Hippocampal MRI Signal Hyperintensity After Febrile Status Epilepticus Is Predictive of Subsequent Mesial Temporal Sclerosis
Abstract
OBJECTIVE. The objective of our study was to test the hypothesis that the finding of hyperintense hippocampal signal intensity on T2-weighted MR images soon after febrile status epilepticus is associated with subsequent hippocampal volume loss and persistent abnormal signal intensity on T2-weighted images (i.e., mesial temporal sclerosis).
SUBJECTS AND METHODS. Eleven children (mean age, 25 months) underwent initial MRI that included coronal temporal lobe imaging within 72 hours of febrile status epilepticus and follow-up imaging from 3 to 23 months later (mean, 9 months). A neuroradiologist blinded to clinical history graded initial and follow-up hippocampal signal intensity on a scale from 0 (normal) to 4 (markedly increased). Two blinded observers measured hippocampal volumes on initial and follow-up MR studies using commercially available software and volumes from 30 healthy children (mean age, 6.3 years). Initial signal intensity and hippocampal volume changes were compared using Kendall tau correlation coefficients.
RESULTS. On initial imaging, hyperintense signal intensity ranging from 1 (minimally increased) to 4 (markedly increased) was seen in seven children. Four children had at least one hippocampus with moderate or marked signal abnormality, three children had a hippocampus with mild or minimal abnormality, and four children had normal signal intensity. The Kendall tau correlation coefficient between signal intensity increase and volume change was –0.68 (p < 0.01). Five children (two with temporal lobe epilepsy and two with complex partial seizures) had hippocampal volume loss and increased signal intensity on follow-up imaging, meeting the criteria for mesial temporal sclerosis.
CONCLUSION. MRI findings of a markedly hyperintense hippocampus in children with febrile status epilepticus was highly associated with subsequent mesial temporal sclerosis.
Introduction
Acute changes in hippocampal size and signal intensity after status epilepticus have been reported on MRI studies by a number of investigators [1–3]. In some subjects the abnormalities appeared to resolve on follow-up MRI examinations [1–4], whereas in other subjects the investigators reported subsequent development of mesial temporal sclerosis (i.e., abnormal signal intensity and decreased hippocampal size) [1, 5–7]. Furthermore, in some studies, decreased hippocampal size on subsequent imaging did not always correlate with increased signal intensity reflected by T2 relaxation times [8]. A general understanding exists that status epilepticus can produce damage within brain structures [9]. However, at present, a clear consensus has not been reached as to whether febrile status epilepticus results in mesial temporal sclerosis and temporal lobe epilepsy. For instance, in one large long-term series, the authors concluded that febrile status epilepticus does not produce a greater likelihood of mesial temporal sclerosis [10]. In other studies, the authors found that children with febrile status epilepticus whose seizures were both focal and prolonged were more likely to have either hippocampal signal abnormality on MRI soon after the ictus or subsequent mesial temporal sclerosis [6, 11]. The role of febrile status epilepticus in the development of a subsequent seizure disorder and mesial temporal sclerosis is actively debated in the medical literature [12–14].
Animal studies using MRI to assess status epilepticus have suggested that the hippocampal signal increase seen acutely on T2-weighted images may be associated with swelling of the neuropil and with neuronal necrosis, ultimately leading to mesial temporal sclerosis [15, 16]. However, a recent MR study of prolonged hyperthermia-induced seizures in infant rats found transient hyperintensity on T2-weighted images without subsequent hippocampal cell loss [17]. Nevertheless, the seizures did produce long-lasting temporal lobe hyperexcitability and eventually temporal lobe seizures in the same animals [18]. Given the varied outcome of both human and animal studies, the prognostic significance of increased signal intensity on T2-weighted images in the human hippocampus after status epilepticus remains unclear and the conditions under which febrile status epilepticus produces mesial temporal sclerosis are still to be determined.
Febrile status epilepticus is defined as a prolonged seizure (i.e., at least 30 minutes in duration) or a series of seizures lasting more than 30 minutes without interictal recovery occurring in the context of a febrile illness [19]. Febrile status epilepticus accounts for approximately 25% of status epilepticus in childhood [20]. One of the most important questions regarding febrile status epilepticus is whether it causes mesial temporal sclerosis and later temporal lobe epilepsy (i.e., partial complex seizures originating in the temporal lobe) [14, 21]. Retrospective studies suggest that febrile status epilepticus is a common cause of mesial temporal sclerosis and temporal lobe epilepsy, but epidemiologic studies have not found a link [22, 23]. Determination of whether increased hippocampal signal intensity is associated with subsequent mesial temporal sclerosis or temporal lobe epilepsy is important in managing patients with febrile status epilepticus.
In this study, we compared the initial and follow-up hippocampal MRI findings in a group of children who underwent MRI for febrile status epilepticus. Our goal was to determine whether increased hippocampal signal intensity on T2-weighted images after febrile status epilepticus is associated with subsequent mesial temporal sclerosis. Some previous studies have been performed in a research environment using specific research protocols, such as signal intensity measurements using T2 relaxometry [24, 25]. Such advanced techniques are clinically feasible using rapid imaging techniques but are not routinely performed in a clinical setting at most institutions; we set out to use the imaging sequences that are performed in most clinical settings. For that reason, we intentionally chose to perform our study, as much as possible, in a clinical environment so that our findings would be relevant to practicing radiologists. Thus, we imaged patients on a clinical MR scanner using routine imaging parameters that are widely used for evaluation of the hippocampus. These images were assessed by a practicing neuroradiologist. We hypothesized that the severity of the signal abnormality on initial clinical MRI would correlate with subsequent hippocampal volume loss and mesial temporal sclerosis.
subjects and Methods
From 1995 through 2002, 38 children who underwent fast spin-echo MRI (febrile status epilepticus) were enrolled in an institutional review board (IRB)–approved prospective study of complex febrile seizures at our institution. Children were prospectively enrolled and parental consent was obtained in all cases. The group reported here consists of 11 of these 38 children who had undergone both initial and follow-up MRI examinations at the time of data analysis. The remaining 27 children were excluded because they underwent a single MRI study. Throughout most of the time period of this study, diffusion-weighted imaging was not available and coronal FLAIR images were not used, which accounts for the absence of those pulse sequences in our study.
The 11 children (eight girls; age range, 11–45 months; mean age, 25 months) underwent MRI examination for clinical purposes within 72 hours of status epilepticus (mean delay between seizure onset and MRI, 36 hours) and a follow-up clinical MR examination during the subsequent 2 years (range, 3–23 months; mean, 9.2 months). The reasons for follow-up imaging included IRB-approved repeat MR studies to determine the effect of febrile status epilepticus on hippocampal growth and signal as well as clinically indicated imaging for possible epilepsy.
We defined febrile status epilepticus as either a documented febrile seizure of at least 30 minutes' duration or intermittent febrile seizure activity without recovery of consciousness between seizures lasting at least 30 minutes [19]. The pediatric neurologist involved in the study reviewed patient charts to determine the duration of status epilepticus in all children. For our study, we defined mesial temporal sclerosis as the loss of volumetrically measured hippocampal volume on the follow-up MR examination combined with persistent abnormal hippocampal signal intensity. For inclusion, children needed to have undergone an initial MRI examination within 72 hours of the end of the status epilepticus episode and a follow-up MR examination. The 72-hour time point, which was chosen by the pediatric neurologist participating in the study, was proposed on the basis of two competing issues: the difficulty of performing imaging in an acutely ill, postictal child; and the desire to obtain images of the hippocampus relatively soon after status epilepticus. One child (patient 8 in Table 1) had a history of meconium aspiration that resulted in hypoxic brain injury. This child did not have a previous MR image. Because this child had no history of seizures and because previous brain injury does not disqualify an individual from being diagnosed with febrile status epilepticus, we included this child in the study. We excluded CNS infections in all children by lumbar puncture as part of routine clinical care. We have previously reported preliminary imaging findings in eight of these children [11]. This study differs from the previous study by inclusion of more complete follow-up MRI, the addition of three patients, assessment of solely children who had follow-up imaging, and correlation of clinical features with MRI findings. HIPAA compliance was observed for all data acquisition, storage, and transmission.
| Ipsilateral Hippocampus | Contralateral Hippocampus | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Patient No. | Initial Signal Intensity Score | Initial Volume (mm3) | Follow-Up Signal Intensity Score | Volume Change (mm3) | Follow-Up Volume (mm3) | Initial Signal Intensity Score | Initial Volume (mm3) | Follow-Up Signal Intensity Score | Volume Change (mm3) | Follow-Up Volume (mm3) | Subsequent Seizure Status |
| 1a | 3 | 3,634 | 2 | –1,084 | 2,550 | 0 | 3,361 | 0 | –232 | 3,129 | — |
| 2a | 4 | 3,978 | 3 | –1,217 | 2,761 | 1 | 3,607 | 0 | –88 | 3,519 | TLE |
| 3a | 3 | 2,883 | 4 | –715 | 2,168 | 0 | 3,075 | 0 | +224 | 3,299 | TLE |
| 4a | 3 | 3,517 | 2 | –7 | 3,510 | 0 | 3,505 | 0 | +322 | 3,827 | CPS |
| 5 | 2 | 3,827 | 1 | +262 | 4,089 | 0 | 3,928 | 0 | +254 | 4,182 | — |
| 6b | 2 | 3,823 | 1 | –38 | 3,785 | 0 | 3,627 | 0 | +335 | 3,962 | CPS |
| 7b | 1 | 3,342 | 0 | –71 | 3,271 | 0 | 3,005 | 0 | +261 | 3,266 | CPS |
| 8b | 0 | 2,222 | 1 | +172 | 2,394 | 0 | 2,303 | 0 | +263 | 2,566 | CPS |
| 9 | 0 | 3,512 | 0 | +115 | 3,627 | 0 | 3,199 | 0 | +234 | 3,433 | — |
| 10 | 0 | 4,356 | 0 | –5 | 4,351 | 0 | 4,012 | 0 | +31 | 4,043 | — |
| 11b | 0 | 2,692 | 0 | +979 | 3,671 | 0 | 2,109 | 0 | +704 | 2,823 | — |
Note—Signal intensity was graded as normal (score of 0), minimally increased (1), mildly increased (2), moderately increased (3), or markedly increased (4). Dash (–) indicates no subsequent seizures. TLE = temporal lobe epilepsy, CPS = complex partial seizures
a
Follow-up imaging was performed for markedly increased signal intensity on initial MRI
b
Follow-up imaging was performed for clinical suspicion of subsequent seizures
Clinical seizure data, which included seizure duration (calculated as total time actually manifesting seizure activity excluding the time between seizures when they were intermittent), time interval from the end of status epilepticus until imaging, temporal duration of status epilepticus, evidence of lateralized clonic or tonic seizure activity and body temperature, were obtained from emergency depart ment notes and interviews with physicians and family members. This information was correlated with hippocampal signal intensity scores on initial imaging. One of the authors, a pediatric neurologist specializing in epilepsy who treated several children, obtained information about subsequent seizures from telephone interviews and clinic visits.
All MRI examinations were performed on one of four 1.5-T scanners (Signa, GE Healthcare) using a standard head coil. The imaging protocol consisted of transverse unenhanced T1-weighted images (TR/TE, 600/25; 2 excitations), transverse T2-weighted and intermediate-weighted (proton density) images (TR/TE first echo, second echo, 2,800/30, 80; 1 excitation), and coronal fast spin-echo T2-weighted images (TR/TE, 4,000/100; 4 excitations) with contiguous 3-mm slices through the hippocampi with the plane of the slices perpendicular to the long axis of the left hippocampus. MR technologists trained in the determination of the long axis of the hippocampus and the neurologist involved in the study verified positioning of these slices.
Scoring of Hippocampal Signal Intensity
Initial and follow-up coronal T2-weighted images of hippocampi were reviewed by a neuroradiologist with many years of experience in pediatric neuroradiology who viewed hard-copy films that included patient identifiers and were magnified approximately 2.5 times relative to standard filming. Standard window width and level settings were not designated for all films. The reader was blinded to clinical data and interpretations of MR studies. The reader first scored the initial imaging studies and the control studies, described later, in a single-day session and 6 months later reviewed the follow-up images and the same control studies in another single-day session.
The reader graded hippocampal signal intensity on a numeric scale using the following possible choices: normal (score of 0), minimally increased signal intensity (score of 1), mildly increased signal intensity (score of 2), moderately increased signal intensity (score of 3), and markedly increased signal intensity (score of 4). The reader determined whether abnormal signal intensity was present in the entire hippocampus, solely the hippocampal head, or the hippocampal body and tail (but not the hippocampal head). Regardless of whether only a part of the hippocampus was involved, a score was provided based on the degree of signal abnormality in the involved hippocampal segment or segments.
MRI examinations from seven control subjects (four girls; age range, 2 months–11 years; mean age, 4.1 years) were randomly intermingled with those of patients during both film reviews. These control examinations were obtained from children who had undergone clinical MRI using the same imaging protocol as febrile status epilepticus patients and whose imaging examinations were interpreted as normal. These studies served as controls for hippo campal signal intensity. Complaints such as headache or paroxysmal spells of nonepileptic nature (i.e., nonepileptic spells that never recurred and for which we were unable to provide an explanation) were the indications for MRI in these children.
Measurement of Hippocampal Volumes
Hippocampal volumes in both febrile status epilepticus patients and control patients were measured using commercially available software (Analyze, Biomedical Imaging Resource, Mayo Clinic) by two readers: one of the authors and a research associate blinded to clinical information about the subjects and trained by that author. Hippocampal volumes of 30 children (19 girls and 11 boys; age range, 1 month–16 years; mean age, 6.3 years) who had no history of developmental delay or seizures were also measured and were used to generate a growth curve of normal hippocampi (Figs. 1 and 2). These children under went MRI for clinical purposes using the same imaging parameters as those used for febrile status epilepticus children and their MR studies were interpreted as normal. The most common indi cations for MRI were headache and suspicion of seizures, which had been excluded on the basis of clinical and EEG data.
Interobserver (i.e., comparing measurements of the blinded observer and one of the authors) and intraobserver coefficients of variation, calculated using the equation [(SD of hippocampal volumes)/(mean of volumes)] × 100, were verified using repeated measures in 10 subjects and ranged from 3% to 4%.
The correlation between initial signal intensity scores and associated hippocampal volume change was measured using a Kendall tau correlation coefficient averaged over the left and right hippocampi. The associated p value was computed by means of the statistical jackknife applied over patients. A scatterplot of volume decreases versus initial signal intensity scores using both hemispheres was gen erated. Formal statistical significance of the association was tested by means of the Kendall tau correlation coefficient. The intervals between the initial and follow-up MRI examinations varied considerably in this sample of 11 children. We assessed for a correlation between the duration of the time interval between MR scans and the degree of volume loss using a linear regression (Origin analysis program, version 7; OriginLab). The correlation between volume change and follow-up interval was tested using linear regression.
Results
Clinical Characteristics
Average rectal temperature at presentation was 38.7°C (range, 36.7–40.3°C; normal = 37.0°C). Eight children initially had fever of 38.3°C (i.e., 101°F), and three children who were afebrile on presentation developed rectal temperatures of > 38.3°C within 24 hours. Inclusion of these patients who were afebrile on presentation but developed fever soon after the seizure is in accord with the definition provided by the International League Against Epilepsy [19] that the seizures occur in association with a febrile illness. The average total seizure duration was 72 minutes (range, 30–216 minutes). Ten children had lateralizing seizure activity and one child had generalized seizure activity.
Scoring of Hippocampal Signal Intensity
On initial imaging of the subjects with febrile status epilepticus, seven children had increased signal intensity on T2-weighted MR images (Table 1). In one child (patient 2), both hippocampi had abnormal signal intensity. Normal signal intensity was found in 14 hippocampi and hyperintense signal ranging from 1 to 4 was seen in eight hippocampi in seven children. One hippocampus was scored as having marked signal increase (score of 4); three hippocampi were scored as having moderate signal intensity increase (score of 3); two hippocampi were scored as having mild signal intensity increase (score of 2); and two hippocampi were scored as having minimal signal intensity increase (score of 1), including the contralateral, less abnormal hippocampus of subject 2 who was the only subject with signal abnormalities noted bilaterally (Table 1). The details of the initial and follow-up signal intensity scores and volume measures for individual patients are outlined below.
The patient with abnormal signal intensity in both hippocampi (patient 2) had marked signal abnormality (score of 4) in the most affected hippocampus (hereafter termed the ipsilateral hippocampus) and minimal signal abnormality (score of 1) in the contralateral hippocampus on initial imaging. On follow-up imaging, that patient had moderate signal abnormality (score of 3) in the ipsilateral hippocampus, which was seen to have a volume decrease of 1,217 mm3 (– 31%); the contralateral hippocampus had normal signal (score of 0) of follow-up imaging, with a volume decrease of 88 mm3 (– 2%). In three patients (patients 1, 3, and 4), on initial imaging the affected hippocampus had moderate signal abnormality (score of 3) and the contralateral hippocampus had normal signal intensity. On follow-up imaging, the ipsilateral hippocampus in patient 1 had mild signal abnormality (score of 2) with a decrease in hippocampal volume of 1,084 mm3 (– 30%); the contralateral hippocampus had normal signal intensity and a decrease in hippocampal volume of 232 mm3 (– 7%). In patient 3, on follow-up imaging, the ipsilateral hippocampus had marked signal abnormality (score of 4) with a decrease in hippocampal volume of 715 mm3 (– 25%); the contralateral hippocampus had normal signal intensity and an increase in hippocampal volume of 224 mm3 (+ 7%). In patient 4, on follow-up imaging, the ipsilateral hippocampus had mild signal abnormality with a decrease in hippocampal volume of 7 mm3 (0%); the contralateral hippocampus had normal signal intensity and an increase in hippocampal volume of 322 mm3 (+ 9%).
In two patients (patients 5 and 6), on initial imaging the most affected hippocampi had mild signal intensity increase (score of 2) and the contralateral hippocampus had normal signal intensity. On follow-up imaging, the ipsilateral hippocampus in patient 5 had minimal signal intensity (score of 1) with a increase in hippocampal volume of 262 mm3 (+ 7%); the contralateral hippocampus had normal signal intensity and an increase in hippocampal volume of 254 mm3 (+ 7%). On follow-up imaging, the ipsilateral hippocampus in patient 6 also had minimal signal intensity (score of 1) but with a decrease in hippocampal volume of 38 mm3 (– 1%); the contralateral hippocampus had normal signal intensity and an increase in hippocampal volume of 335 mm3 (+ 9%).
In one patient (patient 7), on initial imaging the most affected hippocampi had minimal signal intensity (score of 1) and the contralateral hippocampus had normal signal intensity. On follow-up imaging, the ipsilateral hippocampus had normal signal intensity with a decrease in hippocampal volume of 71 mm3 (– 2%); the contralateral hippocampus had normal signal intensity and an increase in hippocampal volume of 261 mm3 (+ 9%).
In the remaining four patients, both hippocampi had normal signal intensity on both initial and follow-up imaging. Hippocampal volume change ranged from a decrease of 5 mm3 (0%) to an increase of 979 mm3 (+ 36%).
Correlation of Signal Intensity with Status Epilepticus Features
In the 10 children with lateralized seizures, the hippocampus with abnormal signal (or, for child 2, the more abnormal signal) was in the hemisphere of seizure origin in seven children, the contralateral hemisphere in two children, and unable to be correlated in one child because of normal signal intensity. A moderately high correlation was seen between the duration of status epilepticus and signal intensity on initial imaging (correlation coefficient of 0.46); the association approached, but did not meet, statistical significance (p = 0.06). No significant correlations were seen between signal intensity on initial imaging and age at presentation or rectal temperature. No correlation was seen between time interval before imaging and signal intensity scores: The two children imaged on the same day as status epilepticus had scores of 0; the two children imaged 1 day after status epilepticus had scores of 3; the five children imaged 2 days after status epilepticus had scores of 0 (n = 3), 3 (n = 1), or 4 (n = 1); and the two children imaged on the third day after status epilepticus (but earlier than 72 hours) had scores of 0.
Correlation of Initial Signal Intensity and Subsequent Volumes
The scatterplot of volume changes versus initial signal intensity scores appeared to show an approximate linear relationship and had a nominal Pearson correlation coefficient of 0.756 (Fig. 3). The hemisphere-averaged Kendall tau correlation coefficient of –0.68 was highly statistically significant (p < 0.01). Noticeable differences were seen between children with, on the one hand, moderate or marked hippocampal signal abnormality (scores of 3 or 4) and those with minimal or mild signal abnormality (scores of 1 or 2). In the four children with moderate or severe hippocampal signal abnormality (patients 1–4), mean change in volume in the abnormal hippocampus was – 756 mm3 (Figs. 1 and 3) compared with 207 mm3 in the 18 hippocampi with lower scores. Three of these hippocampi decreased in volume, and one hippocampus (patient 4) showed essentially no growth (Table 1). A representative case is shown in Figures 4A and 4B (patient 1 in Table 1). Among the three children with minimally increased or mildly increased hippo campal signal intensity (i.e., excluding patient 2), two hippocampi decreased in volume and the third hippocampus grew at a normal rate. The mean volume change in these three hippocampi was 51 mm3. For all four hippocampi having minimal or mild signal abnormality (including the less affected hippocampus in child 2), the mean volume change was 16 mm3.
Data for volume change of the hippocampi contralateral to the abnormal hippocampus in patients 1–7 are shown in Figure 2; these data indicate that hippocampal growth was seen in five children. The mean volume change in the six contralateral hippocampi with normal signal intensity in this group was 194 mm3. Only two of 14 hippocampi with normal signal intensity on initial imaging later decreased in volume, compared with seven of eight hippocampi with initial abnormal signal intensity that decreased in volume.
Among the four children with normal hippocampal signal on initial imaging (patients 8–11), one (patient 10) had no substantial change in volume on follow-up imaging and the other children had relatively normal increases in volume (Fig. 5). In the child with the history of meconium aspiration (patient 8), small volumes on both initial and follow-up imaging were seen. Fourteen hippocampi in the study group of 11 children were judged to have normal signal intensity on initial imaging. The mean volume change in these 14 hippocampi was 261 mm3.
This test for correlation of the interval between MR scans and the degree of volume loss showed only a very small slope in the direction of increasing volume loss with increasing time interval, which was not statistically significant (slope = 6.16 mm3/mo; R = –0.0295; p = 0.931).
Correlation of Signal Abnormality with Subsequent Seizures
Six children, five with abnormal signal intensity on initial imaging and one with abnormal signal on follow-up imaging, subsequently developed seizure disorders (Table 1). Three of the four children with moderately or markedly abnormal signal intensity have seizures and, of these children, two have confirmed temporal lobe epilepsy with EEG findings localized to the temporal lobe containing the acutely abnormal hippocampus. In addition, another child (patient 4) with moderate hippocampal signal abnormality developed seizures, which were classified by their description as partial complex seizures, but has a nonlocalizing EEG.
Three children with only minimal or mild signal abnormality developed seizures. One child (patient 8) has partial complex seizures with right central (extratemporal) spikes on EEG. Another child (patient 6) had a 2-year interval during which she had spells of being unresponsive lasting minutes with postictal sleep compatible with complex partial seizures but with rare bilateral spike and wave activity on EEG. Patient 7 had episodes of feeling frightened and of becoming unresponsive for 1–2 minutes with perioral cyanosis and postictal sleep, suggesting complex partial seizures; however, interictal EEGs showed normal findings.
Development of Mesial Temporal Sclerosis
Our criteria for mesial temporal sclerosis were that the affected hippocampus must show both a decrease in volume and persistently hyperintense signal on T2-weighted images in the follow-up MR examination. In five subjects (patients 1–4 and 6 in Table 1), these criteria were met. Of these five children, patients 2 and 3 have temporal lobe epilepsy and subjects 4 and 6 have had complex partial seizures of unclear origin.
Discussion
Our goal in this study was to correlate imaging findings soon after febrile status epilepticus with subsequent hippocampal volume loss. We adopted a radiologic definition of mesial temporal sclerosis of volume loss on a follow-up MR study with persistent abnormal signal on T2-weighted images [26]. Our study findings in this small patient sample indicate that the presence of markedly hyperintense signal within a hippocampus correlates well with resultant hippocampal volume loss and, in some instances, is followed by the appearance of mesial temporal sclerosis. Our study findings were obtained using standard MRI pulse sequences that are widely used in the clinical environment. Furthermore, our study indicates that the degree of hyperintense signal abnormality can be detected by a practicing neuroradiologist, rather than being dependent on measurement of T2 relaxation times using a multiecho sequence.
The pathologic substrate of the hyperintense hippocampal signal seen after status epilepticus in our patients may be explained by findings in animal experiments. The most detailed information about acute postictal increases of signal intensity on T2-weighted images comes from animal experiments in which neurotoxins or convulsants were used to induce limbic status epilepticus lasting many hours. The initial injury is more pronounced in the pyriform cortex and amygdala but is also seen in the hippocampus [27–31]. A reduction in the apparent diffusion coefficient for water, which is seen in the first 3 hours, precedes the increase in signal intensity and is manifested histologically by swelling of the neuropil. Increased signal on T2-weighted images appears within 12–24 hours after several hours of status epilepticus [27–33].
Histologically, edema is seen, which is rapidly followed by neuronal necrosis and tissue fragmentation [16, 34]. The largest degree of signal abnormality has been correlated with the degree of cortical edema [16]. In the lithium pilocarpine rat model of status epilepticus, increased T2 relaxation times precede development of mesial temporal sclerosis and epilepsy [15]. By extrapolation, the severe hippocampal hyperintense signal abnormality in some of our patients likely indicates the severe cell injury and swelling that can precede mesial temporal sclerosis. In one study, infant rats sacrificed 8 hours after status epilepticus had histologic evidence of degenerating neurons in hippocampi and other limbic structures [35]. MR volumetric measurements months after status epilepticus in infant rats have shown volume loss in hippocampi and other limbic structures in a subset of study animals, a finding that was validated by measurements in postmortem specimens. These animal experiments indicate that increased signal intensity on T2-weighted images can correlate with acute neuronal damage and it is reasonable to suppose that these factors may also be operative in humans.
Development of mesial temporal sclerosis after febrile status epilepticus in humans has been previously reported, but most of the data are derived from case reports or small studies without volumetric analysis. A review of the medical literature shows five case reports having a total of 14 subjects with serial MRI examinations but without volume measurements; five individuals with hyperintense hippocampal signal on initial T2-weighted images developed mesial temporal sclerosis [1–3, 5, 7]. In addition, a few case reports have detailed serial MRI hippocampal volumetric data after status epilepticus. In one report, a 69-year-old encephalitis patient with apparent right temporal lobe electrographic status epilepticus for 5 days had marked right hippocampal enlargement to 4,400 mm3 compared with 1,500 mm3 on the left on initial imaging; 2 months later, right hippocampal volume was 1,800 mm3 [4]. The other two reports with volume measurements are isolated cases of status epilepticus, both of which showed hyperintense hippocampal signal on initial T2-weighted images and subsequent volume loss [36, 37].
Recently, studies involving larger samples of children have used quantitative hippocampal T2 relaxometry after febrile status epilepticus. In one study of 21 children, the mean hippocampal T2 relaxation times were elevated bilaterally within 48 hours of febrile status epilepticus, although none was reported as having a visible signal abnormality on conventional MR images [24]. In a follow-up study of 14 patients from that study performed 4–8 months after febrile status epilepticus, hippocampal T2 relaxation times had returned to normal in MRI examinations in all children [25]. Four patients had unilateral hippocampal volume loss but with normal signal intensity, which prevented them from meeting mesial temporal sclerosis criteria. Unfortunately, the authors did not comment on a relationship between the degree of increase in T2 relaxation time and subsequent asymmetry. The results of these studies indicate that advanced imaging analysis techniques, such as T2 relaxometry, can detect changes that are not visible on conventional MR images alone and that hippocampal volume loss can result. This technique appears to offer great promise as a research tool but is not widely used in the clinical environment, which limits its relevance to the practicing radiologist. On the other hand, our study has shown that visual inspection of T2-weighted images provides information that is relevant to patient outcome. A recent report suggests that other MRI findings on initial imaging after status epilepticus, such as reduced apparent diffusion coefficient values within the hippocampus on diffusion-weighted imaging and increased lactate on MR spectroscopy, may also be predictive of mesial temporal sclerosis [38].
In our study, we found a strong correlation between the severity of hippocampal signal abnormality on initial MRI and subsequent loss of hippocampal volume. The hippocampal signal abnormality seen in our patients probably reflects acute edema that likely is accompanied by the types of neuronal injury and microstructural alteration reported in animals [15, 16]. Such acute edema and swelling could have resulted in an overestimation of the initial volume in hippocampi with abnormal signal intensity [24]. The result would be to also overestimate the degree of volume loss between studies and potentially invalidate our proposed association between degree of hyperintense signal and volume change. Importantly, it appears that the degree of signal abnormality may reflect the degree of neuronal injury, given the correlation between the degree of signal abnormality and the amount of volume loss. The volume loss indicates a loss of neurons; prior studies have shown that the decreased hippocampal size seen in hippocampal sclerosis is correlated with decreased neuronal density especially in areas CA1, CA3, and CA4 segments of the cornu ammonis and the granular cell layer of the hippocampus [39, 40]. We also found that markedly increased hippocampal signal intensity on initial imaging did indeed predict mesial temporal sclerosis. This finding is also supportive evidence that the abnormal hippocampal signal reflects not only acute edema but also underlying structural damage.
Seizure duration in our patients was moderately well correlated with initial hippocampal signal intensity. In a previous study of a larger group of patients with a shorter mean duration of febrile seizure activity than our study group, a stronger correlation between seizure duration and hippocampal signal intensity was found [6]. Because the degree of signal abnormality correlated well with the degree of hippocampal volume loss and subsequent seizures in our patients, the finding of a strong association of seizure duration and signal intensity would be further evidence that aggressive efforts to limit seizure duration in these children are warranted to minimize hippocampal damage.
The bilateral, although asymmetric, loss of volume in patients 1 and 2 suggests that both hippocampi were injured despite the predominantly unilateral appearance of the MRI examinations. Pathologic studies in patients with temporal lobe epilepsy have confirmed that mesial temporal sclerosis, although predominantly unilateral, can be bilateral in 20% and can be accompanied by end-folium sclerosis in the contralateral side in another 40% of cases [41]. Furthermore, bilateral hippocampal volume loss is more common in temporal lobe epilepsy with a history of febrile seizures compared with temporal lobe epilepsy without febrile seizures [42]. In our study, bilateral hippocampal volume loss was seen only in children with moderate or severe hippocampal abnormality and not in children with lower scores, suggesting that relatively severe hippo campal injury must be present for evidence of bilateral injury to be seen on subsequent imaging.
Like any study, our investigation is marked by some limitations. First, the use of 2D coronal imaging, instead of 3D imaging, could have led to miscalculation of hippocampal sizes in some cases due to partial volume averaging of very hyperintense hippocampi with CSF. Second, we recognize that a second observer for the scoring of hippocampal signal intensity would have increased confidence in the scoring process. However, one of the major points of our study is that hippocampi that were scored as having moderate (score of 3) or severe (score of 4) signal intensity increase had a different outcome in terms of size compared with hippocampi with minimal (score of 1) or mild (score of 2) signal intensity increase. Thus, it is appropriate to ask the practical question whether it is likely that two experienced neuroradiologists would differ as to whether signal intensity of a hippocampus falls in the upper end of the scoring range (moderate or severe) as opposed to the lower half of the scoring range (normal, minimal, or mild signal abnormality). We believe the likelihood is low but we admit that, without additional observers, we cannot prove this assumption. Finally, we recognize that the lack of FLAIR images, which are very sensitive to increased signal intensity, may have led to underestimation of signal increase in some patients.
In summary, we found a strong correlation between the degree of hippocampal signal abnormality on initial imaging and evidence of subsequent hippocampal damage. Our data suggest that visual evaluation of the severity of the initial changes by a competent neuroradiologist may be of prognostic value for selecting infants who will develop mesial temporal sclerosis and seizure disorders. Children with marked signal abnormality after febrile status epilepticus might be especially good candidates for neuroprotective drugs in the event that such therapies become available. The findings reported here need to be reexamined in a larger study in an effort to confirm the correlations seen in this small sample.
Footnotes
This research was supported by grants to D. V. Lewis from the American Epilepsy Society and the Charles A. Dana Foundation.
Address correspondence to J. M. Provenzale.
References
1.
Chan S, Chin SSM, Kartha K, et al. Reversible signal abnormalities in the hippocampus and neocortex after prolonged seizure. Am J Neuroradiol 1996; 17:1725-1731
2.
De Carolis P, Crisci M, Laudadio S, Baldrati A, Sacquegna T. Transient abnormalities on magnetic resonance imaging after partial status epilepticus. Ital J Neurol Sci 1991; 13:267-269
3.
Lee BI, Lee BC, Hwang YM. Prolonged ictal amnesia with transient focal abnormalities on magnetic resonance imaging. Epilepsia 1992; 33:1042-1046
4.
Cox JE, Mathews VP, Santos CC, Elster AD. Seizure-induced transient hippocampal abnormalities on MR: correlation with positron emission tomography and electroencephalography. Am J Neuroradiol 1995; 16:1736-1738
5.
Stafstrom CE, Tien RD, Montine TJ, Boustany RM. Refractory status epilepticus associated with progressive magnetic resonance imaging signal change and hippocampal neuronal loss. J Epilepsy 1996; 9:253-258
6.
VanLandingham KE, Heinz ER, Cavazos JE, Lewis DV. Magnetic resonance imaging evidence of hippocampal injury after prolonged focal febrile convulsions. Ann Neurol 1998; 43:413-426
7.
Tien RD, Felsberg GJ. The hippocampus in status epilepticus: demonstration of signal intensity and morphologic changes with sequential fast spin-echo MR imaging. Radiology 1995; 194:249-256
8.
Farrow TD, Dickson JM, Grunewald RA. A six-year follow-up MRI study of complicated early childhood convulsion. Pediatr Neurol 2006; 35:257-260
9.
Bronen RA. The status of status: seizures are bad for your brain's health. Am J Neuroradiol 2000; 21:1782-1783
10.
Tarkka R, Paakko E, Phytinen J, Uhari M, Rantala H. Febrile seizures and mesial temporal sclerosis: no association in a long-term follow-up study. Neurology 2003; 60:215-218
11.
Lewis DV, Barboriak DP, MacFall JR, Provenzale JM, Mitchell TV, VanLandingham KE. Do prolonged febrile seizures produce medial temporal sclerosis? Hypotheses, MRI evidence and unanswered questions. Prog Brain Res 2002; 135:263-278
12.
Tarkka R, Pääkkö E, Pyhtinen J, Uhari M, Rantala H. Febrile seizures and mesial temporal sclerosis: no association in a long-term follow-up study. Neurology 2003; 60:215-218
13.
Scott R. Febrile seizures and mesial temporal sclerosis: no association in a long-term follow-up study. (letter) Neurology 2003; 61:588-589
14.
Shinnar S. Febrile seizures and mesial temporal sclerosis. Epilepsy Curr 2003; 3:115-118
15.
Roch C, Leroy C, Nehlig A, Namer IJ. Predictive value of cortical injury for the development of temporal lobe epilepsy in 21-day-old rats: an MRI approach using the lithium-pilocarpine model. Epilepsia 2002; 43:1129-1136
16.
Fabene PF, Marzola P, Sbarbati A, Bentivoglio M. Magnetic resonance imaging of changes elicited by status epilepticus in the rat brain: diffusion-weighted and T2-weighted images, regional blood volume maps, and direct correlation with tissue and cell damage. NeuroImage 2003; 18:375-389
17.
Dubé C, Yu H, Nalcioglu O, Baram TZ. Serial MRI after experimental febrile seizures: altered T2 signal without neuronal death. Ann Neurol 2004; 56:709-714
18.
Dubé C, Richichi C, Bender RA, Chung G, Litt B, Baram TZ. Temporal lobe epilepsy after experimental prolonged febrile seizures: prospective analysis. Brain 2006; 129:911-922
19.
[No authors listed]. Guidelines for epidemiologic studies on epilepsy. Commission on Epidemiology and Prognosis, International League Against Epilepsy. Epilepsia 1993; 34:592-596
20.
Shinnar S, Pellock JM, Moshe SL, et al. In whom does status epilepticus occur: age-related differences in children. Epilepsia 1997; 38:907-914
21.
Lewis DV. Febrile convulsions and mesial temporal sclerosis. Curr Opin Neurol 1999; 12:197-201
22.
Maytal J, Shinnar S, Moshe SL, Alvarez LA. Low morbidity and mortality of status epilepticus in children. Pediatrics 1989; 83:323-331
23.
Shinnar S. Prolonged febrile seizures and mesial temporal sclerosis. Ann Neurol 1998; 43:411-412
24.
Scott RC, Gadian DG, King MD, et al. Magnetic resonance imaging findings within 5 days of status epilepticus in childhood. Brain 2002; 125:1951-1959
25.
Scott RC, King MD, Gadian DG, Neville BG, Connelly A. Hippocampal abnormalities after prolonged febrile convulsion: a longitudinal MRI study. Brain 2003; 126:2551-2557
26.
Jackson GD, Berkovic SF, Tress BM, Kalnins RM, Fabinyi GCA, Bladin PF. Hippocampal sclerosis can be reliably detected by magnetic resonance imaging. Neurology 1990; 40:1869-1875
27.
Righini A, Pierpaoli C, Alger JR, DiChiro G. Brain parenchyma apparent diffusion coefficient alterations associated with experimental complex partial status epilepticus. Magn Res Imaging 1994; 12:865-871
28.
Ebisu T, Rooney WD, Graham SH, Mancuso A, Weiner MW, Maudsley AA. MR spectroscopic imaging and diffusion-weighted MRI for early detection of kainate-induced status epilepticus in the rat. Magn Reson Med 1996; 36:821-828
29.
Wang Y, Majors A, Najm I, et al. Postictal alteration of sodium content and apparent diffusion coefficient in epileptic rat brain induced by kainic acid. Epilepsia 1996; 37:1000-1006
30.
Wenzel HJ, Born DE, Dubach M, et al. Morphological plasticity in an infant model of temporal lobe epilepsy. Epilepsia 2000; 41[suppl 6]:S70-S75
31.
Bouilleret V, Nehlig A, Mariscaux C, Namer IJ. Magnetic resonance imaging follow-up of progressive hippocampal changes in a mouse model of mesial temporal lobe epilepsy. Epilepsia 2000; 41:642-650
32.
Tokumitsu T, Mancuso A, Weinstein PR, Weiner MW, Naruse S, Maudsley AA. Metabolic and pathological effects of temporal lobe epilepsy in rat brain detected by proton spectroscopy and imaging. Brain Res 1997; 744:57-67
33.
Nakasu Y, Nakasu S, Morikawa S, Uemura S, Inubushi T, Handa J. Diffusion-weighted MR in experimental sustained seizures elicited with kainic acid. Am J Neuroradiol 1995; 16:1185-1192
34.
Hasegawa D, Orima H, Fujita M, et al. Diffusion-weighted imaging in kainic acid–induced complex partial status epilepticus in dogs. Brain Res 2003; 983:115-127
35.
Nairismagi J, Pitkanen A, Kettunen MI, Kauppinen RA, Kubova H. Status epilepticus in 12-day-old rats leads to temporal lobe neurodegeneration and volume reduction: a histologic and MRI study. Epilepsia 2006; 47:479-488
36.
Nohria V, Tien RD, Lee N, et al. MRI evidence of hippocampal sclerosis in progression: a case report. Epilepsia 1994; 35:1332-1336
37.
Gross DW, Li LM, Andermann F. Catastrophic deterioration and hippocampal atrophy after childhood status epilepticus. Ann Neurol 1998; 43:687
38.
Parmar H, Lim SH, Tan NCK, Lim CCT. Acute symptomatic seizures and hippocampus damage: DWI and MRS findings. Neurology 2006; 66:1732-1735
39.
Lee N, Tien RD, Lewis DV, et al. Fast spin-echo, magnetic resonance imaging–measured hippocampal volume: correlation with neuronal density in anterior temporal lobectomy patients. Epilepsia 1995; 36:899-904
40.
Bronen RA, Cheung G, Charles JT, et al. Imaging findings in hippocampal sclerosis: correlation with pathology. Am J Neuroradiol 1991; 12:933-940
41.
Margerison JH, Corsellis JA. Epilepsy and the temporal lobes: a clinical, electroencephalographic and neuropathological study of the brain in epilepsy, with particular reference to the temporal lobes. Brain 1966; 89:499-530
42.
Barr WB, Ashtari M, Schaul N. Bilateral reductions in hippocampal volume in adults with epilepsy and a history of febrile seizures. J Neurol Neurosurg Psychiatry 1997; 63:461-467
Information & Authors
Information
Published In
Copyright
© American Roentgen Ray Society.
History
Submitted: April 12, 2007
Accepted: October 12, 2007
First published: November 23, 2012
Keywords
Authors
Metrics & Citations
Metrics
Citations
Export Citations
To download the citation to this article, select your reference manager software.