JLE

Epileptic Disorders

MENU

Clinical findings, imaging characteristics and outcome in catastrophic post-encephalitic epilepsy Volume 2, numéro 3, Septembre 2000

The incidence of unprovoked and recurrent seizures after previous central nervous infections is high and accounts for 1-5% of patients with epilepsy [1, 2]. Survivors of viral encephalitis have, among other neurological sequelae, a ten-fold increase in the risk of developing epilepsy [1]. Patients with post-encephalitic epilepsy have multiple and varied clinical patterns of seizures with partial seizures accounting for approximately 70% [1].

Radermecker [3], in 1956, was the first to be intrigued by the varied electro-clinical features of acute viral encephalitis and gave a detailed account of the etiological, neuropathological and prognostic issues. Because viral encephalitis affects the brain diffusely, with the notable exception of Herpes simplex encephalitis (HSE), which has a well known predilection for temporal and orbito-frontal structures [4, 5], precise seizure localization is usually difficult. The tendency for temporal focalization in postencephalitic epilepsy has previously been recognized by the St. Petersburg group [6]. Marks et al. [7], from the Yale group suggest that patients with epilepsy following meningitis and encephalitis may have a favorable surgical outcome if the acute infectious disease occurs before the age of four and if mesial temporal sclerosis (MTS) can be identified. The present investigation arose from the recognition that patients with postencephalitic epilepsy, investigated over the years, often had catastrophic disease, and that their surgical outcome was usually unsatisfactory [8, 9]. It became clear that these patients usually had clinical features suggestive of temporal origin, but that in the majority there was additional evidence for extratemporal involvement.

The purpose of this study is to delineate seizure patterns, imaging characteristics, and EEG localization using surface and depth recordings in a group of patients with post-encephalitic epilepsy, and to assess seizure control and pathological findings after resective surgery.

Methods

Study population

All patients identified as having epilepsy secondary to viral encephalitis, studied between January 1982 and December 1999 at the Montreal Neurological Institute, were evaluated retrospectively. The diagnosis of encephalitis was based on clinical history and medical records consistent with CNS infection and rapid brain parenchyma involvement (stupor, coma, focal neurological signs, and convulsions), supported by CSF studies, EEG, and imaging findings as available [10]. Virological proof was not required to establish the diagnosis. Patients with meningitis, brain abscess, febrile convulsions without signs of CNS parenchymal involvement, chronic encephalitis or Rasmussen's syndrome, post-infectious or post-vaccination encephalitis or progressive neurological diseases other than epilepsy were excluded.

Patient evaluation

Forty-two patients were studied. We reviewed their medical and neurological history, age at encephalitis, age at first unprovoked seizure, description of their aura and seizure characteristics. Magnetic resonance images (MRI) were reviewed with special attention to mesial temporal lobe structures. In fourteen patients, volumetric measurements of the amygdala and hippocampus were obtained [11]. When no MRI was available cranial computerized tomography scans (CCT) were reviewed (3 patients). Neuropsychological studies (including intracarotid amytal test when performed) and all interictal and ictal video EEG studies were assessed; fifteen patients underwent, in addition, prolonged intracranial EEG monitoring with a combination of temporal and extratemporal depth electrodes, as well as epidural electrodes, to localize the seizure onset. Patients were classified according to their interictal and ictal epileptiform scalp EEG patterns into three different groups: unilateral temporal (UTLE), bilateral temporal (BTLE) and extratemporal and/or multifocal or generalized (ETMFE).

Surgery and Pathology

Twenty-four patients underwent surgical resection. Procedures included temporal and extratemporal neocortical resections, and callosotomy. The decision for surgery and the site of resection were determined after extensive evaluation and were based on clinical findings, imaging studies, ictal EEG often with stereotactic depth electrodes (SEEG) evaluation, and neuropsychological test results. The pathology of the resected tissue was reviewed in all patients.

Follow Up

Each available patient was interviewed by two of the authors (FD and ET). Handicap and quality of life were assessed by questioning patients about the frequency of residual seizures, the number and amount of antiepileptic medications (AED), and the performance at work, at school and in activities of daily living. Information about patients not available for interview was obtained from relatives or care givers, hospital records, physicians or neurologists following the patients. The degree of handicap and quality of life were graded with the Glasgow Outcome Scale (GOS): 1 = death within the first month; 2 = vegetative state; 3 = severe disability (conscious, needing assistance for some activities of daily living; including patients who were independent in most activities of daily living, but required constant supervision, because they were unable to organize their daily activities or leave their home without supervision); 4 = moderate disability (neurological or intellectual impairment, but independent; some previous occupational and social abilities were no longer present); 5 = good recovery (leading an independent life, with or without minimal neurological impairment [12]. Surgical outcome with respect to seizures was assessed using Engel's modified scale [13].

Statistical analysis

ANOVA and two tailed t-tests, with Bonferroni correction, were used for group comparisons on continuous variables. Pearson Chi square tests were used for the evaluation of frequency distributions of categorical variables accross groups.

Results

Forty-two patients with intractable epilepsy were studied (table I). Twenty-six were men. Eight had unilateral temporal lobe epilepsy (UTLE), 12 bilateral temporal lobe epilepsy (BTLE), and 22 extratemporal, multifocal or generalized epilepsy (ETMFE).

Encephalitis

Age at encephalitis ranged from seven months to 75 years (median, 10.5 yrs; mean, 17; SD = 15.5). The age distribution is shown in figure 1. Four patients developed their infection at or before age four years, and 38 after that. The mean age of infection in patients with UTLE was lower compared to those with BTLE and with ETMFE, but this difference did not reach statistical significance (table II). The causative viral agent was identified in 18 patients (10 Herpes simplex virus, three Varicella, two measles, one mumps, one Coxsackie B4, and one parainfluenza III), and remained unknown in 24 others. All 42 patients had had severe encephalitis: 33 (79%) had status epilepticus (SE) or recurrent seizures during the illness. Profound disturbances of consciousness were found in 32 (76%) patients, including 26 (62%) in deep coma. The proportion of SE was lower in the UTLE group (2/8) compared to the BTLE (6/12) and ETMFE group (14/22), but the numbers in each category were too small for reliable statistical analysis (table II).

There was no latent period from the initial insult to the onset of recurrent seizures in the majority of patients (32/42). In 10 individuals (24%), the first unprovoked seizures appeared between two months and eight years (median, 3 yrs) after encephalitis. The mean latency for the 42 patients was 0.8 years (SD = 1.9). The latency was shorter in the ETMFE group (0.06 yr, SD = 0.2) compared to the UTLE group (2.6 yrs, SD = 2.6) and the BTLE group (0.9 yrs, SD = 2.4) (F2.39 = 6.75; p < 0.01). Two sided t-tests showed significant differences between patients with ETMFE and UTLE (p < 0.01), but not between ETMFE and BTLE, and BTLE and UTLE (p > 0.10).

Clinical seizure characteristics

The patients presented with frequent and usually medically intractable partial seizures, often with secondary generalization (33/42, 79%). Thirty-six (86%) reported an aura. The most common was limbic-mesial temporal (epigastric, déjà vu, fear, olfactory) in 27 (64%) patients. Neocortical temporal (auditory, vertiginous) and extratemporal (visual, somatosensory) auras were reported in five (12%) and nine (21%) patients, respectively. The aura was non-localizing (cephalic, dizziness, "strange feeling") in 11 (26%) patients. All patients with UTLE had temporal and no other type of aura. Extratemporal or non-localizing auras were reported in both the BTLE and ETMFE groups (table II).

EEG/SEEG findings

All patients underwent surface EEG studies with additional anterior zygomatic or sphenoidal electrodes. The interictal abnormalities were non-localizing in six patients, and restricted to the temporal lobe in 21 (50%), seven of them unilateral. The abnormalities were extratemporal only in one, and multifocal or generalized in 14 (33%) patients. An ictal EEG was obtained in 31 patients: twelve (39%) individuals had an electrographic seizure onset in a temporal lobe, five of them unilateral; two patients had extratemporal, and 15 (48%) multifocal or generalized onset; in two patients seizure onset could not be localized.

For better localization of the epileptogenic area, SEEG investigations were performed in 15 patients. Depth electrodes were implanted bilaterally in frontal and temporal regions in nine patients (patient no. 9, 20, 24, 26, 30, 31, 34, 35, 39), unilateral fronto-temporal in one (patient no. 12), and bilateral temporal in four (patient no. 8, 12, 13, 38). Additional epidural recordings over pre- and post-central areas were obtained in seven (patient no. 8, 20, 24, 29, 30, 31, 34,). Patient no. 8 from the UTLE group had seizures recorded from both temporal lobe structures during intracranial evaluation, but more than 90% came from the left side. One patient had only epidural electrodes over pre- and post-central, and temporal regions (patient no. 22). The epileptogenic zone was bilateral temporal in five, multifocal temporal and extratemporal in 10 patients.

Neuroimaging (table I)

In three patients (no. 11, 14, 28), only CCT scans were available; two of them showed diffuse atrophy. MRI scans were reviewed in 39 patients: Seven (18%) had mild diffuse atrophy or normal MRI scans; temporal abnormalities (figure 2 a, b), i.e. mesial and neocortical atrophy, with or without increased T2 signal, or cystic lesions, were found in 12 (31%). Unilateral temporal abnormalities (including hippocampal atrophy and gliosis i.e. increased T2 signal) were present in seven (18%) patients, and bilateral temporal abnormalities in five (13%). Bilateral signal changes in the thalamus and cystic lesions in the basal ganglia were found in two individuals. Finally, the majority (20/39, 51%) had a combination of temporal and extratemporal abnormalities, which included regional and severe diffuse atrophy, with widespread T2 signal changes, more pronounced in the perisylvian and periinsular regions (figure 2 c, d). This pattern was frequently found in all three groups (table II).

Of the 14 patients for whom volumetric measurements of the amygdalae and hippocampi were available, only two had normal volumes. Atrophy was unilateral in four and bilateral in the remaining eight.

Neuropsychological findings

The majority of patients for whom data were available (34/39, 87%) showed some degree of intellectual impairment. The mean full scale IQ (FSIQ) was 86 (SD = 13), verbal IQ 85 (SD = 12) and performance IQ 88 (SD = 16). Patients with UTLE had a higher mean compared to those with BTLE and ETMFE, but this difference did not reach statistical significance (table II). Cognitive dysfunction was widespread in 23 (55%), and predominantly temporal in 10 (24%) patients. The sodium-amytal test was performed in 17 patients and confirmed memory impairment in the majority (11/17, 79%); these impairments were based on unilateral pathology in five and bilateral pathology in six individuals.

Surgery and Neuropathology

Twenty-four patients were surgically treated; twenty-two underwent a temporal resection, and two had, in addition, extratemporal neocortical resections. In two patients, a frontal resection was carried out, one had also an anterior callosotomy. Seventeen hippocampi could be analyzed. Classic mesial temporal sclerosis (MTS) was diagnosed in 12, focal gliosis in 3, and minimal or no changes were found in 2 cases. Inflammatory changes with perivascular lymphocytic infiltration were found in 5 patients; four of them had had Herpes simplex encephalitis (HSE) as the initial brain insult. In one Herpes simplex virus (HSV-1) DNA could be detected by polymerase chain reaction (PCR) in the resected tissue and in the CSF.

Follow-up and postsurgical outcome

A follow-up interview was obtained in 39 patients. The mean duration of the interval between the follow-up and the acute illness was 10.9 years (SD = 8.9; range, 1 to 34; median, 9). The mean Glasgow outcome scale (GOS) score after the encephalitis was 3.6 (SD = 0.8).

The mean GOS score was higher in patients with UTLE (4.7; SD = 0.8), compared to those with BTLE (3.2; SD = 0.4) and ETMFE (3.6; SD = 0.8) (F2.39 = 12.65; p < 0.0001). A two sided t-test showed significant differences in the mean GOS, between patients with UTLE and BTLE (p < 0.001), UTLE and ETMFE (p < 0.001), but not between BTLE and ETMFE. Six of the eight patients with UTLE were able to lead an independent life with full performance at work (GOS = 5), whereas no patient with BTLE and only four of the 22 cases with ETMFE reached the same level of independence. Two patients died during or after a seizure due to cardio-pulmonary arrest (patient no. 12, 22).

In the medically treated group (18 patients), the mean follow-up was 15.9 years (SD = 9.2; median, 15.5 yrs; range, 1 to 34) two patients received monotherapy, and the remaining received 2 or more AEDs. Two patients (no. 23, 41) were eventually well controlled and ultimately entered remission, while the majority remained medically refractory and continued to have a high seizure frequency.

In the surgically treated group (24 patients), the mean follow-up was 5.1 years (SD = 4.4; median, 3.5; range, 1 to 15 yrs). The postsurgical outcome was best in the UTLE group (7/8 patients in class I or II) and poor in the BTLE and ETMFE group (13/16 patients in class III or IV). Four patients with UTLE (17%) became completely seizure-free after surgery (patients no. 3, 5, 6, 7); two of them remained seizure-free after successful AED withdrawal. Thirteen patients were in class III or IV; a worthwhile seizure reduction was achieved in two (patient no. 20, 28), and at least some reduction of seizure frequency or severity in five (patient no. 4, 10, 26, 29, 36). These patients reported improvement in their quality of life, which was confirmed by family members or relatives.

The use of an invasive EEG method did not improve seizure localization and surgical outcome. SEEG investigation confirmed a widespread or multifocal epileptogenesis in all patients investigated with intracranial electrodes. Seven individuals were operated and only two so far (patient no. 8, 39) had a good outcome, but with a follow-up of less than a year.

Discussion

The majority of patients included in this study were referred for presurgical evaluation because of severe, intractable epilepsy. Most of them sustained a catastrophic, acute infectious illness which consequently led to a chronic and severe epileptic syndrome. This study focuses on a highly selected population and, therefore, conclusions about the complete spectrum of seizures following viral encephalitis of varied etiologies cannot be drawn from this group of individuals. This may account for the divergent results obtained in our study compared to previous ones for patients with intractable epilepsy following CNS infections, carried out in other tertiary referral centers: Yale [7], Memphis and Seattle [14], and Cleveland [15].

Bacterial meningitis and even viral encephalitis, irrespective of their causative agent may lead to typical mesial temporal lobe epilepsy with hippocampal sclerosis detected by MRI [16-18] and favorable postsurgical outcome [7, 14, 15]. The relationship between complex partial seizures and neuronal cell loss in the mesial temporal structures has been recognized for over a century, but there is still debate whether they are the cause [19-21], or the result of seizures [22, 23]. Mesial temporal sclerosis (MTS) correlates with early (< 3 years) childhood convulsions [24] and early childhood prolonged febrile convulsions [25]. In these patients an excellent outcome after surgical treatment can be expected [26].

In previous studies of patients with epilepsy, the age of the initial CNS infection was found to be important in predicting the development of MTS following meningitis or encephalitis [7, 15]. In the present series, the mean age at encephalitis in the UTLE group was younger compared to the BTLE group (10.3 versus 25.6 yrs), and the ETMFE group (10.3 versus 14.9 yrs), but this difference did not reach statistical significance given the large variability in each group. Furthermore, all four patients in our series, who had their initial brain insult at four years or younger, developed multifocal or bilateral temporal epilepsy. This conflicting result may be due to the severity of the early insult, which leads not only to mesial temporal lesions, but also to extratemporal or multifocal lesions via direct viral damage.

Nine of 11 patients with, and four of six without postoperatively confirmed MTS had SE or repetitive seizures during the acute phase of the encephalitis. The proportion of SE was higher in the BTLE (58%) and ETMFE group (64%) compared to the UTLE group (25%) but the numbers in each category were too small for reliable statistical analysis. The high incidence of SE (22/42 patients, or 52%) and the inflammatory lesions themselves, are likely to increase the damage to both hippocampi and extratemporal structures, and therefore to obscure the pattern of clear unilateral MTS. This high proportion of SE is in agreement with the study of Yoshioka et al. [27], who found that status epilepticus during encephalitis was the most important predictive factor in the development of epilepsy. Berg et al. [28], also identified a strong association between intractability and a history of SE before the onset of habitual seizures.

The Cleveland group reported a longer latent period from the initial CNS insult to the first unprovoked seizure in a mixed group of patients with meningitis and encephalitis [15]. We found that the latent period was shorter in the ETMFE group (0.06 yr) compared to the UTLE group (2.6 yrs) and the BTLE group (0.9 yr), but there was no difference between the mean latency in the UTLE and BTLE group. The severity and the more widespread nature of the insult may explain the earlier onset of the epilepsy.

Marks et al. [7] reported only four patients or 18% with imaging abnormalities (2 had measles, 1 mumps, 1 unknown etiology), and all of them had unilateral hippocampal atrophy. This is in striking contrast with our series where only 18% patients had a normal MRI or slight diffuse abnormalities. Seven patients (20%) had unilateral temporal atrophy with or without signal changes in the hippocampus. The majority (51%) of our patients had a combination of temporal and extratemporal abnormalities, which consisted of regional temporal atrophy, with widespread T2 signal changes most pronounced in the perisylvian and periinsular regions, extending to the orbito-frontal cortex and combined with severe diffuse atrophy. This pattern is consistent with the involvement by HSV and can be unilateral or bilateral [5]. Furthermore, the tropism of viruses to CNS structures seems to be important for the type of lesion that results. For instance, in the study of Marks et al. [7], a comparatively high percentage (5/21, 24%) of mumps and measles encephalitis cases was included. Mumps and measles are the most frequent causes of post-infectious encephalomyelitis and, at times, distinction between acute and post-infectious disease can be difficult. The MRI changes of post-infectious encephalomyelitis consist of multifocal white matter lesions, which slowly resolve over time [29]. This may explain the reported high rate of normal MRI scans in their patients [7]. In addition, Free and co-workers pointed out that volumetric assessment of the temporal lobes in this patient group is clearly superior to visual analysis alone [30]. This technique was not used by Marks and co-workers [7]. In our study, volumetric measurement confirmed the widespread and bilateral nature of the brain damage in 57% of those where it was performed.

In our study the etiology was established in 18 patients: a quarter of them were survivors of severe (mean GOS, 3.7) Herpes simplex encephalitis (HSE). The clinical course of HSE has often been fulminant with a high mortality and morbidity rate even after the introduction of treament with acyclovir [31]. Although, there are no neuropathological reports concerning the influence of specific antiviral treatment, it is highly likely, that the lesions may be more restricted to the infero-mesial structures of the temporal lobe [5]. Untreated HSE often resulted in very severe cognitive deficits and behavioral disturbances [32], and these patients were often not referred for consideration of surgical treatment of their epilepsy.

Four patients, who had had HSE underwent surgery. All showed signs of inflammation in their resected tissue with perivascular lymphocytic infiltrates and microglial reaction, whereas only one patient with "non-herpes encephalitis" had similar histopathology. In one patient (no. 36), HSV-1 DNA was detected in the resected tissue using PCR, three years after the encephalitis. This unusual finding was previously reported [33] and raises the question of focal chronic HSV encephalitis. Perivascular lymphocytes in surgical specimens from patients with medically intractable complex partial seizures have been described [34-36], some of them with a previous history of meningitis [34] and an unfavorable postsurgical outcome [35, 36]. Aguilar and Rasmussen [37] reviewed the pathological findings in 512 surgical specimens from 449 patients operated on for intractable epilepsy. Thirty two cases demonstrated histological evidence suggesting active encephalitis, and 12 of these had progressive neurological deterioration compatible with Rasmussen's syndrome [38]; the other 20 cases did not show progressive neurological deterioration, but their detailed clinical status was not described. In our series only one of the four operated cases with HSE had a favorable outcome (patient no. 1). Such patients may be part of a spectrum of clinico-pathologic findings ranging from classic Rasmussen's syndrome to complex partial seizures without evidence of progression [36, 39]. The detection of various Herpes viruses (Herpes simplex, Epstein-Barr, cytomegalovirus, human Herpes virus 6) has been reported not only in patients with Rasmussen's syndrome [40-43], but also in brain samples of patients with focal epilepsy [44] and in normal brain tissue [45]. At this time one cannot assume that Rasmussen's syndrome is due to these viruses.

The prognosis after surgical treatment in our patients depended on the seizure localization: seven of the eight surgically treated patients in the UTLE group had a favorable outcome (Engel class I and II) whereas 13 of the 16 with bilateral or multifocal epilepsy did not do well (Engel class III and IV). The main prognostic factor for developing UTLE in this group, unlike that in patients with meningitis and in previous reports on post-encephalitic epilepsy, was not the age at the initial brain insult, but the severity of the encephalitis. The single most predictive factor for development of UTLE was the Glasgow outcome score reflecting the severity of the brain damage. It is not sufficiently clear, whether there was any clinical pattern that permitted recognition of future UTLE patients and distinction from the others. The more prolonged interval between the acute disease to the onset of seizures and the lower incidence of status epilepticus in this small subgroup suggest a mechanism different from that in the majority of patients with postencephalitic epilepsy. Even if the MRI shows unilateral lesions suggesting MTS in a patient, neuropsychological testing may demonstrate widespread or diffuse brain dysfunction and the EEG often reveals multiple epileptogenic foci (neocortical and/or extratemporal), which are negative prognostic factors for seizure remission [46]. Conversely, a normal or only mildly and diffusely abnormal MRI is not necessarily a favorable prognostic sign, since 20% of the patients with BTLE and 24% of those with ETMFE had normal MRI or mild diffuse abnormality on MRI scans.

Only two patients were eventually well controlled and remitted spontaneously. This was an exceptional finding in the present series and may reflect the referral pattern to our institution. Anecdotal observations of other patients with regression of the severity or cessation of the attacks suggest that population-based studies may show a somewhat different pattern of outcome of encephalitis with coma and seizures.

A particular reason for the referral of this group of patients was the severity and catastrophic nature of their epilepsy, its particular intractability, and the associated cognitive and in particular mnemonic abnormalities. In the evaluation of these patients surface EEG studies are often not adequate, though the hope of identifying a reasonably lateralized and localized seizure generator by invasive techniques is rarely realized. Detailed imaging studies using all available approaches (volumetry, T2-relaxometry, FLAIR sequences, magnetic resonance spectroscopy and others) may help identify a subgroup of individuals with unilateral temporal lobe epilepsy, who may be expected to have a favorable outcome. Other groups, from Yale [7], Cleveland clinic [15] and Seattle and Memphis [14] also noted the presence of this "privileged" subgroup. These patients were somewhat younger than the others, but all of those who had encephalitis before the age four in our series, did not follow this pattern. It is unlikely, that this subgroup can be identified without detailed investigation. Zemskaya et al. [6] refer to temporal focalization with many temporal clinical and EEG features, which lead to the referral of such patients and the hope that a surgical approach will be helpful, although the process is more diffuse. The fact that most patients were treated by temporal resections confirms the apparent localization of maximal abnormality, but this is obviously not sufficient in most patients to lead to a good result. The presence of extratemporal clinical features, coupled with knowledge of the etiology should lead to caution in the surgical decision and recognition of the limited goal, that resection will bring about in many of these patients. However, a detailed follow-up interview of patients in class IV showed that five reported a reduction in seizure frequency, and severity, which led to an increase in the quality of their life. The limited outcome that surgical treatment in these patients will bring about may be regarded as palliative rather than curative.

CONCLUSION

In conclusion, our series demonstrates an unfavorable prognosis in the majority of patients with post-encephalitic epilepsy: only a small proportion of these patients develop clear unilateral temporal epilepsy, amenable to surgical treatment with a success rate comparable to that of mesial temporal lobe epilepsy of other cause. In such patients every attempt should be made to study the clinical, EEG and imaging features, which may explain this rather unusual pattern of postencephalitic disease. In severely affected patients who have evidence of bilateral or diffuse disease and particularly those with extratemporal clinical features, the outcome after surgery is usually disappointing, even after extensive localization studies using invasive intracranial techniques.

Addendum: Surgical outcome in a subset of these patients has been previously published [47].