, and 6 months postinjury. As it is moderately stressful to the animal, forced swim testing occurred only at the 6-month postinjury time point (Jones et al., 2008a).

Assessment for the development of epilepsy

Electrode implantation

Upon completion of the behavioral and imaging studies, all rats were implanted with six extradural screw-electrodes. The surgical procedure of electrode implantation has been described in detail (Stroud et al., 2005; Jones et al., 2008b). Briefly, rats were anesthetized with isoflurane in equal parts medical air and oxygen. Electrodes were custom-made by soldering gold “female” connector sockets (Ginder Scientific, Napean, Ontario, Canada) onto stainless steel Teflon-coated wire (SDR Clinical Technology, Sydney, New South Wales, Australia) with a 1.4 × 3 mm stainless steel screw (Mr. Specs, Parkdale, Victoria, Australia) attached at the opposite end. Each screw was placed into the skull via a separate burr hole, and the electrodes were inserted into a nine-pin Acrylonitrile – Butadiene – Styrene (ABS) plug (GS09PLG-220; Ginder Scientific). The entire assembly was secured with dental acrylic and the skin suture around the headpiece.

Video-EEG monitoring

Following electrode implantation and a 1-week recovery period, brain electrical activity was monitored for 2 weeks using continuous video-EEG recording in order to assess for PTE. As described previously (Stroud et al., 2005; Jones et al., 2008b), EEG cables were connected to the electrode headpiece of each rat, and to a computer running Compumedics video-EEG acquisition system (32 channel series; Compumedics Limited, Abbotsford, Victoria, Australia). For each rat, four of the six implanted electrodes were used for EEG recording, plus one electrode each for ground and reference. Using this configuration, each rat used four channels, allowing for a total of eight rats to be recorded at once on a 32-channel system. To monitor behavior during seizures, rats were continuously recorded by an overhead video camera (3–8 mm; Pentax, Tokyo, Japan). The camera was equipped with an infrared light and a wide-angle lens (Pentax), which allowed for day and night video recording of up to eight rats simultaneously (Van Raay et al., 2009).

Video-EEG analysis

Two separate experienced reviewers who were blinded to the experimental groups analyzed all EEG recordings. Determination of seizure expression followed protocols described in our previous work (Stroud et al., 2005; Jones et al., 2008b). Electroencephalographic seizure activity was defined as high-amplitude, rhythmic discharges that represented clearly a new pattern of tracing. This included repetitive spikes, spike-and-wave discharges, and slow waves. The event must have lasted at least 5 s and showed an evolution in the dominant frequency. If epileptic events occurred within an interval of <5 s without return to baseline, it was defined as belonging to the same seizure event (Stroud et al., 2005; Jones et al., 2008b). If an electrographic event was observed, the behavioral severity was determined from the corresponding video recording and scored for seizure class according to the Racine scale (Racine, 1972). For the event to be classified as a posttraumatic seizure both reviewers had to identify the electroencephalographic seizure activity and classify it as such.

Statistical analyses

The MRI and PET data for each ROI were analyzed separately using a two-way analysis of variance (ANOVA) for repeated measures, with epilepsy status and time after LFPI as the independent variables. Neuromotor and behavioral tests were analyzed using two-way ANOVA, with epilepsy status and time after LFPI as the independent variables. Multivariate logistic regression analyses were conducted to determine whether a combination of the parameters affected by LFPI, or the dynamics of changes within these parameters over time, might predict epileptic outcome. Analyses were performed using STATISTICA (StatSoft, Tulsa, OK, U.S.A.) or SPSS (IBM Corp, Armonk, NY, U.S.A.) software. Statistical significance was set at p < 0.05 and all data were expressed as mean ± standard error of the mean (SEM).

Results

Occurrence of posttraumatic epilepsy (PTE)

There was a 98% interrater reliability between reviewers, with both reviewers identifying 44 of 45 seizures. Spontaneous recurrent seizures were observed in 30% of rats (n = 7), with these rats experiencing an average of 6.3 seizures (median 4, range 2–17) over the 2-week video-EEG recording period (see Fig. 2). The mean duration of the seizures was 52.9 s (median 14.6 s; range 6–676 s). Epileptic discharges were observed in an additional 22% of rats (n = 5), with these rats experiencing an average number of 11.5 discharges (median 10.5; range 2–20) over the 2-week video-EEG recording period (see Fig. 2). Unless stated otherwise, these 12 rats (52%) comprised the epilepsy group for all remaining analyses. The remaining 48% of rats (n = 11) given LFPI were not observed to experience spontaneous recurrent seizures or epileptic discharges and were the nonepileptic LFPI group. The sham-injury rats displayed no evidence of spontaneous recurrent seizures or epileptic discharges and were therefore not included in any other analyses. For details regarding comparisons between the sham-injury and LFPI groups, see Jones et al., 2008a and Liu et al., 2010.

Figure 2.

Representative examples of EEG traces of electroencephalographic seizure activity from each of the seven rats that had recurrent spontaneous seizures recorded on video-EEG monitoring (AG), and an example of an epileptic discharge recorded on an additional five rats (H), 6 months after LFPI. Arrows indicate beginning and end of seizure activity. Scale bars = 1 s.

Large-deformation high-dimensional mapping of hippocampal morphometry

HDM-LD analysis revealed that both of the LFPI groups (epileptic and nonepileptic) displayed a significant deformation of the ipsilateral hippocampus at 1-week postinjury relative to baseline measures (see Fig. 3). However, these effects differed between the epileptic and nonepileptic groups. Specifically, the epileptic group displayed an increase in lateral regions of the hippocampus, whereas the nonepileptic rats displayed a decrease in the medial-ventral regions.

Figure 3.

Large-deformation high-dimensional mapping (HDM-LD) of the ipsilateral hippocampus illustrating changes in epileptic (A, B) and nonepileptic rats (C, D) 1 week after LFPI compared to baseline. Epileptic rats displayed significantly thicker surface in the lateral region of the hippocampus at 1-week post-LFPI (red-yellow) compared to baseline (unpaired two-tailed t-test, p < 0.05, corrected for multiple comparisons). Nonepileptic rats displayed significantly thinner surface in the medial-ventral region at 1-week post-LFPI (blue) compared to baseline.

T2-weighted MRI and PET

There were no significant differences found between epileptic and nonepileptic rats in T2-weighted MRI (all p > 0.05, see Fig. 4) or PET measures (all p's > 0.05, see Fig. 5).

Figure 4.

MRI-based analysis of structural damage after LFPI found no significant differences between epileptic and nonepileptic rats. Time course of volume changes from baseline to 6 months in hippocampus (A) and cortex (B) are shown (mean ± SEM). (CE) Representative T2-weighted images at 6 months postinjury demonstrate brain damage in both the LFPI groups (epileptic – C and nonepileptic – D) relative to sham-injury (E; see Liu et al., 2010 for additional details), but no significant differences between the epileptic and nonepileptic groups.

Figure 5.

PET-based ROI analysis of functional changes after LFPI found no significant differences between epileptic and nonepileptic rats. However, a multivariate logistic regression model that incorporated the 1-week, and 1- and 3-month PET parameters from the ipsilateral hippocampus was able to predict the epileptic outcome in all of the LFPI cases. Time course of metabolic changes on PET scans in the hippocampus (A) and the cortex (B) are shown (mean ± SEM). Representative PET images at 3 months postinjury demonstrate a decrease in metabolism in both the LFPI groups (epileptic – C and nonepileptic – D) relative to sham-injury (E; see Liu et al., 2010 for additional details), with worsened hypometabolism observed in the ipsilateral hippocampus of the epileptic group.

Neuromotor and behavioral assessment

There were no significant differences in acute neuromotor scores (all p > 0.05, data not shown), indicating that both groups experienced similar levels of injury severity. Behavioral measures also found no differences between epileptic and nonepileptic LFPI rats (all p > 0.05, see Fig. 6).

Figure 6.

Behavioral outcomes (mean ± SEM) in the elevated-plus maze (A), open field (B), and (C) water maze. There were no significant differences between epileptic and nonepileptic groups. However, rats given LFPI are hyperanxious compared to sham-injured rats, *significantly greater than LFPI rats, p < 0.05, see Jones et al. (2008a) for additional details regarding differences between sham and LFPI rats.

Multivariate logistic regression model

Multivariate logistic regression analyses determined that none of the neuromotor, behavioral, or MRI parameters were significant predictors of epileptic outcome (all p > 0.05). However, a multivariate logistic regression model that incorporated the 18F-FDG PET parameters from the ipsilateral hippocampus at 1 week, and 1 and 3 months postinjury blocked over time was able to significantly predict the epileptic outcome in 100% of the LFPI cases (χ2(3) = 19.121, p < 0.001; see Fig. 5).

Analyses excluding rats with epileptic discharges

To ensure that the sensitivity of the analyses were not limited by the inclusion of rats that experienced epileptic discharges in the absence of spontaneous seizures, additional statistical analyses that excluded these rats were also conducted. Consistent with the previous analyses, there were no significant differences between the epilepsy and nonepilepsy groups on any of the measures (all p > 0.05). Furthermore, the multivariate logistic regression model incorporating the serial ipsilateral hippocampus PET parameters was again able to significantly predict the epileptic outcome in 100% of the cases (χ2(3) = 17.845, p < 0.001), whereas the neuromotor, behavioral, and MRI parameters where unable to significantly predict epileptic outcome (all p > 0.05).

Discussion

Herein we report that a proportion of rats given LFPI display evidence of PTE at 6 months postinjury. Rats were divided into epileptic and nonepileptic groups in order to investigate whether structural, functional, and behavioral changes were predictive of PTE. HDM-LD analysis based on T2-weighted MRI revealed that the epileptic and nonepileptic rats displayed minor but significant differences with regard to changes in the ipsilateral hippocampus at 1-week postinjury relative to baseline, with epileptic rats showing an increased thickness in the lateral region and nonepileptic rats showing thinning in the medial-ventral region. In addition, a multivariate logistic regression model including the serial 18F-FDG-PET parameters from the ipsilateral hippocampus was able to significantly predict the epileptic outcome in all of the LFPI cases. With the exception of these findings, the detailed serial MRI, PET, and behavioral analyses for 6 months postinjury found no other differences between epileptic and nonepileptic rats.

The validity and limitations of LFPI as an animal model of PTE

Previous studies have reported inconsistent findings regarding the use of LFPI as an animal model of PTE. Although D'Ambrosio et al. (2004, 2005) report PTE in 92–100% of rats in chronic stages after severe LFPI, Kharatishvili et al. (2006) report incidence rates of 43–50%. The variability in these findings is likely a result of key methodologic differences (i.e., age and injury parameters), as well as the differences in the definition of what is a seizure between these studies. The determination of what is a seizure in rodent-acquired epilepsy models is a highly controversial area, with no generally accepted criteria (D'Ambrosio & Miller, 2010; Dudek & Bertram, 2010). In this study we have used a rigorous definition that had been used in previous publications on experimental posttraumatic epilepsy in the LFPI model by Kharatishvili and colleagues (Kharatishvili et al., 2006, 2007) and subsequently applied in our previous work (Bouilleret et al., 2009, 2011). The determination was done by two independent experienced reviewers who were blinded to the experimental groups, with 44 of 45 of the seizures identified by both of our blinded reviewers and no seizures identified in sham-injury animals. Therefore, it is unlikely that the identified events are normal oscillations or electrical noise or genetically determined absence seizures. Because our methods and definition of PTE closely align with those of Kharatishvili et al. (2006), it is reassuring that our finding that 30–52% of rats given severe LFPI displayed either spontaneous seizures and/or epileptic discharges are similar to those reported by this group (Kharatishvili et al., 2006). This is also consistent with rates of PTE in patients following TBI (Englander et al., 2003; Frey, 2003; Christensen et al., 2009), and supports the use of LFPI in rats to model and study PTE. However, it is important to acknowledge that some workers in the field would question the relevance of brief seizures, even when lasting longer than 5 s, recorded in rodents postexperimental TBI to human posttraumatic epilepsy (Dudek & Bertram, 2010). The median duration of seizures recorded in this study was 14.6 s, with a range of 6–676 s.

There are also limitations with the current study related to the LFPI–PTE model that must be considered when interpreting the present results. Here we conducted 2 weeks of continuous video-EEG recordings at 6 months post-LFPI to detect epileptic rats. This was done because previous studies had demonstrated that the majority of animals given LFPI who develop PTE do so by 6 months (Kharatishvili et al., 2006), 2 weeks of continuous video-EEG recording was the most detailed analysis at 6 months post-LFPI previously reported (Kharatishvili et al., 2006), and compatibility issues associated with the EEG-recording electrodes and MRI scans meant that the EEG electrodes had to be implanted following completion of the serial MRI acquisitions. However, we acknowledge that post-LFPI epileptic animals can experience prolonged seizure-free intervals, and initial seizures may take >6 months to occur (Kharatishvili et al., 2006). Therefore, we cannot exclude the possibility that our “nonepileptic” group included epileptic animals that experienced seizures outside of the monitoring period, or may have gone on to develop seizures had the study continued to a later time point. This may have contributed to the lack of differences between epileptic and nonepileptic animals on many of the measures, and, as such, the remaining discussion should be interpreted in light of this concern. Furthermore, it is important for future studies investigating PTE to consider these limitations and incorporate the most detailed seizure analysis that is feasible.

Neuroimaging and behavioral predictors of PTE

Here, using advanced analysis of MRI and PET parameters, we were able to identify subtle changes that may be predictive of PTE after LFPI in the rat. Specifically, MRI-based HDM-LD hippocampal morphometry analysis identified significant surface changes in the ipsilateral hippocampus that differed between epileptic and nonepileptic rats, and a multivariate logistic regression model that incorporated the serial PET parameters from the ipsilateral hippocampus at 1 week and 1 and 3 months postinjury was able to significantly predict the epileptic outcome in each of the cases. Although these subtle findings reached statistical significance, we interpret them cautiously given that no other measures in the study comparing the epileptic and nonepileptic groups identified differences. Nonetheless, the strength of HDM-LD and multivariate logistic regression analyses is their ability to detect subtle findings (Csernansky et al., 1998; Hogan et al., 2009). Therefore, the differences reported herein may indicate hippocampal abnormalities that occur at various time points post-TBI that are involved in PTE. Indeed, numerous other LFPI studies have implicated hippocampal abnormalities with PTE (Pitkänen et al., 2009), and some suggest that hippocampal changes identified by MRI-based analyses could serve as a surrogate marker for PTE (Kharatishvili et al., 2007). Furthermore, other studies have found that 18F-FDG-PET hypometabolism in the hippocampus, similar to that observed in the epileptic rats here, may be associated with epileptogenesis (Jupp et al., 2012). However, whether the findings reported herein are truly related to the epileptogenic process will require larger and more detailed studies, perhaps focusing histopathologic and biochemical examinations on the regions that showed significant HDM-LD changes, as well as more temporally sensitive PET imaging and EEG analysis.

Despite the use of serial T2-weighted MRI and 18F-FDG-PET imaging to assess structural and functional changes for 6 months post-LFPI, we found no other significant differences between epileptic and nonepileptic rats. The lack of major structural changes associated with PTE reported herein is consistent with previous reports that cortical damage assessed with serial T2-weighted MRI was not related to seizure susceptibility in rats given LFPI (Kharatishvili et al., 2007). Taken together with the lack of functional differences detected by the serial 18F-FDG-PET, the ROI techniques used here were unable to detect any major structural and metabolic changes post-TBI predictive of PTE, limiting their use as predictive tools and suggest that PTE occurs independent of the major structural and functional changes induced by TBI. However, although the ROI techniques used here may lack the capability (i.e., spatial resolution) necessary to solely detect more subtle abnormalities involved in PTE, other neuroimaging modalities, such as diffusion or functional MRI, may be more useful (Kharatishvili et al., 2007).

The behavioral measures also failed to detect significant changes related to epileptic status, indicating that these are not reliable indicators of PTE. However, it is important to note that although no differences were found between epileptic and nonepileptic rats, as a whole the rats given LFPI did display significant imaging and behavioral abnormalities relative to sham-injured rats (see Jones et al., 2008a; Liu et al., 2010; Figs. 4-6). Therefore, although behavioral and imaging abnormalities would be predicted to occur in epilepsy (Golarai et al., 2001; Jones et al., 2008b; Pitkänen et al., 2011), similar changes induced by TBI may have confounded the sensitivity of our measures. Furthermore, the behavioral testing may have occurred at time-points not sensitive to changes related to PTE, and the behavioral tasks were not comprehensive to the entire spectrum of behaviors potentially associated with PTE. For example, the water maze task used in this study was specific to spatial learning and memory and failed to identify significant group differences, whereas other cognitive tasks (Saucier et al., 2008), or variations of the water maze (Shultz et al., 2009), could be used to assess whether other learning and memory abnormalities may serve as PTE biomarkers. Therefore, future studies that employ more detailed techniques may be able to detect slight differences between epileptic and nonepileptic rats post-TBI that were not observed here.

In conclusion, here we examined rats given LFPI using serial MRI, PET, and behavioral assessments for 6 months postinjury. Based on video-EEG monitoring, rats were identified as either epileptic or nonepileptic, and comparisons were made to examine structural, functional, and behavioral changes related to PTE. A subtle difference was identified between epileptic and nonepileptic rats in hippocampal morphometry, and a multivariate logistic regression model that included the serial PET parameters from the ipsilateral hippocampus was able to accurately predict epileptic outcome. Although these findings provide cautious optimism and might direct future studies, unfortunately no other readily identifiable differences between the epileptic and nonepileptic rats were detected. Taken together, these findings indicate that PTE may not be related to major structural, functional, and behavioral changes, and suggest that other more subtle changes or mechanisms may be involved.

Acknowledgments

We wish to acknowledge the staff of the Howard Florey Institute Small Animal MRI Facility, and the Peter MacCallum Cancer Centre. Funding for this project was provided by the Victorian Government Transport Accident Commission (DP0023). N.J. and T.O. acknowledge the financial support of the NHMRC of Australia.

Disclosure

None of the authors have any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Ancillary